Home

THE CONSTRUCTION OF THE L3 EXPERIMENT

1. ee rn T p od Fig 64 Longitudinal cut of part of the L3 hadron calorimeter One side of the forward backward system is shown dotted area proximate that of a 10 mm thick uranium plate in terms of nuclear absorption lengths The first compartment fine sampling part is 26 chamber layers deep and is equipped with 5 mm uranium plates In this compart ment the first three layers of uranium plates have been Table 6 Dimensions and mechanical properties of the hadron calorime ter endcap modules Container type Total HCl HC2 HC3 System Number of modules 4 4 4 12 Inner diameter m 0 95 0 32 0 42 Outer diameter m 1 60 0 94 0 94 Length m 1 37 0 55 0 48 Compartments 4 2 2 Flange thickness mm Front 25 19 19 Additional shielding for BGO TEC 10415 10 415 Inside each 16 16 16 Back 25 25 25 Wall thickness mm Inner 15 15 15 Outer 4 5 5 Weight t Total per module 8 5 1 4 11 44 Uranium per module 522 105 082 284 Chamber layers per module 77 27 23 127 Total number of chambers 1384 488 412 2284 Number of tubes per chamber 25 24 19 Total number of tubes wires 34600 11712 7828 54140 replaced by steel plates in order to shield the BGO crystals and the TEC chamber from the uranium radio activity background The three remaining compart ments eacn 17 chamber layers deep are equipped with Fig 65 Perspective view of the forward backward hadron calorimeter B Ade
2. C G YANG 9 K S YANG 9 Q Y YANG 9 Z Q YANG 2 C H YE 9 S C YEH 3 ZW YIN C ZABOUNIDIS 9 L ZEHNDER Y ZENG D H ZHANG S Y ZHANG Z P ZHANG B ZHOU J F ZHOU V Z P ZHOU R Y ZHU 2 A ZICHICHI 9 M ZOFKA and J ZOLL y Physikalisches Institut RWTH Aachen FRG J Physikalisches Institut RWTH Aachen FRG National Institute for High Energy Physics NIKHEF Amsterdam The Netherlands NIKHEF H and University of Nijmegen Nijmegen The Netherlands University of Michigan Ann Arbor USA Laboratoire de Physique des Particules LAPP Annecy France Johns Hopkins University Baltimore USA Institute of High Energy Physics IHEP Beijing P R China Tata Institute of Fundamental Research Bombay india Northeastern University Boston USA Central Research Institute for Physics of the Hungarian Academy of Sciences Budapest Hungary Harvard University Cambridge USA Massachusetts Institute of Technology Cambridge USA INFN Sezione di Firenze and University of Firenze Italy ii Leningrad Nuclear Physics Institute Gatchina USSR Ad European Laboratory for Particle Physics CERN Geneva Switzerland World Laboratory FBLJA Project Geneva Switzerland University of Geneva Geneva Switzerland w Chinese Uniuersity of Science and Technology USTC Hefei P R China University of Lausanne Lausanne Switzerland T Institut de Physique Nu
3. 3 no TEC FADC TEC FADC TEC FADC Bi SPEC PURPOSE DM GAS SYSTEM COMMUNICATION FASTBUS 1892 LE CROY MONITORING VAX 750 COMPUTER Z gt DAQ Fig 107 Schematic diagram of the central track detector readout systems B Adeva et al The construction of the L3 experiment Fig 108 A VME module containing two Flash ADC ckannels and one DRP calculated by the DRP and added to its output Each VME crate is controlled by a CM which consists of three PC boards It is built around a Motorola 68000 12 MHz CPU The central function of the CM is to communicate with the DRP units in the crate and to transfer the DRP data to the data acquisition system The CM also contains the FADC main control unit FMC which accepts the primary trigger input signals and controls all FADC functions The data collected by the CM from all DRP are stored in a 64 kByte output memory OM The OM of 15 crates build a chain which is connected to the main DAQ system through a fast readout sequencer and the FASTBUS memory module A special VME crate the databox contains modules which control the Mow of data from the CM to their destinations The readout scanner module RS feeds the data from the OM to the receiving memory module A lookup table flags crates that have to be skipped in the readout and crates whose output data should oe copied into a spy channel Three spy channels equipped with 256 kByte me
4. A barrel counter has a 320 mm long light guide on either end The counters are grouped in 16 pairs Each pair covers one hadron calorimeter sector Both sectors near the rails for the BGO calorimeter are covered only by two wider counters Thus in total there are 30 barrel counters The light guides are asymmetric so that the phototubes o two adjacent counters are close to each other In this way there is enough space for cables between pairs of phototubes he end cap counters are located in front of the end cap hadron calorimeter fig 80 The light guides have a 90 bend and extend to the end of the hadron calorim eter end caps There are 16 counters on either side of the detector each one viewed by a single tube The scintillator is 270 mm long 10 mm thick 275 mm wide on one side and 180 mm on the other It is perpendicu 10 E o 80 A 20 CO A A 50 A 50 CO 4 a D 10 A 90 C02 u a 4 S ag A 100 CU w lt t 5 w i 1500 2100 2100 3300 HIGH VOLTAGE V Fig 78 Collected charge fc minimum ionizing particles as a function of the applied high voltage for different Ar CO mixtures B Adeva et al The construction of the L3 experiment joe point 1 A HA Endcap LEG I gt Gey I I scintillators OP ge Barrel gt gt Gy scintillators P Fig 79 Cut through the barrel and end cap hadron calorimeter indicating the location of the scintillation counter The
5. All of these tasks are performed by the monitoring system which consists of a network of VME crates using an OS 9 68000 operating system They have mul titask multiuser and stand alone capabilities and also communicate with the main online computer 4 10 Results of alignment verification Each of the octants is adjusted as closely as possible to an ideal geometry using the double set of straightness sensors and the laser beacon UV laser runs of 100 events are then taken using the eight different laser beam paths in each octant Reconstruction of the laser trajectories fig 47a should show straight lines The deviation from zero called sagitta which is a measure of our alignment accuracy is defined as S ex T325 2 Xmm where x Xmm and Xmo are the coordinates measured in the inner middle and outer chambers respectively Fig 47b shows measurements from 42 laser runs for one octant Errors on the points are obtained from the residuals of 100 event runs The resulting average of all runs 26 10 um is within the design limit of 30 um The error is derived from the scatter of results from the different runs thus taking systematic errors of the laser into account An independent way of checking octant alignment in the absence of a magnetic field uses cosmic muons Scintillation counter hodoscopes are positioned above and below the octant and moved to either end as desired Thc resulting coincidence trigger
6. The trigger processor searches for tracks originating at the interaction point It is composed of 120 CAMAC modules all controlled by a single source of synchro nous clock pulses The main module is the hit array HA LRS2376 a 1 kb memory with two modes of operation load and search corresponding to the two steps of the track finding procedure defined below Each muon chamber wire sends a signal to the 96 channel FASTBUS TDC The same signals are also sent via the FASTBUS module auxiliary connector to a personality card PC for the trigger purpose Inputs from two adjacent wires are ORed and stored in a hit memory The 48 hit memories are enabled for the duration of the chambers drift time 1 2 ps For the trigger the chamber wires are grouped into trigger cells which for the P chambers coincide with the physical chamber cells 16 to 24 wires while for the Z chambers the information from the adjacent two wires of each double plane are combined together to give a single trigger cell The personality card controller PCC is a FASTBUS module which accesses the PC and decides about the presence of a track in each trigger cell For the P chambers the PCC groups the input data according to the trigger cell segmentation counts the number of fired hit memories within the trigger cell and compares it with a preset threshold number If the number of hits is above threshold the PCC encodes a 10 bit word with the layer number 2 bits the
7. a Three bridges support the v ires Three straightness monitors and actuators are schematically indicated Added is a schematic showing the details of the straightness monitor b The accuracy of the bridges cea be seen from this distribution of errors in the equidistant locations of the Pyrex glass pieces 52 B Adeva et al The construction of the L3 experiment Pyrex glass pieces were glued to carbon fiber supporting bars using a very precise Invar template Bridges thus produced have a surface to surface spacing of 101 500 mm All bridges were measured using an HP laser interferometer system with an accuracy of 2 pm The measured rms absolute position accuracy for all 255 bridges produced was 5 2 ym fig 39b There are three bridges per chamber One bridge at each end precisely positions the wires the bridge in the middle reduces the sag of the wires by a factor of four fig 39a The two end bridges are positioned with respect to external reference surfaces The wire planes are put into position in the chamber and adjusted in length so that the vibrational frequency of the first m AVY i Wn NS zL U M N NN 903 8 905 2 905 7ns 905 8ns 905 8 ns Argon Ethane 62 38 m SIGNAL WIRES W 30m dio 130g FIELD WIRES Cu Be 75 m dia 385g _ m 9058 ns 9057ns 22 9mm Spacing g LN Y S s 50 8 50mm a
8. ing chambers form the layer B The last modules on each side of the barrel do not have a layer A since they can only be reached by particles from the interaction region which have crossed the A layer of other modules fig 117 The resulting segmentation is 16 in and 1 layer A or 13 layer B in 6 with a total of 384 channels These signals are digitized in 4 8 us with a 10 bit range and 50 MeV bit CALQO METRIC TRIGGER SIGNAL SEGMENTATION iN RO VIEW Barre HC Endcop HC TEC Al Leminogity Teel Monitor o t et Be 81 NL Fig 117 Detector segmentation for trigger signals 94 B Adeva et al The construction of the L3 experiment The data digitized by the FERA are read at 120 ns word via the front panel bus the data from two groups of FERA are multiplexed into a single bus and go at 60 ns word through memory lookup units MLU t RS2372 which equalize the energy scale for all the calorimeter parts and remove the coherent noise contri bution The output of these MLU requires only 12 bits and one of the highest bits is programmed to give a signal hit if the energy recorded in the single trigger cell is above a given threshold different for each calorimeter and function of These signals are used by the hit counting trigger as described below The energy data from the MLU are presented on a bus to the second part of the processor at one word every 60 ns The operations of su
9. omputing for Experiments CERN 1988
10. surement We see exceilent agreement between all three methods and find all octants to be well within the range of specification of 30 pm 4 11 Conclusion The L3 precision muon detector is unique in its conception and ability to detect dimuons with 1 4 57 R M S 400um 740 EVENTS EVENTS No cuts SAGITTA b Fig 48 a A cosmic ray track reconstructed in the middle chamber b Sagitta distributions measured without cuts and with 2 mrad cut O COSMICS D LASER pn s e Mh OCTANT 16 poe OCTANT 15 cron E OCTANT 14 OCTANT 13 OCTANT 12 OCTANT 11 OCTANT 10 OCTANT 3 OCTANT 8 OCTANT 7 OCTANT 6 Mar c OCTANT 5 ur c E OCTANT 4 E DAC OCTANT 3 oe INE OCTANT 2 E ui OCTANT 1 40 20 0 20 40 Sagitta microns Fig 49 Compilation of alignment results or all 16 octants The zero sagitta prediction is the setting of the opto mechani cal system UV laser verification results are shown as squares circles indicate the center of cosmic ray distribution measure ments A 7 jum systematic error on each laser measurement IS not shown 58 B Adeva et al The construction of the L3 experiment Fig 50 Perspective view of the L3 hadron calorimeter mass resolution at the Z mass The critical feature that of detecting infinite momentum straight tracks with lt 3 um sagitta error has been demonstrated by three independent methods for all 16 productior
11. 70 PARALLEL DRIFT VELOCITY 60 um nsec b 50 40 sob Z x 20 x Pd 0 250 500 750 1000 1250 ELECTRIC FIELD V cm Fig 40 a Electric field lines in a drift cell of an inner or outer chamber are shown with a 0 5 T magnetic field parallel to the wires Drift times refer to a track 44 mm from the signal wires b Drift velocity computed as a function of the drift field COSMIC MUONS 20 mV Threshold 2 0 2 mm 300 b t 209 2 ET M usd C 7 al Lc cox 100 Dues cuu e eec 0 10 20 30 40 50 DRIFT DISTANCE mm Fig 41 a Distribution of residuals for cosmic muon tracks fitted to 14 of 16 wire measurements b Resolution as func tion of the drift distance measured in 0 5 T with a test chamber harmonic is 27 85 0 2 Hz for signal wires This en sures equal gravitational sags of 95 um for all signal wires Three internal alignment systems are integrated into the structure of the bridges fig 39a This threefold alignment system 4 consists of LED lenses and anaodrant ahatandindoac Tiaht fenes an FET manntal nn UCECEVA P CLIE pirstvUUuduvUnuwo IrL R1AJ311 C1 L E J 1 7 1210 7 L111 8 A UII one end bridge is focused by the lens in the middie bridge onto the quadrant photodiode at the opposite end bridge A displacement 6 of the middle bridge moves the image by 26 on the quadrant diode The imbalance of the photodiode output measures this dis placement Each set has been individually cal
12. 8 lt 35 and 145 lt O lt 174 5 over the full azimuthal range 0 lt 6 lt 360 x The solid angle covered by the endcaps 18 of 4 Fig 60 Readout system for the hadron calorimeter extends the coverage of the hadronic calorimetry to 64 B Adeva et al The construction of the L3 experiment 99 5 of 4n Fig 64 shows a longitudinal R z cut of and HC3 Each ring is split vertically into half rings the central detector while fig 65 shows a perspective resulting in a total of 12 separate modules The mod view of the HCEC The HCEC consist of three separate ularity of the HCEC detectors permits their fast rings an outer ring HC1 and two inner rings HC2 withdrawal to provide access to the other L3 central Table 4a Hadron barrel thickness in units of nuclear interaction lengths for pions BONONIE EM NNNM CMM CM dm M E ME LS Material Differential interaction length Integral interaction length BGO 0 93 Scintillator 0 01 0 94 HC barrel long short long short Fe stainless steel 0 37 0 37 Cu brass 0 70 0 62 U absorber 2 42 2 13 CHO Mylar 0 03 0 03 4 46 4 09 Calorimeter subtotal 3 52 3 15 Muon filter Cu 1 03 5 49 5 12 Support tube stainless steel 0 52 6 01 5 64 Table 4b Hadron barrel thickness in units of radiation lengths Material Differential radiation length Integral radiation length BGO 21 43 21 43 Scintillator 0 02 21 45 HC barrel long short long short Fe stainless steel 3 22 3 22
13. An assembled chamber is shown before it is closed by a Z layer 51 4 2 Precision chambers The momentum measuring or P chambers are constructed of two cast and machined aluminum end frames and two extruded aluminum side panels The inner and outer chambers are closed on the top and bottom by Z chambers The middle chambers are closed by honeycomb panels to avoid degradation of the reso lution due to multiple scattering An exploded view of the mechanical structure of one MO MI chamber is shown in fig 37 and fig 38 shows a photograph of an assembled chamber just prior to Z chamber mounting There one can see the 5 6 m long wire planes positioned by end bridges inside the gas tight box of end frames and side panels The insert in fig 37 depicts the double plane configuration of the Z layers Each P chamber contains about 320 signal wires and a total of 3000 wires The signal and field shaping wires are positioned to about 10 um in the magnetic bending direction and to better than 40 pm in the nonbending direction by precision Pyrex glass and carbon fiber bridges fig 39a These bridges have very small thermal expansion coefficients 1 5 ppm C so that tempera ture effects on the wire positions are negligible The QUADRANT RECEIVER QUADRANT DIODE v 1200 t E Q e ENTRIES 4623 cc RMS 5 2 um gt 800 h LL Oo cc ua coa 2 400 01 0 05 0 0 05 01 Error center of pyrex plate mm Fig 39
14. E GENNARI 29 S GENTILE 9 M GETTNER 9 C GIRARD M GLAUBMAN 9 S GOLDFARB gt Z F GONG E GONZALEZ 9 A GORDEEV 2 Yu GORODKOV P G OTTLICHERP C GOY M GOYOT G GRATTA A GRIMESP C GRINNELL 1 M GRUENEWALD M GUANZIROLI 9 S GUERRA 9 C GUILLON A GURTU D G SEWELL H R GUSTAFSON M HAENSLI 3 M HAAN C HALLER T HAMACHER H HAMMERS P K HANGARTER S HANCKE M HARRIS D HARTING F G HARTJES C F HE 2 A HEAVEY 9 T HEBBEKER M HEBERT 2 R HELLER Ch HELMRATH J HERRMANN G HERTEN U HERTEN A HERVE H HESSER G HILGERS K HILGERS H HOFER H HOFER 33 M HOFER 32 T HOFER gt F HOFFMANN U HORISBERGER L HORVATH gt L S HSU G Q HU 2 B ILLE M M ILYAS 5 G IMPROTA 22 V INNOCENTE E ISIKSAL E JAGEL 15 BN JIN 9 L W JONES M JONGMANNS H JUNG 9 P KAARET 25 O KAELIN W KAESTLI gt Yu KAMYSHKOV 2 D KAPLAN 9 W KARPINSKI P Y KARYOTAKIS W KERTZEK V KHOZE G KIRCHHOFF D W KITTEL A KLIMENTOV P F KLOK M KOLLEK M KOLLER 3 A C KONIG 2 O KORNADT V KOUTSENKO 2 RW KRAEMER V R KRASTEV 9 A KRATEL W KRENZ A KUHN 32 A KUNIN 2 S KWAN J LACOTTE M LaMARRA 9 G LANDI W LANGE 2 K LANIUS D LANSKE S LANZANO 1422 J M LE GOFF 5 14 J C LE MAREC D LEA 25
15. The answers from the three HA are collected at one place and the track is identified if a coincidence condition 3 fold or majority of 2 is satisfied between the three layers The large bending of the low momentum muons can cause the road pattern to interest two neighboring octants For these roads the answers of the HA corre sponding to the two octants are collected together In the R Z plane Z chambers there are four chamber layers and a similar procedure is followed The follow ing conditions give a trigger Single muon trigger muon track is defined in the R 6 plane as a coincidence of all three chamber layers and in the R Z plane as a coincidence of all four chamber layers At least one octant should have a track identified both in the R and R Z planes This trigger is effective in the region 44 lt lt 136 where the chambers have a complete O coverage Di muon trigger A muon track is defined by a twofold coincidence of any two layers both in the R and R Z planes but at least two octants should have a track identified and the two tracks should satisfy a coplanarity condition This trigger is useful for the larger angular region 36 lt lt 144 where the angu lar coverage of the chambers is incomplete Small angle muon trigger A muon track is defined by the presence in the R plane of a single hit in the inner P chambers and in the R Z plane by a coinci dence of both layers of inner Z chambers Furthermore
16. The TEC oper B Adeva et al The construction of the L3 experiment Z DETECTOR PSF CALIBRATION INNER SECTOR QUTER SECTOR 7 BERALIUM TUBE Fig 103 General view of the central track detectors ates with 80 CO and 20 iC H which has a low longitudinal diffusion and thus permits a low drift speed Furthermore this gas has a negligible Lorentz angle To reach the ultimate resolution for drift lengths of up to 5 cm determination of the drift time by a center of gravity method is mandatory Thus the anode pulses are sampled by Flash Analog to Digital Con verters FADC after a shaping of the analog pulses to cancel the ion tail This principle has been tested by prototype chambers in test beams and in the MARK J experiment at PETRA 46 48 11 2 1 Mechanical construction The TEC wires are supported by two end plates with precisely drilled holes for the anode potential and field shaping wires Furthermore the plates contain the gas inlets and outlets and the grooves for the positioning of the grid wire planes fig 104 The end plates are held apart and the enclosure is gas sealed by a 4 mm thick aluminum cylinder at the outer diameter and a 1 5 mm thick beryllium cylinder at the inner diameter The structure has been studied by the finite element tech nique to meet the required precision of about 10 pm for the wire positioning and to minimize the amount of material traversed by the particles entering the
17. This is an average over all positions and slopes we expect for high momentum tracks 53 4 3 Z Chambers Z chambers 5 consist of two layers of drift cells offset by one half cell with respect to each other to resolve left right ambiguities Each cell has two parallel aluminum I beams fig 37 connected to 2 4 kV and one gold plated molybdenum anode wire with 50 pm diameter at 2 15 kV in the center The cell is closed by two aluminum sheets at ground potential and iso lated from the I beam profiles by fiber glass strips The Z chamber gas mixture 91 5 argon and 8 5 methane was chosen because it is not explosive The drift velocity averaged over the cell is about 30 m ns The measured resolution both in a test beam with a prototype and with cosmic rays in production cham bers is typically 500 pm Since these chambers are the covering elements of inner and outer precision cham bers tight dimensional tolerances were needed to ensure mechanical fitting and thus precise machining was re quired on the Z chamber frames Relative wire spacing also depends on these frames Aside from that the design and technical specifications allowed for a rather simple construction All of the 96 Z chambers dimen sions about 6 m X 2 m were built in two years with a production rate reaching six chambers per month at the end of the second year 4 4 Octant stands Octant stands are precision structures supporting the chambers and maintai
18. Cu brass 8 61 7 61 U absorber 87 31 76 78 CHO Mylar 0 07 0 06 120 70 109 16 Calorimeter subtotal 29 25 87 71 Muon filter brass 11 54 132 24 120 70 Support tube stainless steel 5 78 138 02 126 48 Table 4c Hadron barre energy loss for mind ionizing particles in MeV Material Differential energy loss Integral energy loss g BGO 221 221 i Scintillator 2 223 CH HC barrel long short iong short Fe stainless steel 83 83 Cu brass 159 140 U absorber 578 509 CHO Mylar 5 5 1048 960 Calorimeter subtotal 825 737 Muon filter Cu 213 1261 1173 Support tube stainless steel 118 1379 12 1 B Adeva et al The construction of the L3 experiment Table 5 Detector thickness at different angles in units of nuclear ab sorption lengths for pions BGO HB HC MF ST Total 90 0 0 94 3 52 6 99 0 52 5 97 82 0 0 94 3 56 1 00 0 52 6 02 72 5 0 94 3 70 1 34 0 54 6 22 66 5 0 94 3 84 1 08 0 54 6 42 58 5 0 94 4 28 1 16 0 60 6 98 54 0 0 94 4 56 0 49 0 64 6 63 50 5 0 94 4 54 0 52 0 67 6 67 48 5 0 94 4 35 0 25 0 69 6 23 42 5 4 92 0 75 5 68 390 5 22 _ 0 41 5 63 35 0 5 12 1 05 0 45 6 62 320 3 74 1 68 0 48 5 90 2710 1 97 3 13 0 47 5 67 220 _ 6 36 6 36 170 6 01 6 01 120 5 91 5 91 HB hadron calorimeter barrel HC hadron calorimeter endcaps MF muon filter d ST support tube XXXXX detector components The HC1 half rings are separated at 90 270 b
19. E 3 x 20 L Time days PEA AAE APR RNC ay sau iu E RE mee IERI 0 5 10 15 Fig 44 a Principle of referencing the octant center line directly to the wires of the three chambers b A vertical alignment piece assembly with a lens for the middle cham bers c Relative position of the middle chamber in an octant monitored over a two weeks period 4 6 Laser beacon The vertical alignment systems guarantee that the chambers line up at each end of the octant but the two octant center lines must also be parallel to each other We use a laser beacon 9 to measure the degree to which the two ends of the octant are parallel A He Ne laser beam which is reflected by 90 by a highly 55 accurate rotating pentaprismatic mirror assembly sweeps out a plane to an accuracy of better than 30 um The deviation of the octant center lines from this reference plane is measured by six position sensors multichannel photodiode arrays attached directly to the three ele ments of each vertical straightness monitor The laser beacon fig 45 can measure the angle between the two octant lines to better than 25 prad corresponding to an error in the sagitta of less than 10 um The MO and MM chambers are adjusted so that this measured angle iS zero 4 7 UV laser Each of the 16 octants contains a two stage nitrogen ultraviolet laser 10 11 which is operated under com puter control The laser beam is directed up and across the top of th
20. F CESARONI 29 Y H CHANG 1D J W CHAPMAN 3 M CHEMARIN 13 A CHEN 3 C CHEN 2 H F CHEN H S CHEN 9 M CHEN M L CHEN 19 S R CHENDVANKAR G CHEVENIER S CHIDZIK 3 G CHIEFARI 22 C Y CHIEN F CHOLLET 4 M CHUMAKOV 2 C CIVININI I CLARE R CLARE n G COIGNET N COLINO 9 V COMMICHAU D G CONFORTO P CRISTOFORI 9 F CRUNS 2 X Y CUI 515 T S DAI J R D ALESSANDRO 12 M DANIEL 20 X DE BOUARD at B DEBYE G DECREUSE A DEGR K DEITERS 3 E DENES P DENES 25 F DeNOTARISTEFANI 9 M DEUTSCHMANN M DHINA 33 B DIDIERJEAN M DIEMOZ 2 M DIETRICH 1 H A DIMITROV 29 C DIONISI 9 F DITTUS M DOHMEN R DOLIN 5 J F DONAHUE 9 A DONAT E DRAGO K H DREGER T DRIEVER 2 G DROMBY P DUINKER L DURAN 1 M ELKACIMI H EMAMOUNI A ENGLER F J EPPLING F C ERNE 2 I ERNE 3 H ESSER P EXTERMANN 9 R FABBRETTI G FABER S FALCIANO 9 T FALK 25 S J FAN 29 M FAVRE 3 J FAY S FEH R 9 J FEHLMANN gt M FELDMANN 2 H FENKER T FERGUSON M FERNANDEZ 29 F FERRONI 9 H FESEFELDT J FIELD 9 J M FIGAROLA 29 C F FIGUEROA 2 G A FILIPOV 9 B FOLIGN G FORCONI 9 T FOREMAN V FRANZKE W FREI 3 K FREUDENREICH W FRIEBEL M FUKUSHIMA G GAILLARD M GAILLOUD 9 Yu GALAKTIONOV 2 E GALLO S N GANGULI D GARELICK 9 S S GAU 3 G GAVRILOV
21. P G SEILER J C SENS 2 L SHEER V SHEVCHENKO 2 S SHEVCHENKO X R SHI K SHMAKOV v SHOUTKO E SHUMILOV 2 R SIEDLING N SMIRNOV 9 V SOUVOROV C SOUYRI I SPANGLER T SPICKERMANN B SPIESS 3 P SPILLANTINI R STAROSTA M STEUER 49 D P STICKLAND 9 B ST HR 2 H STONE K STRAUCH 9 K SUDHAKARP R L SUMNER H SUTER R B SUTTON 2 H SZCZESNY J TANG 5 X W TANG 9 E TARKOVSKY 2 A TAVENRATH V TCHUDAKOV 2 J M THENARD E THOMAS T THON H THUERIG gt M THULEN C TIMMERMANS Samuel C C TING S M TING F TONISCH 22 Y P TONG M TONUTTI S C TONWAR J TOTH W TOTH G TROWITZSCH K L TUNG J ULBRICHT L URBAN E VALENTE 9 R T VAN DE WALLE H VAN DER GRAAF V VANZANELLA 22 M VERGAIN I VETLITSKY 2 H VEY 9 G VIERTEL gt M VIVARGENT H VOGEL 9 S VOLKOV M VOLLMAR H P VON GUNTEN I VOROBIEV 2 A VOROBYOV 9 L VUILLEUMIER 9 S WALDMEIER 22 W WALK 9 W WALLRAFF C Y WANG G H WANG J H WANG 9 Q F WANG 19 X L WANG Y F WANG 2 Z M WANG Z M WANG D WASSENBERG D WEGMANN 25 R WEILL 9 T J WENAUS P WENGER J WENNINGER 9 M WHITE R WILHELM 2 C WILLMOTT 2 H P WIRTH F WITTGENSTEIN RJ WU 9 S X WU 19 Y G WU 9 B WYSLOUCH E y XI SD ZZ XU Z L XUE 2 D S YAN K D YANEV 9 B Z YANG
22. between chambers of the same octant fig 36 To achieve the design resolution systematic errors in the internal octant alignment must be kept below 30 um As described later this requires complex optical and 5 fo eim ja 3 Fig 33 Schematic view of the assembled muon detector Z CHAMBERS o P CHAMBER _ zit E SER AMPLIFIERS zz I 4 T A m ag I ML o A WS v d ad CHAMBER ae eo Fig 34 An octant module is attached to the supporting torque tube Precision chambers Z chambers octant support stand amplifiers and cables gas system and the UV laser calibration system are also shown 2 9 m 1 I t s 96990980 5999800292 l6 wires 24 iS 250um l_ 2 3 JN Fig 35 Measurement errors on the sagitta of a muon trajec tory curved in a magnetic field QU ER CHAMBER MO A 5 nA d i A s A Fig 36 End view of the three chamber layers in one octant with a schematic view of the alignment system and a detail of the middle chamber B Adeva et al The construction of the L3 experiment NER J X ae cerroensis IU U VIVENTE A ME v n I ENDFRAME Fig 37 Exploded view of an outer MO precision chamber and detail of Z chamber mechanical measurements as well as UV laser and cosmic ray verification Sagitta errors less than 30 pm have been achieved in all octants Fig 38
23. can be realized with a 16 kbit RAM addressed by the 14 data bits from this single d bin The lower limit of 42 on is set by the TEC geometry tracks with smaller polar angles miss the outermost wires Tracks with gt 42 and P 150 MeV c cross up to three adjacent bins The corresponding SD modules provide therefore 5 x 14 bits of information i e one time 14 bits from the reference bin plus two times 2 X 14 bits from the two pairs of bins adjacent to the central one To keep the size of RAM memory to a manageable level it is necessary to reduce this informa tion The 14 bits of each 6 bin are grouped radially two by two The grouping is an AND or OR independently programmable for each pair of bits If the background conditions are good all pairs can be set to OR to maximize trigger efficiency If background conditions are bad all pairs can be set to AND so as to increase the rejection of spurious tracks As a compromise if the background is localized at small radii the signals in this region can be set to AND the others at larger radii to OR Forward tracks with gt 25 and P 100 MeV c The search for small polar angle tracks is performed with the seven innermost wires The al gorithm described above is applied but no AND or OR is now necessary Since the same masks are used the P cut is lower for the forward tracks Because of tle difference in maximum drift time between the inner and outer wires the sea
24. cell number 7 bits and one control bit For the Z chamber the processing is simpler since a trigger cell corresponds to a single hit memory The data are processed by the PCC in parallel for all octants and therefore a total of 16 buses 8 for the P 2nd 8 for the Z cnambers feed the trigger data into the trigger processor at the same time In the load step the hit addresses are presented by the PCC on each bus to a number of HA 4 for the Z chambers and 3 for the P chambers Only one HA whose layer number assignment agrees with the layer number presented on the bus picks up the cell number B Adeva et al The construction of the L3 experiment interprets it as a load address and writes 1 in the corresponding location of its hit memory For the search step a number of possible roads of tracks originating from the interaction point are prede fined Each road is parameterized by its central cell number and its half width in each one of the three chambers layers Each road corresponds to a certain region of the muon production angle in the trans verse momentum and the electric charge All the possi ble tracks with P gt 1 GeV c are defined Track finding starts after all the PCC have finished loading the trigger data A control stack presents road parameters of each layer to the respective HA and interrogates whether the HA finds any hits within the specified range In the R 6 plane the three layers o P chambers are used
25. chamber and reaching the electromagneuc calorimeter less than 195 and 10 of a radiation length respectively Fig 105 shows one of the end plates The anode wire planes are arranged radially forming 12 inner and 24 outer sectors Their sensitive length is 982 mm There are two types of sense wires standard wires to measure precisely the R coordinates of the tracks and charge division CD wires to determine the Z coordinates Additionally ror part of the standard wires groups of five grid wires on each side of the amplification region are read out pick up wircs Com paring the induced signals the left right ambiguity for these anodes LR wires can be resolved 49 The inner sectors include six standard and two CD wires while the outer sectors include 31 standard 14 LR and 9 CD wires 87 s CHARGE DIVISION ANODE Lo s FOCUS L SPACING BETWEEN WIRES ALL DIMENSIONS IN mm Fig 104 Technical drawing of the wire configuration in one inner TEC sector and in part of iwo outer sectors 11 2 2 Readout Fig 106 shows a schematic drawing of the analog part of the TEC readout All anodes are capacitively coupled to the same type of hybrid preamplifiers via HV protection circuitry The CD anodes have an in creased time constant for better signal to noise ratio The shapers type A are set to produce a symmetric Fig 105 One of the TEC end plates 88 B Adeva et al The construction of the L3
26. com posite structure weighs 140 kg for a total load of 10 t 9 1 3 Electronics Since the BGO calorimeter is operating in a 0 5 T magnetic field and space is at a premium conventional photomultipliers vannot be used Instead we use 1 5 em Hamamatsu S2662 photodiodes to detect the BGO scintillation light they are insensitive to the magnetic field and have a quantum efficiency of about 70 Each crystal has two photodiodes glued to its rear face The total diode capacitance is 230 pr for 15 V reverse bias Since they have unity gain each photon detected pro duces one electron hole pair a preamplifier must be added The signal from the photodiode is about 0 2 fC 1200 electrons for each MeV deposited in the BGO 33 The charge sensitive preamplifier 34 is mounted directly behind the crystal and uses a low noise high transconductance Toshiba 2SK147 FET in a cascode configuration The output pulse rise time is 300 ns corresponding to the BGO light decay time and the exponential decay time is 800 ps The rms random noise level of the photodiode and preamplifier combination IS less than 1000 elect ons This is much higher than the value that can be obtained with photomultipliers but significantly affects the resoluucn only for very small signals showers less than a few hundred MeV The 78 B Adeva et al The construction of the L3 experiment amplifier gain is extremely stable however and we do not expect any of the gain vari
27. computer a dedicated FASTBUS crate with a DSM which is on the same cable segment as the DSM of the central crate The data from each detector component can therefore also be monitored in spy mode on the detector compu er An important exception to this procedure is the TEC Data reduction processors DRP are used at the front end to reduce the large amount of data generated by this detector by a factor of 20 Since the DRP need 5 ms to process a typical event and transfer it to the buffer they are only started by the positive decision of the level 2 trigger One of the DSM on the central crate receives from the level 2 processors the trigger data together with the level 2 decision The GPM examines this decision If it is negative the event is simply erased from each subde tector DSM and from the front end memories of the TEC If it is positive the DRP of the TEC are started and the BM is instructed to merge the information coming from the different subdetectors and io send them to one of the three 3081 E computing units which are on its cable segment The central crate is connected to the VAX 8800 with the 4 MB s CERN host interface CHI The CHI can perform any FASTBUS action check and monitor col lected data and transfer them to the VAX 93 12 4 The level trigger The level 1 trigger is simply the logical OR of trigger conditions from different sources calorimetric trigger muon trigger TEC trigger and scintillator tri
28. control valves are checked about 2000 detector signals in total The a er Fig 14 View of a magnet door hinge B Adeva et al The construction of the L3 experiment magnet monitor ng system placed on the plug of the vertical shaft feeds genera information concerning the magnet behavior into the L3 slow control system 3 6 Field measurement Because of the large volume which is furthermore partially obstructed by the support tube the field mea surement has been divided into two parts The inner volume of the support tube was mapped with Hall plates 2 The remaining volume has been mapped with about a thousand magnetoresistors permanently in stalled on the muon chambers In addition five NMR probes monitor the absolute value of the field For the interior of the support tube a mapping device was used which had been developed to map the field of all four LEP experiments Two rods each equipped with 60 Hall plates rotate around the central axis The rotating mechanism is supported by the two rails on which the hadron calorimeter rests removed during the measurements Both the movement in the azimuthal as well as in the axial direction are under remote control The Hall plates are oriented in the Z main component r and 6 directions Each compo nent is measured twice The relative alignment within a pair and io other pairs is known with an accuracy of 0 2 mrad This allows one to determine the minor compo nents
29. data This provides a check of the data integrity at the event building stage The trigger word also provides information for event selection of the online raonitoring and is a guide for the leve 2 and level 3 computations Trigger control also provides tim ing signals for taking cosmic data or synchronizing test signal generators to calibrate the detector and test the DAQ system 12 5 The level 2 trigger The level 2 trigger scheme is shown in fig 116 The multiport multievent buffer MMB is an 8 events deep 256 words per event first in first out input memory 64 After each beam crossing all data 4 k 16 bits words delivered by the front end trigger digitizers are stored in parallel in the 60 ECL input ports at the speed of 60 ns per word The final and some intermediate results of all the level 1 trigger processors are also stored One of the ports receives the trigger word and the event number generated by the trigger control logic The levei 2 trigger receives also information which was not available in time to be processed by the level 1 In particular it receives charge and drift time information from the charge division wires of the TEC which allow to define the track coordinates in three dimensions 62 Data acquisition management is hard wired and synchronized by signals delivered by the level 1 trigger Ali the data banks connected to the input ports are overwritten at each beam crossing until a level 1 signal validat
30. if a track is found in the forward half of the detector another should be found in the backward half There is also the possibility to accept two tracks both in the forward or backward half of the detector This trigger covers the small angle region 36 O 44 136 lt 144 where oniy one layer of P chambers is available 12 4 3 TEC trigger Fer each of the 24 outer TEC segments the trigger uses 14 wires out of 54 62 The data from all TEC segments are processed in parallel a segment divider module SD divides the total drift time into two bins 95 and solves the left right ambiguities by means of sig nals induced on the grids located at the right and left of the anode plane 63 In this manner the R plane is subdivided into 96 bins Thus after processing by the SD modules the level 1 TEC trigger information con sists of a 96 x 14 bit matrix For each bin a track finder module TF searches for the presence of tracks which leave the TEC in this bin and originate at the beam line The 96 sets of 14 signals are used to address random access memories RAM which contain the topologies of all relevant tracks allowing for the effects of inefficiencies and additional hits The TF perform the track search in parallel in less than 1 ps Three types of tracks are defined Tracks with polar angle O gt 42 and transverse momentum P gt 6900 MeV c are contained in a single bin and the track search
31. measuring the rate of Bhabha events It Hadron Calorimeter Endcap Planar hamters 16cm 12 cm Fig 99 The L3 lumirosity monitor 84 B Adeva et al The construction of the L3 experiment Fig 100 An end view of the BGO crystal array is located in an angular region forward enough to become independent of the Z exchange and yet not too far forward so as to allow easy Bhabha event selection unaffected by systematic errors It consists of a charged particle tracking device with good position resolution followed by a highly segmented BGO array of good radiation hardness to measure the energy of the shower ing electrons and photons With this system one can study in detail the phabha process including the radiative tail The trigger 42 will permit the measurement and rerm oval of ba kground events like beam gas interactions Offline analysis will remove the Bhabha events that develop only a fraction of their energy in the BGO detector but otherwise pass the trigger condition By comparing the tracking infor mation with the energy profile deposited in the crystal array one can define a very precise geometrical accep tance region An asymmetric software cut will reduce the systematic effects of variations in the LEP beam parameters The BGO array is cylindrically symmetric The crystals are arranged in eight rings fig 100 each covering 15 mm radially parallel to the beam pipe Azimuthally they are arrange
32. mixture at 2 bar gives an optimum between resolu tion and running conditions A Z coordinate resolution of about 2 cm was re ached by charge division The same R 9 resolution was obtained from the CD wires as from the standard anodes The left right ambiguity is resolved by compar ing signals induced on the neighboring grid wires 49 These signals can also be processed and read out by the FADC DRP system and be used in the trigger system Because of the restricted radial extension of the TEC the two track resolution is important It was studied by superimposing single hits from test beam events Beyond a separation of 450 um in drift distance fuily efficient two track resolution can be expected 53 11 2 6 Results from test beam The complete detector system has been tested in a pion beam at CERN during March 1989 The compo nents were operated under standard conditions as fore seen for LEP operation including proper temperature and cable lengths The quality of the TEC signals was monitored by switching several DRP to raw data mode to display the full DRP memory content from the evenis Fig 113 demonstrates the excellent performance of the digitized signals after pulse shaping from a multitrack event The RUN 433 EVENT 583 0 200 400 600 800 1000 FADC 2 Row DRP 1118 soft 15201045 Fig 113 Display of FADC raw data for a multitrack event and the time marker sign extreme right The triangles indicate the
33. modules This design can be extended to mea ure muons at much higher energies 5 L3 barrel hadron calorimeter 5 1 Motivation and overall structure The energy of hadrons emerging from e e colli sions 1s measured in L3 by the total absorption tech nique calorimetry with the BGO crystals and the uranium hadron calorimeter The uranium hadron calorimeter has two parts the barrel part and the forward backward part The hadron calorimeter barrel covers the central region 35 lt lt 145 it is a fine sampling calorimeter made of depleted uranium ab sorber plates interspersed with proportional wire cham bers it acts as a filter as well as a calorimeter allowing only nonshowering particles to reach the precision muon detector Uranium has a short absorption length thus maxi mizing the amount of absorber material in the available radial space The uranium radioactivity imposes strin gent requirements on the construction and the oper ation f the calorimeter but it also offers a built in gamma source for the calibration of the wire chambers 14 We choose gas wire proportional chambers as detectors because they are stable reliable can operate in the magnetic field and are relatively easy to produce on a large scale Moreover in a multiwire detector the wires can be grouped in any readout pattern By orient ing the wires in alternate chamber planes at right angles to each other better determination of particle tr
34. of the field with the sum of differences al gorithm even in the presence of a misalignment of ihe gear 2 The volume occupied by the muon chambers has been mapped with magnetoresistors main component only Magnetoresistors were chosen because they are economical stable in time and since they are sensitive to B need only to be calibrated for one polarity of the field Their temperature dependence was taken care of by adding compensating resistors in parallel The distri bution of the magnetoresistors on the muon chambers was done in such a manner as to have a measurement whenever the value of the main component changes by 40 G To facilitate their installation on the muon cham bers up o ten magnetoresistors were mounted in a chain in aiuruaum housings up to 1 m long These magnetoresistor ensembles were then calibrated in a smal solenoid 3 7 The detector support structure The support tube ST is a 32 m long 50 mm thick 4 45 m outer diameter heterogeneous tube fig 15 with a flange support at each end to transmit the load to the ground The part of the ST which is inside the magnet 14 1 m is of nonmagnetic stainless steel with a 4 6 m long octagonal double walled central section fig 16 The remaining portion is of carbon steel Each flange rests on two servo controlled mechanical jacks to allow B Adeva et al T e construction of the L3 experiment Fig 15 The support tube with the two captive torque tub
35. overheads of routing through intermediate hosts and terrestrial lines minimizing propagation de lays Except under heavy load the switching delays are still small compared with the cable transit times for coast to coast or transatlantic calls Remote use of full screen facilities such as the VAX TPU editor or iBM VM CMS over 11000 miles of LEP3NET cable is relatively comfortable even though the 200 ms delay for remotely echoed characters is quite visible Witn the connection of L3 s LEPICS IBM 3090 to LEP3NET in February 1988 it became evident that full screen access to LEPICS using character by character remote echoing often places a heavy load on LE 3NET Even with an unloaded network propagation delays maxe this mode of access slow A software package support ing IBM 327x full screen terminal emulation on VAX compuiers has been obtained The package maintains a local scieen image on the VAX at the user s site and only changes are transmitted to and from the IBM whenever the enter or attention key is pressed The package also provides extensive support for key map ping which is needed to make remote use convenient on non IBM remote terminals by emulation of the PF permanent function keys whose use is an integral part of working on an IBM VM CMS computer system 100 B Adeva et al The construciion of the L3 experiment 13 3 2 Electronic mail Electronic mail over LEP3NET has removed almost all need for p
36. shows the light yield versus the uniformity parameter R defined as the relative variation of the collection efficiency of the light produced at the crystal extremities These measure ments were made with cosmic rays for all crystals of the first half barrel Limits were set on these parameters to avoid low light output tails or light collection curves too far away from the optimum value Crystals out of limits were corrected by paint additions 9 1 2 Mechanical structure The mechanical structure of the barrel figs 87 and 88 bears the weight of the 7680 crystals the preampli fier boards and the corresponding cables monitoring devices and cooling circuitry To achieve the best solid angle coverage and to minimize dead spaces between crystals the structural material is confined to thin walls 5 ay q S gt t Cn 5 s 0 30 20 W O 10 R Fig 86 Light output versus uniformity parameter R for all the crystals of the first half barrel The average light output is 14 4 3 8 x 10 e cm and the average value of R is 6 9 3 8 The best energy resolution is obtained for R values between 0 and 10 around the cells and to a cylindrical inner tube attached on each side to a conical funnel trumpet which carries the weight and transmits it to the four bearing pads The whole structure was studied on a computer aided design system 29 For the calibration of the calorimeter it is necessary to
37. the chamber arrangement with the 22 5 stereo angle of the wires b Shows details of the chamber construction the depleted uranium plates within a U or V absorber layer of a HC1 and HC2 half ring An absorber layer is defined to be of the V type when it follows a V type chamber layer The gaps between plates do not coin cide in successive layers and they do not point to the beam axis nor do they coincide with the gaps between chambers compare figs 66 and 67 see also fig 65 6 2 4 Wire grouping The wire signals are grouped to form towers pointing to the interaction region In the R z plane the Fig 67 Hadron calorimeter forward backward system The figure shows the uranium absorber plate arrangement for the HC1 and HC2 modules Displayed is the arrangement for the U layers For the V layers the plate arrangement is mirror symmetric 68 B Adeva et al The construction of the L3 experiment detector is segmented in 31 pointing roads of width AG 1 by grouping two wires or their equivalent at the z position of the first HC1 chamber layer Signal towers are then formed by subdividing each of the three end cap rings in depth and grouping the wires within each road To retain the stereo angle informa tion separate towers are formed for the U and V type layers The azimuthal segmentation is therefore Ad 22 5 In depth HC1 is divided into a total of seven logical segments fig 68 the first tw
38. their tracks determination of the transverse momentum and the sign of the charge for particics up to 59 GeV c reconstruction of the impact point and direction for charged particles at the entrance of the electromag netic calorimeter determination of the track multiplicity originating from the interaction region at the trigger level reconstruction of the interaction point and of sec ondary vertices for particles with lifetimes greater than 107 s These goals and the limited space available for this detector within the electromagnetic calorimeter have determined its design The total lever arm available for coordinate measurements in the chamber is 37 cm ra dially The charge identification of 59 GeV c particles with 957 confidence requires 50 coordinate measure ments with 50 um resolution This is accomplished by two concentric cylindrical drift chambers on common end plates operated in time expansion mode the TEC surrounded by two cylindrical proportional cham bers with cathode strip readout the Z detector fig 103 11 gt d or Following the TEC principle the high field amplifi cation region at the sense wire piane is separated from the low field drift region by an additional grid wire plane This configuration allows to optimize the elec tron arrival time distribution as well as the track length seen by the individual anode wires independently of the drift velocity chosen in the drift region
39. uranium by an order of magnitude The 50 um diameter gold plated tungsten anode wires are crimped into gold plated brass jacks which in turn are al GAS CONNECTOR PLASTIC END PIECE SEE FIG amp b E GAS MANIFOLD Le p PLasti PLUG I MYLAR FOIL SHIELOING PLATE 1 i SHIELDING PLATE hi PLASTIC END PIECE v2 BRASS JACK ee Fig 52 Proportional chamber of the hadron calorimeter bar rel 59 28 o 0 04 24 20 12 THOUSANDS OF WIRES o O e r rope 0 4 0 6 0 8 1 1 2 1 4 1 6 RELATIVE GAIN Fig 53 Relative gain distribution for the wires of the cham bers fitted into plastic end pieces The tension in the wire is 250 g with an rms spread of 17 g The gas is supplied in parallel to all tubes via two channels which are incorporated into the end pieces The gas inlet and outlet are diagonally opposite to each other to ensure adequate flow of gas through all tubes In designing the calorimeter particular attention was paid to minimizing the size of the dead regions such as the chamber end pieces various support structures and space for the services To achieve this 53 different sizes of chambers are used with the number of tubes per chamber ranging from 33 to 58 For the same reason the chambers are operated with the anode wires at ground potential avoiding the use of numerous capacitors All 371764 wires of 7968 chambers were tested in the production line 15 and showed good uniformit
40. yellow report 88 06 voi 2 p 896 44 J F Crawford et al Nucl Instr and Meth 127 1975 173 45 F A Berends and R Kleiss Nucl Phys B228 1983 537 46 H Anderhub et al Nucl instr and Meth A252 1956 357 47 H Anderhub et al Nucl Instr and Meth A263 1988 1 102 B Adeva et al The construction of the L3 experiment 48 H Anderhub et al Nucl Instr and Meth A265 1988 50 49 A B hm et al Nucl Instr and Meth A273 1988 471 50 U R ser et al A shaping amplifier for high resolution drift chambers operated at low drift velocity with Flash ADC readout in preparation 51 H Akbari et al Multianode readout of scintillating fibers for L3 Vertex Chamber calibration to be published in Proc Scintillating Fiber Workshop at Fermilab 1988 52 H Akbari et al Performance of the multianode readout of scintillating fibers used for calibration of the L3 vertex chamb f in preparation 53 G Viertel Research and Development on Time Expan sion Chambers Invited Talk given at the Int Symp on Position Sensitive Detectors in High Energy Physics Dubna USSR 1987 54 K Deiters The Z detector of the L3 experiment ibid 55 W Friebel et al PHE 87 08 Zeuthen 1987 56 H Muller and R Vachon CERN EP Electronics Note 85 02 7 L Pregernig The Block Mover presented at the 1985 JEEE Nuciear Science Symposium San Francisco CA USA 58 H Muller C M Story an
41. 0 300 mean pulse height Fig 69 Hadron calorimeter forward backward system Mean pulse height distribution for ail 2284 chambers mounted into the endcaps 6 mean 0 03 measured with cosmic muons ment is as follows Pressure AQ QA p 0 6 mbar Temperature AQ QAT 1 85 C 5 High voltage AQ QAU 1 2 V Particular attention was given to the long term be havior cf the proportional chambers under radioactive load of the uranium absorber An accelerated aging test has been carried out with a strong 12 Ru source No change in the performance within 3 has been found up to a corresponding lifetime of 800 yr in the uranium environmen of the end caps fig 70 coulomb cm 05 0 6 0 7 0 6 09 10 400 500 600 700 800 70 B Adeva et al The construction of the L3 experiment 6 3 2 Calibration and monitoring An absolute calibration of the calorimeter with cosmic muons has been performed using scintillation trigger counters above and below the end cap modules Additional concrete shielding resulted in a lower muon cutoff energy of 2 GeV A face to face cut was applied to the raw data demanding a coincidence of three consecutive vertically arranged towers to ensure that the center tower was traversed by the muon vertically in full length Preliminary results for one half ring arrange ment one HC1 HC2 and HC3 module are shown in fig 71 The mean pulse height of the center towers corresponds to 45 ADC channels with
42. 2 GeV c Chamber resolution 400 pm Absolute momentum error from bending angle 4x107 magnets to insure that the design parameters were re spected The measured parameters of the spectrometer are given in table 11 The second requirement is to have a geometrical precision in the crystal alignment with the nominal beam direction and position such that no correction should be introduced in the evaluation of the energy deposited in the crystal This requirement is translated in a positioning of the geometrical center of the crystal front face better than 1 mm and an angle between the beam and the crystal longitudinal axis less than 5 mrad These limits were achieved by the mechanical precision of the supporting table and by the electronic controls of the servo motors Finally during the calibration the half barrel was enclosed in an air conditioned tent with a stabilized nominal temperature of 18 0 5 C The calibration constant of a crystal is defined as the ratio between the energy deposited in the crystal and the electronic signal read by the readout system The energy deposited E a single crystal is a fraction of the electron energy which depends upon the impact point In order to reduce this dependence we sum the signals from a matrix of 3 x 3 crystals In this sum we normal ize each amplitude with the value of the signal given by the cosmic ray measurement In subsequent iterations the normalization values are replaced by the cal
43. 20 27 26 48 27 24 29 68 43 9097 1 5283 spected by eye and any defect was recorded Then the optical quality cf the crystals was checked by a mea surement of the transparency spectrum on their full length 26 Typical transmission curves are shown in fig 83 The three reference points at 400 480 and 630 nm are minimum transparency values required in the specifications A departure from the transmission profiles shown in fig 83 especially in the near UV range may indicate a weaker resistance to radiation This effect was observed on a few prototype crystals During the quality control of the barrel crystals several batches of a few crystals each were checked for radiation hardness They were exposed to a dose of 10 rad from a medical Co y ray source This is more than 100 times the daily dose expected for the barrel at LEP in the worst case scenario beam loss etc Immediately after irradiation the transparency in the blue region had decreased by about 40 Then it was observed to recover spontaneously a room temperature Full recovery of the original trans mission profile was reached after one month It should also be stressed that the scintillation efficiency was not affected by the irradiation This was demonstrated by comparing the light output of small and large crystals The cr stal dimensions planarities and angles were checked on a measuring bench 27 simultaneously re Transmission 300 400 500 600 700 Wa
44. 200 day per vertical crystal Whth these rates the precision on the muon momentum measurement in the muon chambers and the determina tion of the energy loss in BGO 40 calibration and monitoring of the light yield along the crystal with the accuracy required to match the detector performances can be achieved in a few days 9 2 The end caps The end caps EC are made of two symmetrical parts with 1536 BGO crystals each giving an angular coverage ranging from 12 to 42 and 138 to 168 respectively fig 97 Each end cap is split into two halves for installation around the beam tube Each half is composed of eight sectors fig 98 in which the crystals are distributed into five modules of three rows and one module of two rows near the beam tube The end caps are built with the same materials as the barrel The electronic readout the thermal regulation system and the xenon light monitoring system are also identical 83 Fig 98 Crystal arrangement in an end cap sector There are 16 sectors per end cap to those used in the barrel In front of each endcap four drift chambers measure the position and the direction of a charged particle after the TEC fiange with a spatial resolution of better than 200 um and angular precision better than 10 mrad 10 The luminosity monitor 10 1 Design description The luminosity monitor fig 99 is designed for reliable luminosity measurements in the Z energy range at LEP 41 by
45. A func tion of the slow control microVAX is to respond to these promptings and alert operators to initiate less drastic responses These two microVAX are on an unin terruptible power source Other functions of the pro grams running on the slow control VAX are to ensure that the data taking conditions are within specified tolerances and alert the operators should they change Should the operators not respond it can inform devices controlling other parts of the detector operation All these m sages are logged for later investigation For precise reconstruction of the evenis it is also necessary to set monitor and record many detector parameters at a muc finer level than that needed for slow control and safety purposes This task is distrib uted fig 118 between the PC and VME micro processors which control the hardware and the VAX computers Periodically or on request th microproces sors transfer the current settings to disk files on the VAX cluster As a debugging tool these files can be immediately dumped to consoles Periodically proces ses on the 8800 pick up this data and format it into ZEBRA banks and insert it into the event data stream rom which it is written to tape An alternate path is to send these banks ur an Ethernet link to the L3 offline computer LEPICS where they are processed for inclu sion in the master database The online copy of the database is also updated by these processes This data base is used fo
46. Bundesministerium f r For schung and Technologie Partly supported by the grant CCA 8411 129 from the US Spain joint committee Science and Technology pro gram sured by four layers of additional wire chambers providing 300 u m single track resolution and 7 mm double track resolution An electromagnetic calorimeter using a new type of crystals BGO to measure energies of photons and electrons with an accuracy of 5 at 100 MeV and better than 1 above 2 GeV A hadron calorimeter measuring hadron energies with 55 VE 5 resolution and A0 2 5 A6 3 5 for jets which also provides a clean muon sample by absorbing hadrons close to thee e inter action point thus minimizing in flight pion decays and by tracking muons through the uranium ab sorber The forward backward part of the hadron calorimeter is specially designed for quick assembly and removal to provide access to the rest of the detectors A muon detector comprised of large drift chambers able to measure the sagitta of muon tracks te provide Ap p lt 1 59 at p 50 GeV The detectors are complemented by a luminosity 28 B Adeva et al The construction of the L3 experiment CONTROL ROOMS i MAGNET POLE u MAGNET YOKE s MAGNET COL e i i 1 i i i SUPPORT TUBE z zo 2 BLOCKHOUSE j e v rt gt HADRON CALORIME TER 1 LE CENTRAL TRACK DETECTOR E LUMINOSITY MON
47. Europe Increases in network band width have been achieved through more sophisticated and expensive modems on the existing LEP3NET lines In July 1986 Codex modems running at 16 8 kb s were installed on the LEP3NET transatlantic lines The Caltech MIT and MI Y LBL links are now running at 19 2 kb s The CAMTEC packet switches at CERN and MIT were upgraded to models capable of switching packets between lines running at up to 64 kb s Most recently the presence of the higher speed X25 TELE FILE switches at MIT and CERN which are key sites both on LEP3NET and ESNET has led to the decision to move some key lines to the faster switches In Sep tember 1988 the speed of the LEP3NET transatlantic B Adeva et al The construction of the L3 experiment SLAC Caltec X 25 Switch CMU Princeton Telenet 99 NIKHEF A Lausanne J ETH Geneva Telepac Florence Naples P Rome Johns Hopkins Fig 119 The LEP3NET and ESNET X 25 line was increased to 56 kb s This became economi cally feasible with the advent of the TAT 8 Trans atlantic optical fiber cable 13 2 LEP3NET design elements The design of LEP3NET takes into account the wurldwide spread of the Colleboration the limited manpower available to construct develop and maintain the network and above all the limited budget Some of the design elements are 1 The data communications protocols are based on the internationally recognized standard X25 whic
48. ITOR Fig 1 Perspective view of the L3 experiment monitor triggering and data taking electronics a cluster of online computers and a mainframe computer for offline analysis Ethernet and packet switching net works are used for local and long distance communica tions SV s ee OM DA V SUN mU Cotes scot nem Ss pr x WB UN B fF RUN dm ee MUON FILTER MUON CHAMBERS N 2 General description of the L3 experiment 2 1 Detector The L3 experiment 1 is installed at interaction point 2 of the LEP e e storage ring A 7800 t octago nally shaped solenoid houses all the detectors The poles of the magnet are split into doors to give access to the field volume The maximum field is 0 5 T and the effective field volume is 11 4 m across the flats of the octagonal aluminum coil and 11 9 m long A water cooled screen separates the coil from the detector volume The magnet rests on a concrete cradle in tegrated into the hall foundation fig 2 The detectors are supported by a 32 m long and 4 45 m diameter steel tube which rests at both extremi ties on adjustable jacks placed on concrete pillars fig 3 The tube is concentric with the LEP beam line and symmetric with respect to the interaction point it is mechanically coupled to the elements of the low f insertion allowing alignment of all L3 detectors relative to the LEP beam The muon spectrometer forms three concentric
49. M LEBEAU P LEBRUN P LECOMTE 2 J LECOQ P LECOQ P LE COULTRE gt L LEEDOM A L GER 9 F LEHMANN L LEISTAM 9 R LEISTE 2 E LEJEUNE 9 B LEONI gt J LETTRY gt X LEYTENS C LI HT LI98 L LI PJ L1 9 XG LIS LY LIAO 7 Zy LIN F L LINDE 2 D LINNHOFER E LOFTIN 9 W LOHMANN 22 s L K S 9 E LONGO 9 YS LU 9 J M LUBBERS K L BELSMEYER C LUCI 9 D LUCKEY 59 X LUE L LUMINARI 20 G LUNADEI 9 F L RKEN H MA W G MA M MacDERMOTT N MADJAR R MAGAHIZ 29 M MAIRE P K MALHOTRA A MALININ C MANA D F MANNA 2 G MANTO 22 Y F MAO 9 M MAOLINBAY P MARCHESINI A MARCHIONNI M MARKWALDER P MARSDEN 9 J P MARTIN 9 L MARTINEZ 9 H U MARTYN F MARZANO V MARZULLO 22 F MASCIOCCHI 9 G G G MASSARO 2 L MASSONNET T MATSUDA P G MAURELLI 9 K MAZUMDAR P P McBRIDE 9 G MEDICI H MEIER Th MEINHOLZ M MERK R MERMOD 9 L MEROLA 2 M MESCHINI 12 W J METZGER 2 M MICKE U MICKE P G B MILLS gt J MNICH 0168 9002 90 303 50 Elsevier Science Publishers B V North Holland 36 B Adeva et al The construction of the L3 experiment M MOELLER A MOLINERO L MONTANET B MONTELEONI R MONTINO G MORAND 9 R MORAND S MORGANTI 2 v MORGUNOV 2 R MOUNT M MOYNOT P MUGNIER W NAGELI 2 E NAGY 9 M NAPOLITA
50. NO 22 S NEBOUX H NEWMAN 2 Ch NEYER K NGUYEN L NIESSEN A NIKITIN 2 W D NOWAK M OKLE 35 P OLMOS 2 J ONVLEE D OSBORNE J OSSMANN 9 D PANDOULAS H PAPROTNY V A PARMENTOLA 22 G PASSEGGIO 22 G PATERNOSTER S PATRICELLI 22 Y J PEI Y PENG 2 Y PENG gt D PERRET GALLIX J PERRIER 9 E PERRIN 9 G PERROT P PETITPAS a P PETSCHNER P A PEVSNER J PIER AMORY M PIERI V PIERI G PIERSCHEL P A PIROUE 25 V PLYASKIN 2 M POHL V POJIDAEV 2 C L A POLS T PONOMAREFF P J POTYKA N PRODUIT 9 P PROKOFIEV F PRUJA G P TZ P J M QIAN R RAGHAVAN P P RAZIS gt K READ 25 P REDDICK 1 K REISSMANN P D REN S REUCROFT P D REY M REYNAUD X RICADONNA J P RICHEUX 6 C RIPPICH 2 U RINSCHE R ROCCO 22 S RODRIGUEZ 29 B P ROE M ROHNER S ROHNER Th ROMBACH L ROMERO 29 J ROSE U ROSER S ROSIER LEES J A RUBIO 1420 W RUCKSTUHL 9 H RYKACZEWSKI 5 pP SAHUC 9 J SALICIO 9 S SARAN G SAUVAGE 92 A SAVIN 2 T SCHAAD 9 B SCHAFHEITLE V SCHEGELSKY A SCHETKOVSKY F SCHILD 3 R SCHILLSOTT P SCHMITT 9 D SCHMITZ P SCHMITZ M SCHNEEGANS M SCHNEIDER 3 E SCHNEEVOGT M SCH NTAG D J SCHOTANUS H SCHUIJLENBURG R SCHULTE A SCHULTZ VON DRATZIG K SCHULTZE J SCHWENKE G SCHWERING C SCIACCA
51. Nuclear Instruments and Methods in Physics Research A289 1990 35 102 35 North Holland THE CONSTRUCTION OF THE L3 EXPERIMENT B ADEVA M AGUILAR BENITEZ 29 H AKBARI 5 J ALCARAZ 29 A ALOISIO 2 J ALVAREZ TAVIEL 20 G ALVERSON 8 M G ALVIGGI 2 H ANDERHUB 33 A L ANDERSON A M ANGELOV 1130 T H ANGELOV 1130 G H ANTCHEV 30 L ANTONOV 3 D ANTREASYAN 15 A AREFIEV 2D LH ATANASOV 30 B AUROY n R AYAD OL AYRANOV 3 T AZEMOON 3 T AZIZ gt U BACHMANN 3D P BAHLER 9 J A BAKKEN 25 L BAKSAY 7 H BALDINGER R C BALL J BALLANSAT S BANERJEE 71419 J BAO 5 G BARBIER 9 L BARONE 9 G BASTI 29 A BAY 9 F BEAUVAIS 3 U BECKER R BEISSEL S BENDIG P BENE 16 J BERDUGO 1420 p BERGES M BERTHET Y BERTSCH B L BETEV 32 A BILAND 2 A BISCHOFF 32 M BISCHOPS R BIZZARRI 9 J J BLAISING M BLANC P BL MEKE B BLUMENFELD 5 G J BOBBINK 2 M BOCCIOLINI K D BOFFIN W BOHLEN 39 A BOHM D T BOHRINGER 9 H BONNEFON C BOPP 25 B BORGIA 2 K BOSSELER J F BOTTOLIER 4 M BOURQUIN 19 D BOUTIGNY P BOWDITCH ny J G BRANSON 2 D BRAUN LC BROCK F BRUYANT 14 M BUCHHOLZ D B B CKEN W BULGERONI 31 R BUREL ID BURGER C BURGOS 20 J P BURQ 9 L CAIAZZO 22 M CAILLAT B CAMBERLIN 4 D CAMPANA 22 C CAMPS V CANALE 29 M CAPELL F CARBONARA 22 F CARMINATI A M CARTACCI 2 M CERRADA 29
52. STBUS ADC modules located in the blockhouse table 7 For triggering purposes the hadron calorimeter end cap modules are longitudinally divided into two trigger planes A 1 5 A and B for the remainder up to 6 5 A Each plane is azimuthaly 9 subdivided into 16 elements while the polar 0 subdivision contains 4 elements 6 3 Detector performance 6 3 1 Proportional chamber performance The chambers are operated with an 80 argon 20476 CO gas mixture The single layer signal for a minimum ionizing particle at normal incidence is 100 fC with a charge collection time of 240 ns 9595 At an opera ing high voltage cf 1650 V the gas amplification is 1 5 x 10 In fig 69 the mean pulse height distribution for all 2284 chambers mounted into the endcaps is displayed show ing an excellent manufacturing homogeneity of 6 mean of 3 The sensitivity due to changes of the environ 1 20 110 0 80 0 109 200 300 years in uranium calorimeter Fig 70 Hadron calorimeter end cap chamber behavior under irradiation Accelerated aging test with Ru Normal chamber operation conditions see text Q is the signal from 55Fe taken in the non irradiated part of the chamber Qes is the signal at the center of irradiation The linear charge density of 1 C cm accumulated in 30 days of 1 Ru irradiatior corresponds to 800 years of operation in the L3 experiment 69 m 200 160 I counts bin 40 O i50 180 200 220 240 260 28
53. a variance of 8 The monitoring and relative calibration of the hadron calorimeter end caps will be achieved by using the radioactive background radiation of the uranium ab sorber The total ionization yield a amp to electrons and photons a s are absorbed in the chamber covers can be measured accurately with our gaseous detector al though we cannot detect a single photon line from the chain of radioactive decays of the uranium absorber nuclei In our chamber absorber arrangement this ioni zation rate amounts to 40 Hz cm yielding 10 counts s for a typical tower At this high rate we can randomly capture uranium signals with high efficiency while open ing the ADC for 500 ns with 2 kHz repetition rate This mode of operation is highly practical in our 4000 chan nels system as we avoid additional circuitry for the calibration gate formation The approximately exponen tial spectrum of the ionization yield observed in this fashion is shown in fig 72 At higher chamber gain the slope of the spectrum decreases while its end point is counts bin cn E g 20 40 60 80 100 ADC channels most probable pulse height cosmic myons Fig 71 Hadron calorimeter forward backward system Mean pulse height distribution obtained with cosmic muons for a fully assembled arrangement of a HC1 HC2 and HC3 con tainer after applying a face tc face cut to the raw data counts bin TII zs poe asi SSES NE on x
54. addition each counting room is equipped with two 25 kW air condi tioners The ventilation system of the experimental area has a normal capacity of 40000 m h stabilizing the temperature to 1 C with a dew point at 12 C In an emergency the capacity of the ventilation system can be doubled A 1780 nY surface hall equipped with a 65 t over head crane and a 16 m wide 8 m high door covers the vertical access shaft It is used as a test and assembly hall for all equipment to be lowered into the experimen tal area It also houses the magnet power supply the electricity and water distribution and the ventilation for the experimental installations To permit on site assem bly of large and heavy parts of the experiment a 1420 n assembly hall was built near the vertical access shaft It is equipped with a 65 t traveling gantry crane 2 3 Gas systems Special gas systems have been built for the different wire detectors used in L3 table 1 Recirculation is used where possible to keep gas cost low for better control of impurities and for leak detection by balancing overali input and output A very good long term stability of the mixture ratio is achieved by using infrared analyzers with regular recalibration against a reference mixture 2 4 General safety and gas safety The design construction and installation of L3 fol lowed the general CERN safety rules In addition spe cial restrictions are applicable to underground experi me
55. aim an electron beam of known momentum at each crystal individually with the beam passing through the geomeirical center of tne barrel Therefore the barrel is split into two halves along a plane normal to its axis As a consequence each half barrel has to be reinforced at the cui with a 0 5 mm austenitic steel membrane which is raostly needed during the coupling operation of the two half barrels and remains in place afterwards The cell walls as well as the inner tube are made of epoxy resin carbon fiber composite 30 This material MM Fig 87 Isometric view of the mechanical structure for the BGO barrel 1 Trumpet 2 Middle flange 3 Middle mem brane 4 Barrette 5 Slice 6 Row of 24 crystals 7 Roller bearing 8 TEC supporting rail B Adeva et al The construction of the L3 experiment 6 N 8 T ee l O 10 20cm Fig 88 Longitudinal section through the BGO barrel 1 Crystals 2 Cell walls 3 Trumpet body out of carbon fiber composite 4 Reinforcement pads for titanium bolts 5 Mid dle flange with stainless steel membrane 6 Thermal shield 7 Preamplifier board 8 TEC supporting rail offers excellent mechanical properties and is of rela tively low density The minimum value of the Young s modulus is E 45000 N mm and the yield stress is o 350 N mm The tube thickness corresponds to 0 04 radiation length at normal incidence Moreover this composite materia
56. ajecto ries is possible The barrel hadron calorimeter has a modular struc ture consisting of 9 rings of 16 modules each figs 50 and 51 The innermost ring is centered at the interac tion vertex and is flanked on either side by one ring of long modules followed by three rings of short modules B Adeva et al The construction of the L3 experiment The hadron calorimeter barrel is 4725 mm long has an outer radius of 1795 mm and an inner radius of 885 mm for the three inner rings and 979 mm for the outer rings The 261 t assembled barrel was lowered to the experi rental area in one piece by the same giant crane used for the support tube 5 2 Proportional wire chambers The design of the chambers 15 is presented in fig 52 The chamber gap is made as thin as possible without loss of mechanical stability for the anode wires at the working high voltage Each chamber is made of a plane of brass tubes of equal length with 0 3 mm thick walls and 5 mm X 10 mm inner dimensions The length of the tubes ranges from 347 mm to 605 mm depending on the position of the chamber inside the modules The struct ural strength of the chamber body is assured by 0 7 mm brass plates glued onto both sides of the chamber plane with self adhesive Mylar sheets The Mylar also in sulates the tubes which are at a high potential during operation The brass plates also shield the chamber from the uranium radioactivity reducing the counting rate due to the
57. aper mail and the directness and speed of mail exchange has also decreased the need for long distance phone calls In many cases the exchange of many mail messages within a short time takes the place of a telephone discussion with the added advantage of giving time to think and to formulate ideas clearly The electronic mail over LEP3NET has the advantages of immediacy in most cases and reliability Mail can be sent to computers which are down for maintenance or disconnected due to temporary line problems delivery of the mail will take place as soon as the connection is reestablished 13 3 3 Fiie transfer The Coloured Books file transfer system supports transfer of programs and binary files between many types of computers e g VAX IBM Gould etc The system offers data compression automatic re tries and automatic resumption from marks after any interrup tion File transfer requests are sent to a manager pro cess and control is immediately returned to the user while the transfer is handled by a background sub pro cess These features make it easy to transfer programs and small data samples of up to a few tens of mega bytes Even though the line bandwidth makes such transfers last many hours they can be confidently left to run while other work continues 1 3 3 4 Decnet DECNET data link mapping circuits have been established over LEP3NET between Caltech and CERN and between I arvard and CERN These circuits link the E
58. astic supports 8 Scintillation counters The good time resolution lt 1 ns of the scintillation counters will be used to distinguish di muon events from cosmic muons A single osmic muon which passes near the interaction point resembles a muon pair event produced in e e interaction but the time of flight difference between opposite scintillation counters is 5 8 ns for cosmic muons and zero for muon pairs 8 1 Counter dimensions The scintillation counters are located between the electromagnetic and hadronic calorimeters fig 79 In this position nearly complete coverage of the solid angle V 20 M qr 2700 V ae A 10 0 90 90 W AN Lore 400 600 900 1000 ADC Channel Fig 77 Typical charge distribution obtained with different Ar CO mixtures can be achieved and the scintillator can be used in the trigger on hadronic events The barrel scintillation counters are bent to follow the shape of the hadron calorimeter barrel they are 875 mm away from the beam at the position of the hadron calorimeter rings RO R1 and at 969 mm at the position of rings R2 R3 R4 We use a 1 cm thick Bicron BC 412 plastic scintilla tor The counter is 167 mm wide in the middle and 182 mm at the ends in order to cover the same solid angle for the inner and outer rings of the hadron calorimeter The projected length of the scintillator is 2200 mm We use adiabatic light guides made of UV transparent Plexiglas GS218 Roehm FRG
59. ation is finished the XOP arbi trates for mastership on the output segment in order to B Adeva et al The construction of the L3 experiment write data and results includirg the final decision on the acceptance of the event into the DSM on the central crate The same data can be read in spy mode by the DSM connected to the trigger computer and used to monitor the trigger 12 6 The level 3 trigger The level 3 trigger made by three 3081 E emulators is embedded in the main flow of the data acquisition The 3081 E is a computer developed at SLAC and CERN which emulates a subset of the IBM 370 instruc tion set and runs as fast as an IBM 370 168 Each 3081 E is equipped with a FASTBUS interface with two cable segment ports completely symmetric and in dependent one used for data input another for data output I O operations can be done by one port at a time while the control and status register CSR 3 0 is common to both ports and can be simultaneously accessed by them to know the status of the machine Through this interface the 3081 E becomes a FAST BUS slave geographically addressable on cable segment As a consequence the levei 3 architecture relies on the use of intercommunicating masters on both input and output segments to control all the spects of the data flow When a 3081 E is free its input responds to T pin broadcast and receives a complete event at a speed of 16 MB s It then starts a computation for the event selec
60. ations that are normally experienced when using photomultipliers A test pulse input to the preamplifier is provided The analog to digital converter ADC units one for each crystal fig 89 are mounted 3 m away just outside the hadron calorimeter The ADC has been designed to satisfy two basic requirements to measure signals accu rately over a wide dynamic range from 100 MeV to 100 GeV and to have a short memory time so that the tails from large signals do not mimic small signals in later beam crossings The signal from the preamplifier is differentiated with a pole zero circuit which replaces the 800 us exponential decay with a 1 1 us decay time and split three ways into a programmable attenuator chan nel for the trigger see section 12 and into two separate ADC channels one for small signals and one for large signals Each channel has its own independent resetta ble integrator and a sampie hold circuit The low range has an additional gain of 32 before the integrator After each beam crossing the signal is integrated and stored by the sample hold circuit Then the integrator is reset in preparation for the next beam crossing This provides the short integrator memory time for large signals The sample hold with the stored signal is not released until just before the next beam crossing allowing maximum time for the first level trigger to operate The sample hold circuits are fol lowed by two amplifiers each providing a ga
61. barrel counter is viewed by a phototube on each side only half of the counter is shown lar to the beam line 1030 mm from the interaction point Each counter covers one of the 16 sectors of the hadron calorimeter endcaps The barrel counters angular coverage is cos O lt 0 83 34 lt lt 146 where O is the polar angle with respect to the beam line They cover therefore the acceptance of the middle muon chamber MM The end cap counters extend the coverage down to cos 0 0 90 25 lt O lt 155 In the azimuthal angle 9 93 of the solid angle is covered by scintillators 8 2 Photomrultiplier The Hamamatsu R2490 01 16 stages mesh dynode photomultiplier operates in the 0 5 T field region with high quantum efficiency 14 at 430 nm the maximum in ihe emission spectrum of the BC 412 scintillator high gain 4 x 10 at 0 5 T magnetic field compared to 1 5 x 10 without magnetic field and good time resolu tion The rise time of the anode pulse is 2 7 ns with a c i w zu EM sad Fig 80 The end cap scintillators are mounted on the end cap hadron calorimeter HCl 73 transit tim2 jitter of 0 9 ns for one photoelectron and 0 6 ns for 0 photoelectrons 8 3 Readour The 92 signals are digitized in LeCroy FASTBUS TDC 1875 with a resolution of 50 ps and a dynamic range of 15 bits The time resolution of the counters can be improved by a factor of two because we also digit
62. barrel octants one barrel counter and one end cap counter one forward and one backward end cap counter two forward or two backward end cap counters in different quadrants The scintillator signals are also sent to the calorimet ric trigger to contribute to the decision of the cluster trigger 12 4 5 Trigger control The trigger control implemenis the final level 1 trig ger decision and synchronizes the data acquisition anc the level 1 trigger with the beam crossing signal Before every beam crossing the trigger control checks if all the detector components are ready for a new event If they are it sends a CLEAR signal to reset the whole DAQ system increments the event number counter and starts the next cycle then the beam crossing signal is sent out to sample the data of the whole detector and to initiate the level 1 trigger processors The results from the trigger processors are returned and evaluated before the next beam crossing If the event is rejected the system is reset by the CLEAR signal and ready for the next event If the event is accepted an ACCEPT signal is sent to the subdetectors to start the data conversion and buffering During this time the subdetectors musi set FALSE on their READY line to prevent further events from being accepted The results of the level 1 trigger processors are combined into a 16 bit trigger word Together with a 16 bit event number it is sent to the subdetectors and attached to the event
63. better than 0 5 V cm is neces sary For a safe start the 210 power supplies are ramped at a speed of 100 V s Since the high voltage resistor chains and the preamplifiers are attached to the end plates of the chamber their temperature has to be controlled by a cooling and heating system There are 96 temperature sensors distributed over the inner and outer cylinders as well as over the end plates The three magnetic field components are also measured at two positions on each end plate 11 2 4 TEC calibration Each TEC segment is equipped on its outer surface with a plastic scintillating fiber ribbon to monitor the low drift velocity 6 uu m ns to an accuracy of 0 1 fig 110 A ribbon is comprised of 143 fibers each 700 yu m wide 1000 pm thick and 1 3 m in length The fibers in each ribbon are read out using two multianode micro channel photomultiplier tubes The signals from each tube are multiplexed serially using the MX4 microplex chip and are interfaced to the standard data reduction processor of the TEC 51 MONITOR CHAMBER CO C H 4 to MIXING Fig 109 Schematic diagram of TEC gas system ANODE i DETAIL A eeeespesse BEAM TUBE INTERACTION POINT Fig 110 TEC calibration using plastic scintillating fibers PSF The drift distance time relationship is obtained for every anode by averaging over the fitted tracks for every fiber using the e e interaction point and the fib
64. by charge to time converters QTC for readout of the Z chambers and by multiplexers to read the scintillating fibers Special modules have been designed to interface the crates with the monitoring computer the L3 trigger system and the DAQ system The TEC data are written into a LeCroy 1892 FAST BUS memory where they are picked up by ihe L3 DAQ system program running on the host computer ini tializes and supervises the complete system The shaper output signal of every anode wire is digitized by one FADC with a 100 MHz clock rate Each FADC consists of a 6 bit ADC chip TRW TDC 1029 J7C and 1 kB ECL memory Hitachi HM2112 1 This allows one to digitize the incoming pulses in 10 24 us covering the maximum TEC drift time The FADC input can be switched under program control to a signal generated by the crate master for calibration Two FADC channels are connected to one DRP to reduce the amount of data transferred downstream in the read out system The two FADC channels and one DRP are built as one VME board fig 108 The DRP is built around a Texas Instruments chip TMS 99105 with a 24 MHz clock It has a 32 kByte RAM memory for program data and multievent buffer ing Started by a trigger level 2 accept the DRP reduces the FADC raw data to about 5 The output is written into a dual port memory from where it is collected by the crate master CM For a typical TEC event the computation time is less than 5 ms A check sum can be
65. can be ob tained Each wire contributes proportionally to its length to the integral F of the uranium spectrum fig 75 of a detector cell tower and the detector performance can be monitored at the wire level B Adeva et al The construction of the L3 experiment relative gain uranium random 1 2 3 A 5 6 relative gain Fe 5 Fig 73 Gain determination by observing uranium noise sig nals compared to gain determination by observing the maxi mum of the pulse height distribution from Fe 5 9 keV gamma rays in the same chamber The gain was varied by changing the high voltage 28 refintegrat 41000 e refintegral 170000 2 4 20 x 16 c LQ 12 UO n oo OC 0 01 02 0 3 0 4 0 5 06 1 integral x 10 Fig 74 Precision of the gain determination as a function of the inverse number of entries in the randomly sampled U spec trum With high statistics in the reference spectrum relative gain determinations to better than 1 can be achieved 8000 6000 uranium rate Hz 4000 i 08 6 2000 712 01 08 5 o 712 03 08 5 0 1000 2000 3000 400 5000 wire length mm Fig 75 U rate for two different HC2 towers as a function of the operative length of the sense wire change in length by connecting the appropriate chambers to high voltage selec tively The very low end of the U spectrum Qu lt Qmip has been cut out resulting in a slope of 26 Hz cm 71 6 3 3 Detector response to pions an
66. chamber layers around the beam It consists of two ferris wheels each weighting 86 t Each of the ferris wheels has eight octants and each octant has five cham bers The muon spectrometer covers 7695 of the solid angle The central section of the support tube houses the inner detectors arranged as barrel elements around 15930 mm Fig 2 Transverse view of L3 B Adeva et al The construction of the L3 experiment Free passage for RFQ insertion wary hh MAGNET YOKE Zea Bane 39 141 80 mm COIL ls Muon Filter BGO Luminosity Monitor T j T Experiment s on a slope of 1 39 Fig 3 Longitudinal view of L3 the beam pipe and as end cap elements in the for ward and backward directions The barrel elements consist of Muon Filter Hadron Calorimeter Barrel Barrel Scintillators BGO Electromagnetic Calorimeter Vertex Chamber and the beam pipe The Luminosity Monitors are situated immediately in front of the low 8 magnets 2 2 Experimental area The 21 4 m diameter 26 5 m long experimental hall is oriented longitudinally with respect to the LEP beam line which enters the hall with a slope of 1 39 The hall is equipped with a 40 t overhead crane an 11 t Table 1 L3 detector gas systems Detector Mixture Detecior volume m Time expansion chamber CO 20 i butane 0 75 Central Z chambers AR 20 CO 0 05 Muon filter AR 20 CO 3 LM forward chamb
67. cl aire de Lyon IN2P3 CNRS Universit Claude Bernard Villeurbanne France bn Center of Energy and Environmental Research CIEMAT Madrid Spain Institute of Theoretical and Experimental Physics ITEP Moscow USSR H INFN Sezione di Napoli and University of Naples Italy x California institute of Technology Pasadena USA 2 Carnegie Mellon University Pittsburgh USA Princeton University Princeton USA B Adeva et al The construction of the L3 experiment 9 INFN Sezione di Roma and University of Roma La Sapienza Italy d University of California San Diego USA ju Union College Schenectady USA a Shanghai Institute of Ceramics SIC Shanghai P R China 37 30 Central Laboratory of Automation and Instrumentation CLANP Sofia Bulgaria si Paul Scherrer Institut PSI W renlingen Switzerland on High Energy Physics Institute Zeuthen Berlin Germany di Eidgen ssische Technische Hochschule ETH Z rich Switzerland 34 National Science Council Taiwan Received 8 November 1989 The L3 experiment is one of the six large detectors designed for the new generation of electron positron accelerators It is the only detector that concentrates its efforts on limited goals of measuring electrons muons and photons By not attempting to identify hadrons L3 has been able to provide an order of magnitude better resolution for electrons muons and photons Vertices and hadron jets are also studied The constructio
68. coil 5930 mm Width of the conductor 890 mm M Fann epg Ae ae Outside radius of the yoke yuu Hii Totai length of the coil 11900 mm Electrical power at the taps 4 2 MW Central field 0 5 T Stored magnetic energy 150 MJ Rated current 30 kA Current density in the conductor 55 5 A cm Cooling water 150 m h Coil weight aluminum 1100 t Shielding weight soft iron 6700 t B Adeva et al The construction of the L3 experiment aini Fig 4 Edge cooled plate for the magnet coil The manufacturing procedure of the coil has been determined by the dimensions of the detector The sectors were welded using electron beam technology A set of special tools centered on a welding gun of 45 kW at 50 kV fig 5 was developed to support the quasi in dustrial production In a first step tour sectors were welded together forming half turns of 3 2 t figs 6 and 7 After completion of this work the equipment was moved from the CERN SPS site to the experimental site where by welding 12 half turns together the unit weight 41 Fig 6 First coil workshop assembling plates into half turns was brought to 38 t figs 8 10 Each 6 turn package has four cooling circuits this design is a compromise between the ideal case of having two circuits per turn but an enormous amount of piping or two circuits per package but water pressure problems The cooling cir cuits include more than 6000 welded joints Numerous checks concerning dim
69. ctor The length of a bundle is 35 m The light mixer consists of a polished lucite rod 200 mm long with a 5 x 5 mm section Upstream and downstream of this rod slightly depolished lucite plates improve the mixing In order to match the spectrum of the BGO scintillation light the ultraviolet and infrared components are suppressed by appropriate filters The secondary bundles typically 2 m long are composed of 200 230 pm quartz hard plastic fibers type HCP200 produced by Ensign Bick ford Optics Company At one end the fibers are glued together in a connector and at the other end each fiber is terminated by a brass ferrule which snaps into the crystal capsule The mechanical couplings of the primary and secondary bundles to the casing of the mixer are made to be strong and very precise Two different types of references are used to moni tor the intensity of the light pulses coming out of each mixer namely photomultipliers equipped with Harshaw Nal Am pulsers and Hamamatsu 2662 photodiodes Because of the temperature dependence of the Nal Tl scintillation efficiency the photomultipliers are housed in temperature controued cabinets located outside the L3 magnet about 30 m away from the BGO barrel whereas the photodiodes are mounted on the BGO barrel structure itself and need not be decoupled from the mixers during transportation between the calibra tion site and the L3 experimental hall For each photo diode the 59 5 keV y ray line
70. d events is expected from this third level trigger Only when this last scrutiny arrives at a positive decision the event is transmitted to the main data acquisition com puter to be written on tape at an expected rate of 1 Hz at the Z peak 12 3 Data flow The digitized data for each detector component may be stored in several event memories situated in the detector FASTBUS crates fig 116 Data from all parts of the detector are then collected together and sent to the central crate The merging of the data from separate substreams is performed by the combination of three FASTBUS modules a general purpose master GPM 56 a block mover BM 57 and a dual slave memory DSM 58 The GPM is a FASTBUS master which monitors the status of the source memories LRS1892 and DSM and sets up the block transfer parameters in the BM The BM then takes up master ship of the crate and cable FASTBUS segments and performs a series of block moving from several source memories to one destination memory DSM The func tion of the BM is controlled by a microcode and a B Adeva et al The construction of the L3 experiment MUON CHAMBER BOOK TRIGGER DATA 2 Cad NE HADRON CALORIM S DETECTOR OO Qe CRATE FJ LJ B 5 m oen s er C VAX 8800 Fig 116 The L3 data acquisition system variety of FASTBUS operations can be easily imple mented Furthermore each detector component has interfaced to its
71. d S Falciano CERN EP Elec tronics Note 87 04 59 F Cesaroni S Di Marco E Gennari and S Gentile Nucl Instr and Meth A260 1987 425 60 LAPP SCAIME Ecline Driver User s Manual Technical report 61 F Cesaroni E Gennari S Gentile and P Pacchiarotti Nucl Instr and Meth A260 1987 546 62 M Bourquin et al Charged particle trigger for the L3 experiment Proc Int Conf on the Impact of Digital Microelectronics and Microprocessors on Particle Physics ICTP Trieste 1988 63 A Boehm et al Nucl Instr and Meth A273 1988 4712 64 J Lecoq M Moynot and G Perrot F 682B Multi port Multi event Buffer MMB LAPP Internal Report 23 01 85 revised 10 04 87 65 T Lingjaerde Proc Topical Conf on the Application of Microprocessors to High Energy Physics Experiments 1981 CERN Geneva CERN report 81 07 66 J Lecoq M Moynot G Perrot P Baehler and C Ljus lin Proc Fastbus Software Workshop 1985 CERN Geneva CERN Report 85 15 67 G Perrot XFMI XFSI Specifications LAPP Technical Document revised 1987 68 H Newman Caltech DoE and NSF Research Proposal CALT 68 1109 1984 69 R Mount L3 Technical Note No 351 1985 7 0 High Energy Physics Computer Networking Report of the HEPNET Review Committee 1988 71 D Williams ed The Computing Needs of the LEP Experiments the MUSCLE Report CERN D and EP Divisions 1988 72 D Williams ed Green Book II Study
72. d electrons The response of thc hadron calorimeter endcaps to pions up to 25 GeV and electrons up to 10 GeV has been studied in a prototype setup which had a su lar internal structure but llowed for larger shower contain ment 18 The results are compared with Monte Carlo simulation calculations 19 giving good agreement with the data 7 The muon filter The muon filter is mounted on the inside wall of the support tube and adds 1 03 absorption length to the hadron calorimeter It consists of eight identical oc tants each made of six 1 cm thick brass 65 Cu 35 Zn absorber plates interleaved with five layers of proportional chambers and ollowed by five 1 5 cm thick absorber plates matching the circular shape of ihe supporting tube fig 76 Each octant is 4 m long 1 4 m wide and 0 2 m thick in the radial direction The first four layers of an octant each contain 16 chambers whereas the outermost layer contains 14 chambers The muon filter proportional chambers are made of a 4 m long comb l ike brass profile covered with a brass lid thus forming eight tubes and are enclosed in a plastic box to ensure gas tightness Each tube is 8 4 mm wide and 5 mm high aad is equipped with a resistive 470 Q m wire 50 um in diameter The tubes are separated by 1 6 mm thick walls They are extruded industrially wires are strung manually and supported every 20 cm by plastic holders The chambers are terminated by a board equipped with eig
73. d in 16 sectors of 22 5 each Each sector consists of 19 crystals which are 26 cm long and which range in cross section from 1 5 X 1 5 cm to 1 5 x 3 0 cm The crystals are wrapped first in teflon tape and then in 25 pm copper foil Each crystal is viewed by an Hamamatsu photodiode and a yellow light emitting diode is mounted opposite the photodiode to monitor any radiation Gamage and recovery To ensure optimum shower containment a software cut is defined that limits the acceptance to the inner six of the eight rings This also matches the full efficiency ranze of the monitor chambers The BGO array is split into two halves that are separated fig 101 during each filling of the LEP ring by an hydraulic device with a positioning accuracy of 10 pm A lead shield between BGO and beam pipe provides further radiation protector T P 1 ams 0 uh Inr Ln E E dee n Fig 101 The support structure for the BGO array main characteristics of the system are summarized in table 12 Apart from an energy trigger demanding a large amount of energy deposited in the two BGO arrays e g for tags in two photon physics a geometrical trigger will require a coincidence having a minimum energy in each of the two BGO arrays The azimuthal width of the overlap region for the coincidence is defined as two BGO sectors i e 45 This trigger scheme ailows the Table 12 Main characteristics of the L3 luminosity monitor Distance from th
74. detector provides coarser and less accurate trigger data available in digitized form within a few us The level 1 trigger analyzes the trigger data and either initiates the digitization of the main data or clears the front end electronics before the next beam crossing i e within 22 ps so that negative decisions at level 1 do not contribute to the dead time After a positive decision the detector data are digitized and stored within 500 ps in multievent buffers and the system is readied for the next beam crossing The expected level 1 trigger rate is 100 Hz corresponding to a dead time of 5 but the system can cope with a trigger rate up to 500 Hz with a 2595 dead time After a level 1 trigger the trigger data are further analyzed by four programmable processors XOP each one disposing of an average time per event of 8 ms under the most adverse condition of a 500 Hz trigger rate This level 2 trigger should reduce the rate of accepted events by a factor of 10 The level 3 trigger is performed by three 3081 E which have at their disposal the original trigger data the results of the calculations performed by the level 1 and level 2 processors and the complete sei of digitized data fium all the detector components These data are then used for a further filtering of the events The time available for each event is now about 10 times longer and ine information much more complex thus a further reduction of a factor 10 on the number of accepte
75. di i w 22901 ATR 90 TO 50 30 POLAR ANGLE Fig 115 Spatial resolution of a Z detector prototype as a function of polar angle 92 B Adeva et al The construction of the L3 experiment 200 um glass fiber epoxy Both cover layers of the middle cylinder and the outer inner layer of the inner outer cylinder consist of 50 um Kapton foil carrying the aluminum cathode strips of 4 4 mm pitch On both ends of the cylinders two glass fiber reinforced epoxy rings hold the wire feedthroughs In order to avoid electrostatic displacements the chamber is operated in drift mode with every second wire at ground potential The Z detector has an effective length of 1068 mm an outer diameter of 980 mm and a thickness of 21 5 mm representing about 1 3 of a radiation length An 80 argon and 20 CO gas mixture is used for operation 54 The high voltage is supplied on one end of the cylinder the signals from the 920 cathode strips are read out from the opposite end Preamplifier boxes carrying four 4 fold hybrid circuits are positioned on the conical end flange The signals are fed via 45 m coaxial cable line drivers and another 70 m twisted pair cable to the ADC inputs The digital readout forms part of the TEC readout system fig 107 the FADC DRP units being replaced by 16 fold 8 bit ADC A calibra tion system will pexiorm a quick check of all channels measure the gain and rms noise of individual channels and the cro
76. e which is the reference support for the coil figs 19 21 b Second phase the 28 coil subassemblies were mounted in this cradle aligned with respect to the LEP beam and electrically connected in series by welding The 5 8 of the second crown together with the two vertical walls of the barrel were then erected and the piping for the coil completed figs 22 and 23 B Adeva et al The construction of the L3 experiment Fig 21 5 8 of a crown and 3 8 of the barrel ready to receive the coil c Third phase the doors the crown top arcs and barrel roof were mounted fig 24 and the door ele ments were welded together d Fourth phase filling of the two poles was com pleted figs 25 and 26 After the magnet had been assembled and operated for 100 h at nominal current the mounting of the ST TT unit inside the magnet began The ST TT unit was pulled out of the assembly hall and lifted vertically fig 27 by a 1000 t crane helped by an 800 t foot Fig 20 3 8 of the barrel installed in the cradle Fig 22 The beginning of coil installations B Adeva et al The construction of the L3 experiment 4 Fig 23 The coil installation progresses crane it was then brought over the opening in the roof of the building covering the vertical shaft lowered through the shaft down to the experimental rea fig 28 and inserted into the magnet fig 29 The same crane was used two days later to lower the fully ass
77. e fluid temperature fig 93 To avoid accidental fluid leaks the circuit is operated at 800 mbar below atmospheric pressure The cooling fluid must have low density low viscosity low vapor pressure and high specific heat it must be an insulator to avoid mixing up the electrical CIRCULATION ON PUMP HEAT HEAT EXCHANGER _ ns j TANK Fig 93 Schematic view of a cooling circuit B Adeva et al The construction of the L3 experiment BGO CRYSTAL XENON FLASHER REFERENCE P D REFERENCE PM Fig 94 Schematic of the BGO xenon monitoring system PFB primary fiber bundle SFB secondary fiber bundle SF spec tral filter F attenuation filter optional 1 Fiber from the other lamp system 2 Fibers from other mixers grounds of the system We use about 1 t of a silicon base liquid Dow Corning DC 200 5C with a density of 1 g cm a viscosity of 5 cst at 25 C a vapor pressure of 40 mbar and a specific heat of 1400 J kg The temperature at the front and back of the BGO crystals is monitored by 1280 AD590 sensors so that there is one front and back temperature measurement for every 12th crystal These sensors are read out through the BGO level 1 readout system Another set of 960 AD590 sensors monitors the temperature of the pre amplifier boards of the level 1 readout boards and of the cooling fluids They are read out by an independent system to permit safe operation of the electronics eve
78. e for triggering pur poses These lines are grouped into 288 independent trigger signals which are digitized by Fast Encoding and Readout ADC These data are then used by the energy and cluster triggers Table 3 Thickness in cm of calorimeter and other components mea sured normal to the beam line Long Short modules modules Inner ring inner flange Stainless steel 4 0 4 0 First plate stainless steel 1 5 1 5 Base plate stainless sicci 1 5 1 5 Shield plates brass 8 4 7 4 PWC tubes brass 5 3 4 7 Uranium 27 94 24 57 CHO Mylar 2 0 1 8 BGO 24 0 Scintillator 1 0 Muon filter brass 16 5 Support tube stainless steel 10 0 B Adeva et al The construction of the L3 experiment 63 UH ELK RE ee U U X Z 5 Ero eu ms oe juli ese usur LIU VLL S de s ui 103 lU 10s 106 107 api 108 RES P e M eet ete Sots ae dem m d mn M rFTP J rra me e rl j piece x x WOUWCORE UNES pr ane ee a E sss Ese WS EN FR ee L ss s sns Fig 59 The wire grouping for the long module a projection b Z projection 5 7 Physical properties response to cosmic ray muons a typical response of a wire group to muons is given in fig 61 The response is The relevant properties of the barrel in terms of the in good agreement with the expectcd distribution thickness of various materials of which it is constructed The response of prototypes ana f the finished
79. e interaction point cm 265 Beain pipe radius cm 6 0 Radial extent of physical BGO array R min R max cm 6 8 19 0 Radial extent of acceptance area R nin R max cm 8 8 17 5 Effective polar angle coverage Omin Omax mrad 31 62 Effective Bhabha cross section o nb 100 Length of BGO crystal cm 26 0 Length of BGO crystal Xo 24 Tracking chamber resolutions entire track AR pm lt 250 AO mrad lt 0 10 A deg 0 6 Calorimetry AE E 0 5 1 0 AP a um lt 800 AO mr d lt 0 5 A deg ne a a v Has been measured in a 50 Cc lectio B Adeva et al The construction of the L3 experiment observation of radiative and nonradiative Bhabha events as well as of background interactions Software studies will then control the amount of background to be admitted into the luminosity event sample The forward tracking system in front of the BGO array consists of a stack of four planar multiwire pro portional chambers with cathode strip readout The chamber dimensions are 400 mm x 200 mm x 10 mm and each stack of four chambers has the cathode strips arranged in both R 6 two chambers and x y two chambers configurations Data on efficiency and spa tial resolution were obtained by exposing a stack of four prototype chambers to 50 GeV electrons in the X3 beam at the CERN SPS The efficiency per wire plane was better than 98 Requiring a track to have at least 3 out of 4 hits giv
80. e of 1000 m It consists of two ferris wheels each having eight independent units or octants fig 33 Air pads on the support tube allow rotation during the installation phase Rails are used to roll the assembled ferris wheel inside the magnet The octants are attached to the torque tube fig 34 Each octant consists of a special mechanical struc ture supporting five precision chambers There are two chambers MO in the outer layer two chambers MM in the middie layer and one inner MI chamber They measure track coordinates in the bending plane In addition the top and bottom covers of the MI and MO Fig 28 The support tube arriving at the bottom of the acce s shaft B Adeva et al The construction of the L3 experiment PA M chambers are also drift chambers and measure the Z coordinate along the beam There are a total of six Z chambers per octant Prime consideration was given to the accuracy of the sagitta determination Our detector has been designed to minimize the contributions from the major causes of errors in the sagitta measurements which are a intrinsic resolution of ihe drift chambers b multiple scattering c accuracy of alignment of chambers belonging to different layers An intrinsic accuracy of 250 ym per wire is sufficient to reach the design resolution Careful chamber optimi zation studies have led to smaller values 3 We average Distance from beam axis mm oq ay WAP 0 5 j u
81. e outer chamber layer by an addressable movable beam directional element fig 46 Mirrors direct the beam down through a quartz window into selected drift cells of all layers of the octant which are connected by tubes pointing roughly to the interaction point Photodiodes at the bottom of the MI chamber measure the intensity and position of the beam centroid Each octant has eight laser beam trajectories which simulate infinite momentum particles coming from the interaction point The sagitta of laser events should be zero 10 11 and thus is used to verify the alignment Two of the laser beams have movable mirrors and can produce parallel trajectories of exactly known sep aration allowing us to measure and constantly monitor the electron drift velocity 4 8 Electronics The signals from 27648 P chamber wires are con nected via 82 Q decoupling resistors to 13 824 amplifiers 12 in corresponding pairs of wires from both detector wheels Amplifiers are located in the median plane of the detector The amplifiers convert incoming currents to voltages with a conversion factor of 25 mV pA The differential output typically 200 mV for a muon is sent via 30 m of twisted pair cable to discriminators 13 set to a 20 mV threshold The logical time over threshold signal is conducted through about 14 m of twisted pair cable to 500 MHz FASTBUS time digitizers LeCroy LRS 1879 which continuously record until the common ston from the beam crossi
82. effort between L3 and LeCroy Corporation A preamplifier board con tains 24 channels as weil as power regulators snark gaps and test pulse injection circuits The differ ential output signal is transmitted to the ADC over approxi mately 40 m of 116 Q twisted pair cables grouped into shielded bundles of 24 pairs At the receiving end of the cable there is a transformer and an attenuation network which sets the energy scale in the ADC The ADC cnosen are the 96 channe LeCroy model 1882 12 bit FASTBUS ADC with a sensitivity of 50 fC per count The conversion plus readout time is slightly more than 500 ps The ADC are placed into FASTBUS crates in groups of 18 together with the Segment Manager module and Calibration module Since the beam crossings occur every 22 us the ADC hold the charge on a capacitor befwe Jigitizing until the first level trigger gives a decision 1f no first level trigger occurs the ADC are cleared before the next beam crossing In the event of a first level trigger the ADC digitize the data and store them in a multiple event buffer on the ADC board In the meantime any event left over in the buffer from a previous gate is piped out to Multiple Record Buffer memories over 70 m ECL cables Finally the data are moved to Dual Slave Mem ories for event building and for transfer to the VAX pending the second and third level trigger decisions The ADC provide immediately a fraction 1 8 of the input charge on a separate lin
83. eights 340 t rests on grease skates positioned under the center of gravity and rotates around large hinges fig 14 The hinges can be mecha nically disconnected from the doors to prevent over stressing due to the magnetic pressure on the poles 3 4 The bus bars and power supply The power supply is installed in the surface hall and connected to the magnet through the shaft with a set of 43 Fig 11 A steel skeleton for a door 82 m long water cooled bus bars made of 30 t of aluminum tubes To reduce thc fringe field and the radiated electrical noise both polarities of the bus bars are interleaved The power supply is a thyristor con verter delivering a maximum current of 31 5 kA at 150 V It consists of two transformers followed by six banks of water cooled thyristors equipped with passive filters and fre wheel diodes During the magnet tests the current was stabilized within 0 5 of the rated value Precision and reproducibility will be improved by the Fig 12 Filling material wg e T NS WM Da Fig 13 One of the magnet poles door and crown during trial assembly addition of an NMR probe The magnet coil is grounded in the middle through a resistance of 1 Q J 5 The magnet monitoring system The magnet system includes 159 cooling circuits and 29 interior electrical connections all monitored by em bedded detectors In addition potential and field moni toring devices water flow meters and
84. em bled 261 t hadron barrel calorimeter 3 9 Properties of the magnet The coil axis has been aligned to within 2 mm of the Fig 24 Some of the L3 collaborators in front of the partially assembled second pole The top of the coil is still visible Fig 25 The filled magnet doors beam axis fig 30 The measured field of the magnet fig 31 agrees with the design value 4 Muon detector 4 1 Design considerations The L3 muon detector has been designed 1 to measure high energy muons to an accuracy of Ap p 2 at 50 GeV thus providing a 1 4 dimuon mass Fig 26 The completed magnet Notice the heat shield 48 B Adeva et al The construction of the L3 experiment Fig 27 The support tube being lifted by a giant crane resolution at 100 GeV This is achieved using a config uration of three layers of drift chambers which very precisely measure the curvature of the muon trajectory in the region between the support tube and the magnet coil In this region the 0 5 T magnetic field makes a 50 GeV muon track deviate from a straight line by a sagitta s 3 4 mm To get Am m 1 V2 XA p p 1 4 we must mea sure As s to 2 i e As 70 pm Very good mass resolution is required for the missing mass Higgs search according to the reaction e e gt Z Higgs Z9 utu Fig 32 shows a computer simulation of such an event in our detector The muon detector must be modular to fill the large volum
85. ensions thermal mechanical electrical pressure and corrosion behavior have been conducted during the manufacturing period Fig 5 The 45 kW electron beam welding gun 42 B Adeva et al The construction of the L3 experiment Fig 8 Second coil workshop assembling half turns into 6 turn packages 3 3 The magnetic structure The magnetic structure is made of soft iron with 0 5 carbon content The poles are made of 1100 t of self supporting steel structure fig 11 giving the re quired rigidity and serving as a support and reference frame to mount the 5600 t of filling material which provides the mass needed for the magnetic flux return both in the poles and in the barrel The filling material supplied by the USSR is made of 50 mm and 40 mm Fig 9 Lifting the first 6 turn package thick soft iron plates cut to shape and tack welded to form individual masses of about 40 t for the barrel fig 12 and 15 t for the poles A pole consists of two parts the crown and the double doors fig 13 All the parts are made of open frames bolted together and positioned with expansion keys The crown forms a complete ring and each door a half ring The frame elements of the doors are welded together in situ Two rails on each side of the open frames are used to guide the stacks of filling material B Adeva et al The construction of the L3 experiment Fig 10 Stacks of finished 6 turn packages Each filled half door w
86. er position For every anode an average drift time is plotted for all fibers of the segment Since each fiber yields azimuthal position information with an rms of 700 y m 12 the slope of such a plot is the inverse of the drift velocity of the specific anode Monte Carlo studies have shown that for five tracks per fiber 6 h run at a luminosity of 10 cm s the drift distance resolution is of the order of 10 pm determining the drift velocity for each anode with a 0 1 accuracy x 70 E 9 50 2 EN O e i amp H a 7 30 gt 3 lt E d Q S 45 AT O z z E IO 30 20 70 3C POLAR ANGLE amp e SPAT AL RESOLUTION ym B Adeva et al The construction of the L3 experiment NUMBER OF EVENTS 40 20 O 20 40 FIBER NUMBER OBSERVED PREDICTED Fig 111 Difference between the fiber position with a bit from the predicted fiber position using drift time information The feasibility of this calibration system was demon strated in a test beam during March 1989 52 Fig 111 shows the difference between the positions of fibers with a hit and the predicted fiber positions using the drift time information of the nearest anode wire of TEC These residuals peak about one fiber position The residuals at large deviations are due to the presence of more than one beam track per trigger 11 2 5 Test resuits from prototypes Since the most important design goal an average
87. ers AR 20 CO 0 01 Hadron calorimeter end caps AR 20 CO 1 Hadron calorimeter barrel AR 20 CO 16 Muon p chambers AR 38 5 ethane 250 Muon Z chambers AR 8 5 methane 50 monorail fixed to the ceiling and a 10 t jib arm mounted on the back cavern wall There is also a 14 t gantry crane inside the support tube The 23 m diameter and 52 m deep access shaft con nected to one end of the experimental hall serves as access for experimental equipment and personnel for hall services water cables gas ventilation and pro vides in its upper half space for four counting rooms with a total area of 325 m directly above the LEP beam line and protected from radiation by a 1 7 m thick shielding of concrete beams A blockhouse at the hottom of the access shaft provides shielded space for electronics near the detectors fig 1 The detector electronics is powered via a dedicated Operating Fresh gas Recircula Number Special pressure flow tion flow of requirements bar gauge m h m h parallel Relative circuits stability 1 discontin 0 2 1 O lt 1 ppm uous A drift v lt 0 1 0 005 0 02 0 2x 2 0 005 0 24 0 16x10 0 005 0 02 0 2x 2 0 01 0 05 0 06 0 4x 7 density 0 01 0 05 0 8 0 16x 9x6 lt 0 05 0 001 25 50 16x 5 A mix lt 0 1 0 001 0 5 10 16x 3 40 B Adeva et al The construction of the L3 experiment 2 MV A transformer and cooled by water air heat ex changers mounted inside the racks In
88. es in the assembly hall continuous alignment of the ST with respect to the LEP beam fig 17 The muon chambers are supported by two torque tubes TT on rails attached to the exterior of the support tube Each TT made of nonmagnetic stainless steel has a mass of 29 5 t and supports eight 7 t octants of muon chambers A TT is a cylindrical shell with octagonal ring flanges and eight webs extending from flange to flange on the outside of the shell fig 18 The Fig 16 Installation tests of the barrel hadron calorimeter inside the support tubes Fig 17 One of the servo controlled jacks flanges hold the octant end frames in a stable configura tion while the shell resists the torsion of one flange with respect to the other The two TT have been fitted over the cylindrical part of the ST before the welding of the last flange support and are now captive on the ST fig 15 The finished ST TT unit has thus a mass of 340 t Stresses and deformation of the ST TT assembly have been studied by the finite elements method It was found that under full load the tube sags by 7 mm 3 8 The magnet assembly The assembly of the magnet progressed in four phases Fig 18 View of a torque tube before machining Fig 19 The bottom of the hall serves as a cradle for the magnet a First phase the lower 3 8 of the barrel and 5 8 of the first crown were assembled and aligned with respect to the LEP beam to form the coil cradl
89. es a tracking efficiency of 99 8 The spatial resolution was 350 pm per chamber Again requiring at least 3 out of 4 hits gives a spatial resolu tion per track of better than 250 ym The measured pulse height matching between opposed cathode planes within an individual chamber has an rms spread of 1795 this will aid in resolving ambiguities for multihit events Because of the steep dependence of the Bhabha cross section on the angle from the beam axis a crucial factor is the turn on of the efficiency near the inner radius of ihe chambers The system was found to be fully effi cient fcr radii greater than 88 mm which is adequate for the fiducial voiume used in the energy trigger 10 2 Precision goals for the L3 luminosity monitor The luminosity monitor will accept an angular re gion of 30 62 mrad with full efficiency corresponding to an effective Bhabha cross section o 100 nb At an average luminosity of 10 cm 2 s a trigger rate of about 1 Hz will result to be compared with the 0 3 Hz expected rate from Z events Thus a Statistical error of about 1 in the luminosity will be achieved in a 3 h run We aiin at a precision including systematic effects of better than 2 in the integrated luminosity on a run to run basis The main limitation will then come from systematic errors which can be separated into four major sections as discussed in the following subsec tions 10 2 1 Theoretical uncertainties These include unce
90. es the event This increments the input event pointer and the status register Four XOP processors work in round robin mode between two FASTBUS segments XOP is a fast trigger processor designed at CERN 65 and fully integrated in FASTBUS 66 It is microprogrammable and executes 192 bits wide microinstructions in 100 ns on 16 bits data words It communicates with the external world via a VMF interface or program loading and a FASTBUS interface 67 for data taking and processing When idle each XOP competes for mastership of the input crate segment The winner reads the MMB status word If positive it reads the earliest event in each memory and decrements each MMB status register thus building a complete event with its event number This readout is decoupled from level 1 operation since the MMB mem ory provides a simultaneous random read write access When the readout is completed the XOP releases mas tership and starts computation The main purpose of this computation s to reduce the event rate by a the detection of the clustered energy in the BGO and two lateral layers of hadron calorimeter by correlat ing them in a full O plane b the longitudinal and transverse energy balance of the clustered energy c the recognition of the vertex along the beam axis by using the charge division wire of the TEC chamber The programmability of the XOP provides ample flexibility in the selection criteria After the comput
91. eveloped at CERN Soon after the startup of LEP3NET the complete software base needed to run the Monte Carlo some 50 MB was transferred to MIT and regularly updated More recently with the creation of a CERN based branch of the muon chamber software group there has been a frequent interchange of new software Coordination of the L3 computing and software tusks requires an active involvement in many areas Network access is also used for examining and resolving problems with the network In addition specific pro jects at participating institutions require access to soft ware developed ai other institutes in the collaboration For example the Monte Carlo study of light collection in crystals under calibration by the RFQ system re cjuired the collection of software from L3 s European collaborators 13 5 Future network needs and LEP3NET The success of LEP3NET in offering a high quality general purpose service at a cost comparable to that of the line rentals has made it an attractive model for the future of HEP networking in the US The HEPNET Technical Coordinating Committee incorporated much of LEP3NET s experience and design principles in its recommendations for the ESNET X25 backbone In the near future the main problem of LEP3NET will be its limited speed Early analyses of the L3 network needs made in 1984 and 1985 68 69 showed that the principal liaks to CERN should reach speeds in the range of 224 kb s soon after LEP star
92. experiment HIGH VOLTAGE SIOE CHAMBER BLOCKHOUSE 70m US 25 PRE ANP BOARD MAIN BOARD ANALOG ELECTRONIC SIDE 378 WAIN BOARD PRE AMP BOARDS ne 24 BLOCKHOUSE BOm COUNTING ROOM P4 a e FADC Fig 106 Schematic diagram of TEC analog electronics output signal with minimal overshoot 50 For the pick up grid wires a different preamplifier and a special shaper type B with a differential input are used The shapers for the LR anodes the pick up wires and the CD anodes provide additional output signals to the level 1 and level 2 trigger processors Calibration of the CD channels is done by injecting known charges at the corresponding preamplifier inputs The design parameters of the digital readout system of the TEC are given by the 45 kHz bunch crossing rate and the 50 Hz events rate accepted by the trigger level 2 To achieve this the readout system has to buffer 10 events at trigger level 1 The 2 Mb of raw data per TEC event coming from 2000 readout channels have to be reduced considerably before being transferred to the L3 data acquisition system Therefore a fast data reduction processor DRP is attached to each pear of digitization units The TEC readout system fig 107 consists of 60 VME crates organized in 4 chains of 15 crates Each crate houses up to 38 FADC channels and its control unit the crate master CM In a fifth chain the FADC units are replaced
93. f the detector and can also depend on the ratio of electromagnetic to hadronic energy Cluster trigger A cluster is a localized deposit of energy observed in different detector layers at the same coordinates the energy threshold depending on tre calorimeters involved A trigger is given if at least one cluster is found Clusters zre searched for indepen dently in the two projections R O and R The information from TEC and the scintillation counters can contribute to the definition of a cluster in the R VIEW Single photon trigger A cluster in the BGO 1s accepted even if its energy is very low provided it accounts almost for the totality of the e m energy detected No track must be detected by the TEC in coincidence with the cluster in the R projection Hit counting trigger A trigger is given if the number of hits detected by the MLU in the first part of the processor is above a certain threshold In addition the luminosity monitor data are pre sented on one bus and the data corresponding to the two opposite sections of the detector are stored in two data stacks A search is made for two high energy hits with strong back to back correlation which is an indica tion of Bhabha events and gives rise to a luminosity trigger If a high energy hit is measured in only one monitor and in addition some energy is measured in the central part of the detector the event is accepted as a single tag trigger 12 4 2 Muon trigger
94. for L3 s highest priority needs References 1 L3 collaboration Letter of Intent January 1982 L3 collaboration Technical Proposal May 1983 2 D Lehm G Petrucci and G Stefanini technical note ALEPH 87 55 note 87 10 August 1987 3 U Becker et al Nucl Instr and Meth 180 1981 61 P Duinker et al Nucl Instr and Meth 201 1982 351 4 P Duinker et al Nucl Instr and Meth A273 1988 814 5 U Becker et al Nucl Instr and Meth 128 1975 593 C Willmott Nucl Instr and Meth A263 1988 10 6 W E Toth Prototype octant construction and evaluation with production phase recommendations CSDL R 1885 1987 unpublished 7 N K Chnabra and K Narender Computerized structure analysis of the muon octant frame CSDL R 2011 1987 unpublished 8 M G Dix et al NASA Technical Briefs 1981 319 9 P G Seiler et al The Laser Beacon L3 Muon Chamber Group Internal Note 85 4 1985 unpublished 10 H Anderhub M Devereux and P G Seiler Nucl Instr and Meth 176 1980 323 11 J C Guv F Hartjes and J Konijn Nucl Instr and Meth 204 1982 77 F Hartjes and J Konijn Nucl Instr and Meth 217 1983 311 F Hartjes J Konijn and Y Peng Nucl Instr and Meth A269 1988 544 12 P Rewiersma NIKHEF H Internal note L3 wire ampli fier 1986 unpublished hybrid made by Philips 13 P Rewiersma NIKHEF H Internal note DISC DOC 27 03 87PR unpublished 14 S Bu
95. gger 12 4 1 Calorimetric trigger The calorimetric trigger processes the information given by the electromagnetic and hadronic calorimeters and by the luminosity monitor 256 384 and 32 chan nels respectively it reaches a final decision 16 8 ps after the beam crossing and is built with 350 CAMAC modules forming a digitizer part and a processor part The CAMAC bus is only used for initial loading and for testing purposes The trigger data flow on front panel differential ECL buses 16 bits wide Data from differ ent buses can merge together making synchronous oper ation of all the modules necessary This is achieved with a centralized and programmable timing source which distributes the strobes to the whole trigger system CAMAC 16 channels fast encoding and readout ADC FERA modules LRS4300 are used for a fast digitiza tion of the information whereas the processor part is built around memory lookup units MLU LRS2372 and arithmetic logic units ALU LRS2378 For the electromagnetic calorimeter the outputs of 30 BGO crystals are grouped together to obtain a segmentation of 32 in 4 and of 8 in These 256 signals are digitized in 8 5 us with an 11 bit range and a sensitivity of 25 MeV bit The luminosity monitors give 16 signals on either side corresponding to 16 segments in Two signals are extracted from each hadron calorimeter module The first layer A corresponds to the chambers in the first interaction length The remain
96. gion with a small angular offset 10 mrad in 4 to suppress photon leakage Their cross section is approximately square and they have a constant front face area This geometry involves 24 types of crystals with shape changing slowly along O due to the variable distance to the interaction point fig 81 Aa accurate 50 pm fast cheap and safe method for cutting and polishing the crystals was developed 24 consisting in sawing directly the crystals to the required dimensions with a diamond disk The surface finish obtained allows us to proceed directly to mechanical polishing of nine crystals simultaneously on a spinning table Systematic tests 25 were performed on all crystals arriving at CERN in monthly batches of 130 to 400 crystals during three years Each crystal was first in lt AS gt re BSP o z E Mie Ll ee TYPE 2 pe gt R TYPE z t FA gt M AP View J View J in Fig 82 Typical crystal shapes for the barrel See alsc table 10 B Adeva et al The construction of the L3 experiment Table 10 75 Dimensions of a few representative crystals in the barrel NCR is the crystal row number deg its polar angle witb respect to the beam axis and d its angular aperture The meaning of the other quantities in mm is shown in fig 82 NCR A AS B AP ASP BP 8 de 2 20 12 20 17 20 54 29 52 29 59 30 14 87 7040 2 2898 12 20 12 20 46 20 39 28 57 29 05 28 95 65 6825 2 0434 24 20 12 20 70
97. h sends the data from source to destination in packets including information for error checking and error re covery procedures to ensure that the data is received error free The X25 protocol was chosen for LEP3NET for a number of compelling reasons X25 is the onl way to reach ma ay of L3 s European collaborators and it allows transparent connections tc established across combined leased line and publ networks The X25 protocols are supported and off the shelf interfaces are provided by all major computer manufacturers 2 In order to support the full range of services needed by the physicists several sets of higher level protocols run on top of X25 This includes DECNET and the Coloured Books protocols The Coloured Books protocol has been implemented on many machines to provide a high level user interface on top of the X25 protocol and has particularly valuable features for re tries and resumption after interruptions 3 The data communications equipment ts all stan dard commercially available This includes the packet switches and computer interfaces The CAMTEC packet switches were chosen on the basis cf performance cost programming flexibility and proven long term reliabil ity 13 3 LEP3NET network services The facilities offered by LEP3NET are detailed be low 13 3 1 Terminal access to remote computers Any LEP3NET computer can make terminal calls directly to any cther LEP3NET hos The X25 protocol avoids the
98. h module is fed into the housing maintaining a small over pres sure in the chambers A long module weighs 1860 kg a short module 1720 kg 5 4 Barrel mechanics and services The 16 modules are mounted on a 17 mm thick stainless steel ring and are bolted together at their outer 61 radii for added rigidity fig 57 Each assembled ring can move on two rails inside the support tube on two sets of rollers attached to the modules just below its mid plane The nine rings of the barrel are bolted together through the support plates of the rollers In the assembled barrel fig 58 the 144 base plates form the outer surface of the barrel The connections for the high voltage preamplifiers temperature sensors and gas pip ing are all situated on this surface Preamplifier boards are plugged directly into the connectors in the base plate The preamplifiers dissipate about 10 W per mod ule and this heat is removed by liquid cooled fins mounted on the base plate 5 5 Tower structure In order to measure the energy of hadrons and hadron jets to separate two jets from each other and to determine the energy loss and the trajectory of muons passing through the calorimeter a high degree of read out segmentation in the calorimeter is required This need is accentuated by the fact that in some cases it is necessary to confine the measurement to the immediate region in which the energy is deposited in order to reduce the background from the natu
99. he level 2 computer Each ADC cea also measure the leakage current of the photodiodes a d if desired the tempera ture of the BGO crystal Several test modes are availa ble to ensure the integrity of the system All the data words have a parity bit added before they leave the level 1 microcomputer Fig 90 shows the organization of the upper levels The level 2 computers are organized in groups of 16 aurem 7 7775000 ame art S p a pore E S ga 5 et TT t a N De _ s 1 i LN ee levee fi ave iiem a f e SN t2 oe lt PR s O P E peo c7 x are Q ty 79 and controlled by a level 3 computer also Motorola 68010 which is the VME crate master the level 2 computers are VME slaves Communication between level 2 and level 3 is via dual ported memory located in each level 2 computer The programs in these computers are stored in RAM and are loaded during an initializa tion phase either by downloading from a higher level or from local non volatile memory in each VME crate Each VME crate provides readout and control for 960 crystals Eight crates are required for the barrel thirteen crates for the full calorimeter with endcaps and includ ing the luminosity monitor These crates are connected to a level 4 crate via a VME to VME link which incorporates FIFO memory buffers The level 4 com puter combines the data for each event into one block and sends it to a FASTBUS memory modu
100. ht 700 pF decoupling capaci tors and eight 10 MQ resistors through which the High Voltage is applied to the wires Chambers are tested f gas leakage less than 100 ml h at 10 mbar overpres sure wire strength 190 310 g and dark current less than 100 nA at 3600 V The chambers use the same gas mixture as the hadron calorimeter barrel at about 1800 V to obtain 400 fC for minimum ionizing particles The 8064 channels readout system is the same as that of the barrel hadron calorim eter For three central layers of each octant we also digitize the charge pulse from the other end of the chamber so that the coordinate along the wire can be obtained by charge partition By coribining the mea surements on the thre layers the precision on this coordinate ranges from 0 8 3 1 cm to 1 2 4 7 cm for a gate going from 250 ns to 400 ns 20 The muon filter chambers were tested both in laboratory with cosmic rays and with the X3 beam ai the CERN SPS 21 22 In tne laboratory tests several Ar CO mixtures were used Figs 77 and 78 show some results that demonstrate the very satisfactory chamber behavior The overall chamber efficiency measured dur ing the tests on the SPS test beam was 91 giving 97 72 B Adeva et ai The construction of the L3 experiment I i 1390 m i m F h Fig 76 Section of one muon filter octant when unfolded for geometrical effects wall thickness pl
101. ibrated on an optical bench and the systems are linear over a range of 250 pm The middle bridge is moved by actuators B Adeva et al The construction of the L3 experiment in x and y perpendicular to the beam direction in response to computer readout of the system The bridges are aligned when all four quadrants of the photodiode receive equal amounts of light These systems allow us to position the bridges and thereby the wires to an accuracy of 10 um Deviations from zero occurring as functions of time or temperature are continuously re corded Wires are put into position along the Pyrex surfaces of the three bridges using a template They are then attached to the Pyrex surfaces with an insulating wax Since the Pyrex edges are optically flat the average of the 16 or 24 signal wires has little systematic error and one can determine the local track slope to 1 2 mrad Fig 40a depicts a chamber cell with 16 signal wires showing the computed drift paths in a 0 5 T magnetic field The chamber cell has been designed to have a very uniform electric field throughout the active region Sense wires are spaced 9 mm apart and are interspersed with field wires Eight additional wires beyond the last sense wire equalize the drift time behavior of all the sense wires within 0 2 ns 10 pm A plane of cathode wires spaced 2 25 mm apart is at 50 75 mm from the sense wire plane Four different high voltages are applied to the sense field cath
102. ibration constants obtained in the previous step Two quantities computed from the pulse height dis tributions can be used to evaluate the calibration con stants the peak and the mean values We use the mean value to compute the constants and the peak to control systematic errors possibly due to data selection or to the influence of malfunctioning or absent channels Both methods agree within 0 2 fig 95 Each crystal con tributing to the tail of the distribution is reevaluated A third method based on single crystal signals was also used as an additional check It agrees weli with the other methods lt 0 4 The main correction was due to the actual tempera ture profile in the BGO crystals The thermal sensors allow the determination of a temperature map which is used to correct the measured pulse height applying the average temperature coefficient of 1 55 C The reproducibility anc the stability of calibration constants 200 C 100 Number of events gi O O 0 98 1 102 Ratio Fig 95 Ratio of the mean value to the peak value for all crystals in the first half barrel were tested with a prototype matrix 37 within 0 3 rms systematic error Since one of the most important parameters of the BGO detector is its low energy resolution this was tested using one of the half barrels at a specially desig ned beam line providing 180 MeV electrons at the LEP injector linac A representative sample of appr
103. igher level computers is done as a background task The microcomputers can buffer up to 41 events internally This system allows a peak instantaneous event rate of 4000 triggers per second and an average rate of 500 triggers per second the transfer rate to the master The microcomputers also allow several system features to be added at the lowest level Each crystal has an individual sparse scanning threshold trigger attenuator constant for the analog output to the trigger system trim HANDSHAKE TOKENS RESET INTERRUPT MICRCCOMPUTER PREAMPLIFIER PHOTODIODE POLE ZERO x pp MULTIPLEXER AP AMPLIFIERS INTEGRATORS ure rc aes oy y SAMPLE HOLDS SLOW AMPLIFIERS COMPARATORS TEMPERATURE INPUT Fig 89 Diagram of the level 1 ADC system B Adeva et al The construction of the L3 experiment TO L3 MAIN DATA ACQUISITION SYSTEM LEVEL 4 COMBINES ALL DATA INTO ONE e STREAM TO OTHER LEVEL 3 VME LINK ae C BUS EMEN JLI HEY IILI IL I ICE TE 16 LEVEL 2 CPUs EACH CONTROLS 60 ADCS L IL LOL IL OIL JL JU TL IL IL JL EACH LEVEL 3 CPU CONTROLS 16 LEVEL2 CPUs TO OTHER TOKEN RiNGS _ H TH 60 LEVEL MICROCOMPUTERS 1 ADC FOR EACH IN A TOKEN RING NETWORK BGO CRYSTAL H H IL A H JL H HR H Fig 90 Organization of the upper levels of the readout system constant to adjust the pedestal value test pulse enable etc all of which are downloaded from t
104. ility each crystal is separated from its neighbors by a composite wall made of two layers of 100 um pre impregnated carbon cloth The final thickness varies between 200 um and 250 um A nominal clearance of 100 pm is kept between the crystal faces and the walls in the and O directions It allows for the paint coating and the expected elastic deformation of the structure due to thc changes in position of the detector from assembly to calibration and to experiment without stressing the crystals Cellu lar walls and clearances represent about 1 7596 of the solid angle covered by the barrel The supporting structure is made of two symmetrical shells connected by titanium bolts in the middle plane Each shell is a composite trumpet 1 m long object with a 10 mm thick cylindrical part in front of the crystals and a 5 mm thick conical part along the side of the last barrel crystals at 47 A 25 mm thick solid ring with fixtures for the bearings reinforces the cone edge The cylinder inside diameter is 1015 mra the solid ring outside diameter is 1600 mm Individual crystal weights and pulling resultants are gathered on the trumpet cylindrical shell and transmitted to the bearings through the conical shell and the solid ring The maximum trumpet deformation calculated 32 and measured dur ing a preassembly loading is less than 1 mm including bending and section deformation This deformation is uniformly shared on to the cellular structure The
105. imension All crystal faces are polished The parameter R is the relative light output difference for the source at 21 and 3 cm from the photomultiplier The dots correspond to aluminized mylar wrapping and the circles to white paint coating see text 76 B Adeva et al The construction of the L3 experiment Light yield 10 e cm Distance to photodiodes cm Fig 85 Typical light collection curves obtained with the cosmic ray bench for a sample of 96 crystals The light yields are measured by the number of photoelectrons per cm of ionized track To guarantee the performance of the crystals for energy measurement it was necessary to check their light output and the uniformity of light collection The light output is an important parameter for the resolu tion at low energy 100 MeV to about 2 GeV which is dominated by the signal to noise ratio We used the ionization by cosmic muons for these measurements 28 A test bench capable of simultaneous measurement of the response of 40 crystals per day and reconstruc tion of track length and position along the crystals was set up The painted crystals were equipped with caps ules holding two photodiodes Hamamatsu S 2662 1 5 cni active area followed by preamplifier shaping amplifier and ADC The data obtained fig 85 can te considered as a precalibration of the crystals and were very useful for understanding the final beam calibra tions of the two half barrels Fig 86
106. in of four to yield the desired wide dynamic range A single chip microcomputer with its program in masked ROM Hitachi 6305 chooses one of the six signals available TO TRIGGER 8 BIT DATA BUS BIDIRECTIONAL SANAA to digitize using a 12 bit digital to analog converter DAC and six comparators A simple successive ap proximation algorithm is used on the chosen signal The least count on the most sensitive range the low energy channel and a gain of 16 after the sample hold is 5 uV corresponding to less than 100 eV Full scale on the least sensitive range no gain after the preamplifier is 10 V about 200 GeV The digitizing range of the ADC is equivalent to a 21 bit ADC with resolution of at least 10 bits 1 1000 for signals greater than 100 MeV The linearity is better than 1 over the full range The actual dynamic range achieved for BGO signals is 200 000 1 from full scale to the noise level The microcomputers level 1 readout complete the digitizing and store the data within 250 ps after the trigger The microcomputers one for each BGO crystal are organized in token ring networks of 60 crystals and are controlled by another computer level 2 readout which is a single board Motorola 68010 in a VME crate These are located more than 100 m away in the count ing room Communication is via differential TTL drivers and receivers Only the actual analog to digital conver sion is done at the trigger time readout by the h
107. ion of the L3 experiment in the group The gas 80 Ar 20 CO mixture is supplied serially to groups of five chambers in the same projection Groups are connected in parallel to a com mon inlet and outlet located on the module base The services are fitted into the space between the absorber chamber stack and the hood In four modules of each ring top bottom left and right the internal temperature is monitored at three positions with PT1000 sensors embedded into one of the spacer bars The measurement accuracy is better than 0 5 C The information is used in the offline analysis to correct for the change in the gas gain due to temperature variations The hood is made of stainless steel plates welded along all edges The slanting walls of the hood point to the beam axis and produce a gap in the acceptance They therefore were made as thin as practical 4 mm The parallel walls are 15 mm thick while the top plate has a thickness of 22 mm The hood is bolted gas tight to the module base and locked to the 15 mm top plate so that in any orientation of tke module the space between the stack and the hood served for the services is preserved The smallest chamber of the stack is situated between the hood and the 15 mm plate Gas fittings are provided in the base to permit the gas flow into and out of the space between the stack and the hood independent of chamber gas supply and return In practice the exhaust from the chambers in eac
108. is used as a TDC stop Runs of about 10000 events are taken at B Adeva et al The construction of the L3 experiment a LASER RUN 5059 EVENT 6 150 9 b i 1 100 m p 26 10 um 50 E um 0 r t oF e y oe 8 50 i E 100r x 150 NM ERRARE MEN roe ERN CENE luci 950 975 1000 1025 1050 Run number Fig 47 a A computer reconstructed laser event b Measured sagitta from a series of 42 runs of 100 laser shots each both ends of the octant Tracks are reconstructed from the chamber segments as shown in fig 48a Accepting all cosmic rays we obtain the wide histo gram in fig 48b which has an rms of 1 3 mm This large width is due to multiple scattering from the predomi nantly low energy cosmic muons The L3 muon chambers are unique in that they also measure the local slope of the particle trajectory to an accuracy of 1 mrad Demanding that the local slope in a chamber agrees with the overall particle trajectory to within 2 mrad eliminates events with large multiple scatters A narrow distribution of 760 events is obtained in fig 48b The centroid confirms that this octant is aligned to s 2 400 760 um 2 14 5 um Fig 49 summarizes the results for 16 octants s 0 is given by the setting of the two independent opto mech anical systems The squares show verification of the geometry by UV laser measurements The cosmic ray data are shown in circles as a third independent mea
109. ize the charge of the pulse so that time slewing corrections can be applied A paddle card has been built to convert the photomultiplier pulse to the differential signals needed for the LeCroy 1885 ADC 8 4 Test results Acceptance tests have been performed with cosmic rays on all counters with the following results The mean attenuation length is 1 7 0 2 m A cosmic ray which penetrates the middle of the counter yields on average 41 photoelectrons A time resolution of 0 35 ns has been measured for the mean time of both photo tubes We have also tested a counter with phototubes in a magnetic field of 0 5 T The number of photoelectrons is not affected by the magnetic field but the gain is reduced by a factor of 2 2 compared to zero field Similar results have been obtained for the end cap counters Here we have measured 80 photoelectrons for a minimum ionizing particle penetrating the scintillator 9 The electromagnetic detector The electromagnetic detector has excellent energy and spatial resolution for photons and electrons over a wide energy range from 100 MeV to 100 GeV It uses bismuth germanate BGO as both the showering and detecting medinm BGO is a particularly attrac ve material for an electromagnetic calorimeter because it has hien stopping power short radiation length for photons and electrons and large nuclear interaction length Furthermore it has low afterglow and is not hygroscopic The electromagnetic ca
110. l distances between reference surfaces are measured to 5 um by means of special tools calibrated with a laser interferometer Angles are measured with electronic bubble levels 5 which have an intrinsic resolution of better than 1 prad A completed octant is shown in fig 43 After chamber loading preliminary alignment and functionality tests each octant is subject to a detailed alignment verification see section 4 10 Then it is rotated to the angle corresponding to its final position in the ferris wheel and alignment is performed to the specifications of fig 42b The elastic nature of the structure is verified by the absence of hysteresis in octant rotation This is important in view of the installa tion scheme on the torque tube For these reasons each octant is rotated through 360 then its alignment is rechecked All octants have been processed in this way and have shown full reproducibility hence elastic be havior TENSION ACOS _ Oct b P 597 parallel in X Y plane 8 2005 mrad Psa t2 perpendicular s 0 5mm 2005 mra Fig 42 a The octant stand structure b Some of the alignment tolerances are shown for assembled octants The alignment tolerance on the chamber center lines is only 25 pm Fig 43 End view of an octant 4 5 Opto mechanical alignment vertical alignment sys tem The accuracy of the bridges inside the P chambers and of the internal alignment systems built into them a
111. l has good molding properties and is very suitable for building a complex structure with the required precision Metallic parts are intro duced in some places for very specific purposes Titanium alloy screws to connect the two half barrel structures low density high resistance alloy ASTM 3 7164 High resistance aluminum alioy Anticorodal 100 pads to connect the supporting structure to the four roller bearings behind the crystals BGO crystals are extremely sensitive to permanent loading with risks of subcritical crack propagation 31 Each crystal is thus held in a separate cell with clearances such that normal structural deformation does not affect any crystal and that the weight of any crystal is not transferred to its aeighbors A total of 160 modular molded slices with 24 cells each are glued side by side on the supporting structure in order to produce a com plete half barrel A step shaped part called a barrette closes the slice bottom and provides accurate longitudi nal crystal position Each crystal is pressed from its back onto the front of its cell by a spring loaded device which pulls on the walls with a force of about 20 N crystal weight 1 kg ensuring that a positive force is exerted on the bottom of the cell in any position and that the crystal cannot slide sideways nor press on the cell wall TI The thickness of the cellular structure has been mini mized to the limit of technical feasib
112. le in the main data acquisition system The system has a maxi mum average data transfer rate greater than 6 Mbyte s This corresponds to 500 triggers per second with 20 of the crystals having data above the sparse scan threshold 9 1 4 Thermal regulation The light produced in a BGO crystal by a particle of a given energy is strongly correlated to the crystal temperature The light output variation is 1 55 C Therefore we must maintain the BGO at the lowest possible temperature but above the dew point The required energy resolution lt 1 at 50 GeV implies maintaining the crystal temperature constant within a few tenths of a degree as well as the temperature difference between the two crystal end faces below 0 5 C c e gt ED rs a mam MT a a ce m o TA ge Te S s co c pb M we iudi 14 m i gt ku Fig 91 Exploded view of a preamplifier board for the BGO barrei 1 Thermai stabilization cover 2 Cooling pipe 3 Cables to the level 1 curd 4 Preamplifiers S Thermal sersors 80 B Adeva et al The construction of the L3 experiment Fig 92 Schematic view of a level 1 readout box 1 Thermal screen 2 Cooling pipe 3 Level 1 readout board 4 Connectors for the readout cables from the preamplifier boards 5 Support frame for the readout box and for the cables The preamplifiers located at the rear face of the crystals dissipate about 0 2 W each This heat must be removed
113. lection was undertaken 42 We have foreseen the use of an asymmetric softwar trigger 44 one of the detector arms requires the nomina geometrical trigger between 30 and 62 mrad whereas the other arm has to satisfy the geometrical trigger within a loosened angular range from 30 d to 62 d milliradians where 86 B Adeva al The construction of the L3 experiment x mm 1 b 2 c 3 Calibration change 4 5 6 la a d 0 Ox Oy 0 as for c b d 0 6 mrad Oyx G 0 0 c d 0 6 mrad o 03 mm d 0 012 mm o 33 mm Fig 102 Effect of asymmetric trigger d is a parameter to be suitably chosen As an example the event rates drop almost linearly with x when d 0 With increasing d the dependence becomes parabolic until for a particular value of d there is no dependence for a wide range of x values With n larger d values one obtains even a rise in rate with x for small values of x The variation of the event rate for particular values of d as a function of x is shown in fig 102 In our sir ulation a chamber resolution of 400 um was used The energy showers were not simulated but lateral non fiuctuating shower spreading was allowed The trigger was simulated with an energy trigger threshold of 78 of beam energy per arm The events dhemselves were produced by a standard generator 45 including first order radiative events it is important to realize that the resuits fr
114. lorimeter 23 consists of about 11000 BGO crystals pointing to the iriteraction region Each crystal is 24 cm long and is a truncated pyramid about 2 X 2 cm at the inner end and 3 X 3 cm at the outer end Two silicon photodiodes and asscci ated linear electronics detect the light The energy reso lution is 596 t 100 MeV and below 1 for encrgies above 2 GeV the measured spatial resolution above 2 GeV is better than 2 mm and the hadron electron rejection ratio about 1000 1 The detector fig 81 surrounds the vertex chamber and consists of 74 B Adeva et al The construction of the L3 experiment Fig 81 Longitudinal cut through the BGO detector i two half barrels EB made of BGO crystals ii two end caps EC made of BGO crystals with track chambers FTC in front to be installed in Phase ID 9 1 The barrel The 7680 crystals of the barrel table 8 are arranged in two symmetrical half barrels giving a polar angle coverage 42 lt lt 138 9 1 1 BGO crystals The BGO crystals table 9 are produced by the Shanghai Institute of Ceramics in China using a mod ified Bridgeman method Very pure Bi O4 and GeO powders impurities 1079 are mixed in the correct stoichiometric proportions The resulting polycrystalline powder of Bi Ge 0 contained in a platinum foil crucible is introduced into an oven Then a temperature gradient is slowly displaced relative to the melt starting from a BGO monocry
115. m and accumulate addition and subtraction are performed by fast arithmetic and logic units ALU LRS2378 59 more complex operations are performed by MLU intermediate data are stored in data stacks DS LRS2375 The patterns of timing signals necessary to drive the system are stored in fast memories ECLine Drivers or ED 60 Extensive use is also made of the bus switch BSW 61 module which latches two 16 bit ECL words and multiplexes data from two buses onto a single bus In order to speed up the data flow data from the hadron calorimeter are presented in parallel on seven different buses Since the BGO data need a longer time to be digitized they are ready when the flow of the HC data is terminated and they are sent on the last four of the same buses For each calorimeter BGO HC layer A HC layer B the data corresponding to a fixed and different values are summed together sums and similarly the data corresponding to a fixed and different values of O O sums The 640 trigger data are therefore reduced to 96 elements which are used to perform the following trigger calculations Total energy trigger The sums are added over to give the total electromagnetic total hadronic and total energies A partial sum is also performed limited to the central part of the detector A trigger is given if the total energy is above a predefined threshold which can be different if the energy is localized in the central part o
116. mod are given in table 3 In tables 4a 4c the thickness of the ules to beams of hadrons and electrons was studied at barrel components and the integral from the beam axis the CERN and ITEP accelerators between 1 and 50 perpendicular to the axis are given in interaction GeV A typical pulse height distribution for 20 GeV lengths and minimum ionization energy loss Values in pions is given in fig 62 The response as a function of table 4 are calculated from the data published by the energy is linear The resolution defined as the standard Particle Data Group except for pion interaction data deviation of a Gaussian fit to the pulse height distribu which are taken from ref 16 Table 5 shows how the tion versus energy is shown in fig 63 where data from calorimeter thickness varies with the polar angle 0 prototypes and completed modules ottained both with The properties of the hadron calorimeter utilizing and without BGO in front are used Extensive studies Ar CO gas proportioral chambers with uranium ab have also been made with other gas mixtures including sorber plates have been studied extensively both as isobutane in various mixtures 17 prototypes and for compieted detector modules One phase of the study was the characterization of the 6 The hadron calorimeter forward backward system Twisted Pair Transformer f Amplifier Cable AE 6 1 Introduction Tae endcaps of the hadron calorimeter HCEC cover the polar angle regions 5 5 lt
117. mory each enable the monitoring program to investigate TEC data without interfering with the main data stream 11 2 3 Infrastructure To achieve the spatial resolution of 50 pm the TEC is operated with a gas mixture of 80 CO and 20 iC H o a gas with a low drift velocity at 2 bar The drift velocity and hence the spatial resolution depends strongly on the pressure and the stability of the gas mixture thus the iC4H content must be stable within 0 1 Fig 109 summarizes the design of the TEC gas system which operates in closed loop to guarantec long term stability The mixture is stored in a 1 m barrel and circulated through a filter and an Oxysorb purifier to keep the oxygen content below 1 ppm The purified gas passes through TEC at a flow of 200 bar h The 89 gas quality is continuously monitored by a mass spec trometer and the water content is measured by an hyzrometer A small drift chamber monitors changes in drift velocity efficiency and mean pulse height due to small changes in the gas composition A VME computer supervises the gas system and passes status information to the monitoring computer and to the L3 slow control system The high voltase ecvctam cumnhas tha panic mae chains YA AG I y OVVARS OMS LIU resistor Cnains for the field shaping wires as well as the anode and potential wires Since the TEC is operated with a low drift velocity linearly dependent on the value of the electric field stability
118. n in case of failure of the central BGO computer The reading accuracy is about 0 1 C for each sensor 9 1 5 Xenon light monitor The light collection efficiency of each crystal gether with the gain of the corresponding readout chzin are monitored by means of xenon light pulses distrib uted by optical fibers fig 94 The light flashes are generated by a set of 16 x ion flash lamps From each lamp the light is transported by four bundles of optical fibers primary bnadles to four light mixers and then to the crystals by four secondary bundles of 240 fibers each From each mixer additional fibers carry light to reference photomultipliers and photodiodes Each crystal is illuminated by two fibers from two independent systems one for high energy pulses typi cally 35 GeV equivalent the other for low energy about 1 5 GeV equivalent These two pulses are used to compare the behavior of the low and high gain channels of the BGO readout electronics Furihermore the sectors of the barrel covered by neighboring low and high energy fiber bundles are slightly offset This en ables the investigation of systematic effects The driving circuits of the Hamamatsu L2453 xenon lamps have been tuned to produce the same pulse length as the BGO scintillation The primary bundles 81 are made of 14 quartz fibers type PCS600 core diame ter 600 nm produced by Fibres Optiques Industrie which at each end are glued together in a conne
119. n of L3 has involved much state of the art technology in new principles of vertex detection and in new crystals for large scale electromagnetic shower detection and ultraprecise muon detection This paper presents a summary of the construction of L3 1 Introduction The L3 experiment is designed to study e e colli sions in the 100 GeV range with emphasis on high resolution energy measurements of electrons photons and muons It is an effort involving a worldwide col laboration of 460 physicists belonging to 34 institutions from 13 countries The preparation of the experiment took eight years from its conception to the beginning of data taking in summer 1989 The total cost was 200 MSf and 1100 technical man years The detectors are installed within a 7800 t magnet providing a 0 5 T field We choose a relatively low field in a large volume to optimize muon momentum resolution which improves linearly with the field but quadratically with the track length From the interaction point outwards the follow ing detectors are installed fig 1 A central detector tracking charged particles with a 50 um average single wire accuracy in the bending plane and with 450 um double track resolution In the nonbending plane the Z coordinates are mea 3 Metareg b Formerly C S Draper Laboratory 9 Also LAA 9 Also Bologna Also F Bitter National Magnet Laboratory Also KEK 8 Also Nanjing t Deceased Supported by the German
120. ng arrives the TDC cover a range of 1100 ns with 2 2 ns least bit accuracy The system has proved stable to 0 2 ns correspond ing to 10 am when checked by our standard 7 calibra on system 4 which electronically induces pulses onto the wires All wires of the Z chambers are similarly processed by 7680 time recording channels Parallel outputs without time processing are used to form fast road triggers S6 B Adeva et al The construction of the L3 experiment Rotating LASER mirror Fig 45 a The laser beacon references six points in the center plan e b Two laser beacon sensors with precision templates for light transmission 4 9 Control and monitoring In addition to readout of the alignment systems the laser beacon system and the UV laser system there are other critical parameters which must be monitored and controlled the chamber high voltage system the signal pulse height the time zero T calibration the octant temperature map discriminator thresholds preamplifier power supplies POSITION e SENSITIVE PHOTODIODES Fig 46 UV laser alignment schematics The beam from the nitrogen laser is directed by a beam directional element into eight trajectories Position sensitive photodiodes which mea sure the location of the beam at the bottom of the inner chamber are schematically indicated actuator motor position monitors magnetic field measuring probes
121. ning long term chamber align ment to lt 30 um The structures have been designed 6 to avoid tensor force transmission thus octant behavior is fully predictable under all conditions of stress load and temperature 7 The main elements of the octant support structure are the A frames fig 42 They support each P cham ber at four points two in each A frame The MI chamber is mounted directly onto the A frame while the other chambers are mounted on special support bars A combination of support points with zero one or two degrees of freedom ensures that chambers can move following temperature variations without introducing unwanted mechanical stresses A longeron connecting the two A frames provides mechanical stiffness in the Z direction In addition to the large structural components there are approximately 300 small precision parts per octant Materials were selected for strengih thermal character istics and long term stability Special materials such as titanium and copper beryllium have been used for chamber support feet chamber tie plates torque tube joints and other highly stressed areas During the assembly phase each octant undergoes detailed dimension and referencing checks some of the design tolerances are listed in fig 42b The four support 54 B Adeva et al The construction of the L3 experiment feet on the floor which simulate torque tube attachment holes have been positioned in a plane to 50 jum Critica
122. ntal halls with limited access in particular minimiza tion of flammable material and of halogenated plastics At an early design stage the L3 collaboration made various studies to limit the use of flammable hydro carbons in its detectors A considerable reduction of fire risk was obtained by replacing for the uranium calorimeter the initially planned mixtures with high concentrations of n pentane by the inert argon CO mixture However to achieve the required performance in the precision muon chambers the ethane concentra tion could not be reduced below 38 5 resulting in the use of slightly more than 100 kg of hydrocarbons in L3 This choice was accepted by CERN Safety in view of a whole package of complementary safety features such as all metal detector walls double gaskets fusing of electric power sensitive leak detection recirculating gas systems with sensitive oxygen alarms and a dedicated ventilation system for the detector volumes The latter is combined with a system of smoke detection and a facility for injecting sufficient quantities of inert gas to stop any fire inside the detector 3 The magnet 3 1 General description of the magnet All the L3 detectors are mounted inside the huge solenoid coil which is surrounded by an iron yoke and closed at its ends by two poles equipped with hinged doors The main parameters of the magnet are listed in table 2 3 2 The coil The coil is made of industrial plates welded toge
123. o forming the fine sampling part while HC2 and HC3 are divided into three and two segraents respectively This segmentation scheme results in a total of 3960 signal towers for the two endcaps The tower structure of the HCEC is summarized in table 7 6 2 5 Services High voltage is provided to the chambers by 72 supply channels which are fanned out into a total of 180 lines Each line is connected to a group of 11 to 14 chambers Distribution boxes mounted on the back flanges of the d gt tector allow us to disconnect individ ual chambers from the high voltage system e g if a particular chamber draws excessive current Gas is supplied to the chambers via four input lines feeding a total of 20 circuits for the HC1 containers and 4 each for HC2 and HC3 Chambers served by the same circuit are connected serially to the gas supply There are between 38 and 128 chambers per circuit because due to the various chamber sizes within the HCEC it was possible to equalize to within 30 the resistance of Table 7 Segmentation tower structure of the hadron calorimeter end caps Longitudinal segment 1 2 3 HCl Number of U chr layers 7 6 5 Number of V chr layers 6 7 4 Number of towers U V 520 468 396 Number of amplifier boards Number of ADC modules HC2 Number of U chr layers 4 Number of V chr layers 5 Number of towers U V 344 Number of amplifier ooards Number of ADC modules HC3 Number of U chr layers Number of V chr laye
124. ode and guard wires allowing us to control the drift field the gas amplification and the zero potential position At nominal voltage settings in a 0 5 T magnetic field and at 740 mm Hg pressure the gas gain is 5 x 10 With an electric field of 1140 V cm in the drift region the drift angle due to the Lorentz force is 18 8 Fig 40b shows the computed drift velocity for con stant pressure temperature and magnetic field More precisely the time to distance conversion function x 1 v 6 B P x t in the drift cell has been mapped in test beam runs and its dependence on the track slope 6 magnetic field B and barometric pressure P was mea sured Corrections are at most a few hundred pm near the sense and cathode planes Without magnetic field the measured cosmic ray residuals of a chamber with a threshold equivalent to the 10 12th drifting electron reaching the wire and two hits dropped is 136 pm fig 4la The chamber resolution in a magnetic field depends on the distance from the wire plane and on the slope of the track Resolution across the cell varies from 110 um to a maximum of 250 um close to the sense wire Fig 4lb shows the rms resolution as a function of the distance from the sense wire The region of degraded resolution near the cathode plane is due to sloped tracks which are largely in the adjacent cell and to inhomo geneity in the electric field From the data we calculate that the overall rms chamber resolution is 168 um
125. of an lAm source allows us to monitor the gain of the amplifying chain The stability of the optical fiber system is estimated as follows n each sector of 240 crystals connected to the same mixer the xenon light amplitude measured in a given channel is normalized to the sum of the ampli tudes in the sector Over a period of 45 days the mean variation of these relative amplitudes for one complete half barrel is 0 2 rms This demonstrates the quality of the optical fiber system and of the electronics 9 1 6 Energv calibration The calorimeter was calibrated at CERN in the SPS X3 beam where an accuracy better than 1 was ob tained In turn each of the two fully equipped half bar rels was installed on a rotating table Sufficient statisti cal accuracy was achieved by recording about 1500 electrons for each crystal at 2 10 and 50 GeV c momenta Owing to the high resolution of the calorime ter this measurement requires a well tested procedure and control of systematic deviations from the nominal energy deposition The firs requirement is a beam spectrometer of adequate resolution stability and momentum repro ducibility comparable to the energy resolution of the BGO We measured the magnetic field of the bending 82 B Adeva et al The construction of the L3 experiment Table 11 Parameters of the X3 beam spectrometer Field reproducibility lt 0 1595 Field peak to peak ripple 10 at 10 GeV c Field slow drift 5x10 at
126. om the study on the systematic errors apply only in case that a particular beam parameter setting cannot be measured or predicted If a parameter can be measured e g by Table 13 Systematic uncertainties in relative luminosity measurement Parameter Typical Known Absolute at IP value to change in o for typical value x 100 um 15 pm 0 1 e 300 pm 10 um 1 5 yY 100 pm 5pm 0 1 0 12um lum 0 06 25 1 mm 0 7 mm 0 1 o 33 mm 0 5 mm 1 5 x 0 2 y rad 0 0 175 urad 5 prad 0 05 y 0 10 prad 0 v 175 urad 5 prad 0 05 the central detector it can clearly pe corrected for and its effects on systematics removed A study is under way to investigate the feasibility of measuring beam parame ters with the luminosity monitor itself Hence the stated results are worst case only Table 13 shows typical values for the parameters as they have been estimated by the LEP instrumentation group 43 The column labelled absolute change shows the corresponding change in the luminosity calibration Using a value of d 0 6 mrad the systematic error on the absolute iuminosity due to ignorance of beam parameter settings can be kept below the 1 level for values several times larger than the column labelled typical value in table 13 11 The cen rai track detector 11 1 Introduction The L3 central track detector is designed with the following goals detection of charged particles and precise measure ment of the location and direction of
127. on Monitor Monitor amp Carb Sonic e angola Console Histogram and s Event D sptay LEP Control i M m E m B3 AW LEPICS I I CERN Ona te Ethernet ee OOOO BOLIC Fig 118 The L3 online computer system Sutb Detector Consc e and Display server HSC70 The main data acquisition computer is a VAX 8800 Five smaller VAX devoted to each of the main detector components and one to monitor the trigger system are attached to the cluster These VAX also form via Ethernet a local area VAX cluster LAVC with a number of VAXstations used as intelli gent command consoles and as multiwindow graphics displays One of the VAXstations is used for the run control The slow control data are gathered by a series of VME crates and PC which are linked to the cluster via a second Ethernet The on ine software has been developed using the MODEL library package developed at CERN Starting stopping etc of both the event data flow and monitor ing processes is under operator direction via a master process running on the ruri control VAXstation When the run is active the event data arrive from the 3081 E into the CHI A process running on the 8800 transfers the event out of the CHI and into a buffer in the 8800 memory The process then assigns an event number and finishes the formatting A subsequent process picks up tne event from memory and writes it to the tape in machine independent format Concurrently with the event writing but at a lowe
128. oximately 200 crystals were exposed to the beam The measured resolution is shown in fig 96 which also include the high energy measurements as well as results from an early prototype detector The resolution at 180 MeV is better than 4 which is within the design goals of the detector Cosmic muons are used to monitor the calibration constants as measured at the test beam and to perform periodic calibrations in situ to ensure the stability of the energy response of the calorimeter 38 39 They allow a measurement of the possible variation of the light re sponse of the crystal along its major axis The expected muon rate useful to the calibration i e those muons crossing opposite faces of a crystal is about 500 day 5 4 d 3 S uad o2 1 M e o O Da Og O 1 2 4 10 20 40 E GeV Fig 96 Energy resolution Diamonds prototype 1985 squares barrel 1988 B Adeva et al The construction of the L3 cxperiment 4 N S l Sap SH NG on Sacre DES a A T A oo 5 4 M l M ene i We i miu CAL i cae f cm T e Beam axs i9 10 20cm Fig 97 BGO end cap longitudinal section 1 Crystals 2 Cell walls 3 Mechanical structure 4 Stiffening radial spokes 5 Supporting ring 6 Preamplifier board 7 Cooling ducts 8 Thermal shield 9 Optical fiber bundle N per horizontal crystal and
129. peak position found by the peak finding program n1 100 1 i V PRELIMINARY SINGLE WIRE RESOLUTION SECTOR 1 ANODE 30 gol 375 90 E og z 60 4 x 3 50 i p Ro O O OOOO O O O Ch ul x x z uj LI 4 20 FIT o A x B A l2 um cm B 3524m EN ME Esas aoe Ss O 5 10 5 20 25 30 35 DRIF TLENGTH x mm Fig 114 Single wire resolution from the time expansion cham ber operated at nominal conditions in a 15 GeV c pion test beam pulse at the extreme right is the time reference signal The triangles indicate the peak positions found offline by the same peak finder program running in the DRP operating in normal mode Fig 114 shows an example of the single wire resolution as a function of the track distance from the anode This has been obtained with a straight line fit on drift times determined from the DRP information in one of the TEC sectors The average resolution stays well below the design goal of 50 um 11 3 The Z detector The Z detector fig 0 consists of two cylindrical proportional chambers with cathode strip readout covering the outer cylinder of the TEC fig 103 The strips of the four cathodes form angles of 90 70 1 70 1 and 0 with respect to the beam direction respectively The three supporting cylinders are built of polyurethane foam as filling material reinforced by oe Z CHAMBER i 800 e z e 2 600 e Ww uJ C anni P
130. r priority monitoring processes are reading events out of the memory buffer on a sampling basis On the 8800 only those monitoring tasks which require data from more than one detector component are performed including the partial recon struction of events using the L3 offline software The outputs of these tasks are buffered on disk for event display and for further analysis A bank of eight VAX stations is available for this The data for the monitoring and control of individ ual detector components are acquired by the detector computers which can access the detector component data on a sampling basis possibly subject to predefined conditicns While the 8800 receives only the events accepted by the level 3 trigger the individual detecior computers can have access to all the events accepted by 98 B Adeva et al The construction of the L3 experiment the level 1 trigger nd also during calibration and setup periods Each one of these computers has a console VAXstation running processes for analysis and display of the data The L3 detector is monitored for overall safety and detector integrity The LEP General Safety Services microVAX continuously monitors conditions in the ex perimental area to detect and prevent hazards such as the accumulation of explosive gases This microVAX is linked to the LEP control room via Ethernet and should a dangerous situation begin to develop and not be responded to the power to the area is cut
131. r the event reconstruction and display mentioned above Filling it during data taking ensures that reconstruction results acquired online will be repro ducible offline Another function of the online database is to allow monitoring of these parameters over time The online database has a validity interval reaching back about one week which allows these parameters and their correlations to be histogrammed versus time 13 LEP3NET L3 s intercontinental computer network 13 1 The origin and evolution of LEP3NET L3 recognized in 1981 that completion of the detec tor and preparation of the data analysis would require a sophisticated system of computer links between the members of the Collaboration 68 69 Prior to data taking L3 physicists enginee s and technicians in the US Europe and Asia would need a means of instant electronic communication with their colleagues at CERN and at the collaborating institutions Because of the precision and diversity of the L3 detection systems and the richness of the L3 physics program the offline software development task is significantly larger than for any previous high energy physics experiment requir ing several hundred man years of work It was therefore vital to get the means to allow physicists to work efficiently while at their home sites and to coordinate with and contribute to the mainstream of software development work going on at CERN This led to LEP3NET L3 s intercontinental computer ne
132. ral radioactivity of uranium To this end the wires in each module are grouped to form readout towers fig 59 In the projection the towers point to the beam axis wiih a constant angular interval The segmentation is 9 in and Z for both kinds of modules and 10 8 in the radial direction for the long short modules In the Z projection the towers have a constant width whereas in the projection they point to the beam axis The number of wires in each tower depends on the position of the tower and ranges form 3 to 28 The granularity is Fig 57 An assembled hadron calorimeter ring 62 B Adeva et al The construction of the L3 experiment Fig 58 The assembled hadron calorimeter barrel highest at the front end of the module where higher shower densities are expected 5 6 Hadron calorimeter readout and electronics At a working high voltage of 1 6 kV the gas gain is about 10 This gives rise to an anode signal of about 50 fC per minimum ionizing particle passing through the chamber The signal wires within each tower are con nected in parallel The signals from the towers are brought via ribbon cables to connectors fixed to the base plate The total number of charge sensitive readout channels in the barrel is 23040 Electronically each channel consists of a preampli fier a 40 m cable a passive receiving network and a charge integrating ADC fig 60 The charge integrating preamplifier is a joint development
133. rch for che forward racks is performed first in 5 us from the beam crossing time After a further 5 us the search for the central tracks can begin The search is completed within 11 us At the end of the search a track adder module counts the total number of tracks found the number of clusters a cluster being defined as any number of contiguous tracks and the number of pairs of tracks with an acoplanarity angle smaller than some program mable value The level 1 TEC trigger decision is taken 96 B Adeva et al The construction of the L3 experiment on the basis of these numbers and the information on the tracks found is sent to the calorimetric trigger processor as well as to the level 2 trigger 12 4 4 Scintillator trigger The scintillator triggers based on the signals of 30 barrel and 32 end cap counters are the following Multiplicity trigger It requires a coincidence of two out of the 30 barrel counters and is used to trigger on cosmic muons in calibration runs During LEP oper ation this trigger is efficier for events with two muons or two hadronic jets thus it serves as a simple backup trigger for the 4i muon and cluster triggers and is useful to monitor their efficiency Coincidence pattern This allows us to select prede fined patterns of hits chosen from the barrel and the end cap counters and in coincidence with the beam gate The following trigger conditions can be pro grammed two counters in different
134. resolution of 50 ium in the measurement of the R coordinate was reached on a prototype 47 much time has been spent to understand the variation of this resolution Tie sampled track length influences the width of the electron arrival time distribution and the number of primary electrons collected at the anode An accep tance of about 2 4 mm was found to be a good compromise As the polar angle deviates from 90 the resolu tion improves due to increased electron statistics fig 112a When the azimuthal angle deviates fr m 0 with respect to an anode plane the resolution de teriorates due to a stretching of the arrivai times This effect is small for the L3 TEC since the maxi mum value of is 7 5 fig 112b r gt e W o tor C US 70 50 P o a 30 2 a a b O C tong C iBOu m IO AT 90 O 5 IO 5 20 25 AZIMUTHAL ANGLE Fig 112 Impact of changes in the polar angle A and in the azimuthal angle B on the A 4 resolution B Adeva et al The construction of the L3 experiment The range of drift velocities is given on the one hand by the wire geometry and the resulting high voltages and on the other by the possible 8 bunch operation of LEP Inside this range a measurement of the resolution as a function of drift velocity indicated an optimum of about 6 um ns A tuning of the shaping amplifier for symmetric output and the use of a 80 CO 20 iC H gas
135. rov et al CERN EP 88 84 15 A Arefiev et al Nucl Instr and Meth A275 1989 71 16 A Bobchenko et al Sov J Nucl 30 1979 805 S Denisov et al Nucl Phys B61 1973 62 J Allaby et al Sov L Nucl Phys 13 1971 295 Yu Gorin et al Sov J Nucl Phys 18 1974 173 17 Yu Galaktionov et al Nucl Instr and Meth A251 1986 258 A Arefiev et al Nucl Instr and Meth 285 1989 403 18 B Bleichert et al Nucl instr and Meth A254 1987 529 19 U Martyn and J F Zhou Nucl Instr and Meth A256 1987 143 101 20 R D Alessandro preprint DFF 72 1988 E Gallo preprint DFF 76 1988 21 M Bocciolini et al Nucl Instr and Meth A257 1987 509 22 M Bocciolini et al Nucl Instr and Meth A257 1988 548 23 R Sumner Nucl Instr and Meth A265 1988 252 24 For BGO machining see M Lebeau et J C Le Marec Usinage des cristaux de BGO LAPP Dossier ANVAR 2101 1985 M Lebeau et H Vey Rodage et polissage des cristaux de BGO LAPP Rapport interne 19 02 85 M Lebeau Recherche fondamentale et transfert tech nologique une collaboration LAPP Annecy SIC Shanghai Rev Technique APAVE 236 1986 67 25 M Schneegans Nucl Instr and Meth A257 1987 528 and Test and preparation of BGO crystals for the L3 calorimeter to be submitted to Nucl Instr and Meth 26 C Laviron and P Lecoq CERN report L3 416 1986 27 M Lebeau LAPP repor
136. rs Number of towers U V Number of amplifier boards Number of ADC modules stainless steel uranium tubes of detection chamber Lay ETE pP d po m L9 Wd RI hea SET M 3 f 1 EX h b LA a P LP Fig 68 Hadron calorimeter forward backward system In the lower part of the figure the segmentation of the HC con tainers in and z are shown separately for the U and V layers The upper part of the figure shows enlarged the O segmentation into roads at the border of the fine and coarse sampling parts All un C ON 39 344 324 288 252 2592 108 28 324 288 956 40 10 232 180 412 20 B Adeva et al The construction of the L3 experiment the various circuits The half ring containers themselves excluding the chamber volume can also be flushed serially i e one gas line per end cap either with the counting gas or with an inert gas e g CO N At the chamber outlet the gas density is stabilized using the signals of pressure transducers and temperature probes For temperature monitoring 84 Pt100 probes are mounted on chambers located at top bottom and along the centerline of each half ring container 6 2 6 Readout and trigger segmentation The 3960 tower signals are amplified bv 168 LRS 2724 preamplifier boards mounted on the rear flanges of the HCl and HC3 containers The signals are then fed through an approximately 55 m long w d pair cable to 43 LRS 1885F FA
137. rtainties in the Bnabha cross sec tion such as weak inierac on effects vacuum polari a tion or multiple photon emission beyond oder gt They contribute to the systematic errors fer an esti mated 1 10 2 2 Detector performance The limitations of tne luminosity detector contribute below the 1 level to the systematics Tbe total energy trigger and the requirement of good lateral shower 85 containment will produce no significant loss of Bhabha events Longitudinal shower containment is almost complete in 24 radiation lengths and can be ignored against the more incomplete lateral containment The tracking chamber resolution of better than 250 yum taken at the critical inner radius R 8 8 cm produces a luminosity error ot less then 0 6 per event and this will be made negligible by the statistics of a 3 h run Cham ber production tolerances as well as final alignment and survey should be below 100 um and thus contrib ute 0 2 to the systematics The chambers will be mounied in a fixed position on the beam pipe and they will be surveyed with respect to the LEP quadrupoles Chamber efficiencies have the most critical effect in the systematics They must be known to better than 1 per wire and must also be stable at that level Requiring hits on 3 out of 4 chambers reduces the systematic error to 0 3 Thus by monitoring any inefficiencies espe ciallv near the onset of full efficiency at R 8 8 cm the systematic error shc
138. s with approximately 20 urn thick copper 0 6 0 4 0 2 r RELATIVE GAS GAIN OF THE CHAMBER e i m 0 0 0 50 100 150 DAYS OF OPERATION Fig 55 Gain stability measurement T Fig 56 Hadron calorimeter module nickel alloy They are clamped into place in the support ing spacers The chambers are put onto uranium plates and fixed with epoxy at four points Long modules contain 60 planes of proportional chambers and 58 uranium plates plus the top stainless steel plate This plate which lies between the BGO calorimeter and the uranium plates is part of the 54 mm shielding for the BGO photodiodes Short modules contain 53 chambers and 51 plates of uranium plus the siainless steel plate The dimensional tolerances of the absorber plates and the chambers are very tight especially for the thickness The chambers with wires parallel to the beam axis are referred to as 9 chambers and the ones with their wires normal to the beam are called Z chambers All services ot the module are brought out through feedthroughs in the base plate The high voltage distribution system which is em bedded in the base has four independent channels serv ing the odd and even numbered chambers in each projection It contains one fuse per chamber which can be blown in a controlled fashion Thus if a chamber develops a serious problem it can be remotely discon nected without compromising the rest of the chambers B Adeva et al The construct
139. sstalk between them The resolution as function of was studied with a prototype of nearly the same length 828 mm but with a reduced diameter 246 mm fig 115 The average single track resolution over the full range 45 lt lt 135 was about 300 um The double track resolution was 7 mm at 90 confidence level 55 12 Trigger and data acquisition 12 1 Introduction After each beam crossing the trigger decides whether an e e interaction took place and if so whether the event should be recorded This function is performed at three levels of increasing complexity reducing the 45 kHz beam crossing rate to a few Hz of tape writing rate The quality of the accepted data is monitored The detector parameters are also monitored for detector calibration and for safety slow control The online computer system consists of a VAX 8800 clustered with five smaller VAX monitoring each of the main detector components and the trigger system The cluster is connected via Ethernet to VME crates and personal computers which acquire the slow control data and to VAX stations for event display and run control The data acquisition uses mostly FASTBUS chosen for its speed and flexibility The system includes ample buffering capacity to allow asynchronous operation without contributing to the dead time 12 2 Trigger ievels All the detectors are read by the front end electron ics for each beam crossing In addition to the main data each
140. ssures that wires within a chamber are precisely posi tioned with respect to each other The next step is to relate the wires in one chamber to those in the rest of the octant Straightness monitors similar to that of the precision bridges are part of the octant alignment system 4 fig 44 A precision piece containing two LED is attached to each end frame of an inner chamber An insulated brass pin referenced to the LED touches one wire of a signal plane The end bridge can be moved so that the wire just makes or breaks its electrical contact with the pin In this way the end bridge positions are set to within a few pm The middie and outer chambers have a similar system of pins touching wires These pin assemblies and thus the wire planes of opposing cham bers are kept at the precise cell separations of 101 500 mm by gauge blocks The assembly between middle chambers contains a lens and that between the outer chambers contains two quadrant diodes Each end thus has two straightness monitors which were calibrated on an optical bench Based on the readout of these systems the middle chamber can be moved on its titanium fiexture feet to bring the chamber centers into a straight line with an error smaller than 10 pm Readout of this system over a 4 day period 1s shown in fig 44c B Adeva et al The construction of the L3 experiment Photodiode MO Lens a LED M 20 X microns 0 Sate uiri cmm uz Ga
141. stal seed The crucible has the shape of the final crystal but is somewhat larger to allow for later machining Table 8 Main parameters of the BGO barrel Inside radius of barrel 52 cm Inside length of cylinder 100 cm Material in front of crystals 0 05 0 1 X Angular coverage 42 39 137 7 Number of crystals 2 X 3840 Number of crystal types in O 24 Number of crystals per typein 160 Crystal dimensions length 24 0 cm front face 2x2 cm back face 2 6 X 2 6 to 2 9x 2 9 cm volume 130 150 cm M M M M U i Table 9 BGO and Nal Tl properties BGO Nal TI Density g cm 7 13 3 67 Radiation length cm 1 12 2 59 Moliere radius cm 23 4 4 dE dx MeV cm 9 4 8 Interaction leagth cm 22 41 Refractive index 2 15 1 85 Waveicngth of maximum emission nm 480 410 Relative light output 8 15 100 Temperature coefficient of light yield 1 55 0 22 C at 25 C Lumin lifetime at room temperature us 0 3 0 23 Afterglow at 3 ms 0 005 0 5 5 Hygroscopic no yes The produced ingots are then cut to size and polished The tolerances in dimensions are dictated by the need to have a safe mounting in the support structure to gether with a minimum dead space between crystals from 300 um to 0 um in transverse dimensions from 400 um to 0 um in length less than 50 pm in planarity of all faces All crystals are truncated pyramids fig 82 and table 10 pointing to the interaction re
142. t L3 584 1986 28 M Chemarin IPN Lyon CERN L3 394 1985 M Chemarin and M El Kacimi CERN L3 88 621 1988 29 EUCLID a computer aided design software from MATRA DATAVISION France 30 Technical specification LAPP April 5 1985 Selected company Elicotteri Meridionali Centro Com positi Sud Gruppo Agusta Frosinone Italy CERN contract LB CR 987 June 6 1987 31 M Lebeau J Mater Sci Lett 4 1985 779 32 Finite Element Calculation code MODULEF developed by INRIA France 33 Diodes for the L3 BGO Calorimeter Quality tests and mouniing PITHA 88 21 34 M Goyot B Ille P Lebrun J P Martin Nucl Instr and Meth A263 1988 180 35 M Bosteels LEP IM MB Y N 1985 36 Roll bound Aluminum inflated cooling panels hot rolled by ALCAN SPA Milano lialy 37 J A tDakken et al CERN EP 89 16 P E Kaaret thesis Princeton University 1989 ref DoE ER 3672 50 38 J A Bakken et al Nucl instr and Meth A275 1989 81 39 B Borgia et al Internal Report Dept di Fisica Uni versit di Roma La Sapienza 925 1988 and Nucl Instr and M th A278 1989 699 40 J A Bakken et al Nucl Instr and Meth A270 1988 397 41 C Rippich Monte Carlo Results for the L3 Luminosity Monitor 1984 and L3 Technical Proposal 1983 chap 9 42 C Rippich CMU HEP 86 14 1986 this reference deals with an earlier version of the monitor design 43 G Von Holtey CERN
143. ther on the CERN site The selected material Anticorodal 041 with heat treatment 71 has 6 less conductiv ity than pure aluminum but better mechanical and welding properties Cooling is provided by two indepen dent circuits made of Extrudal 050 an aluminum alloy with high resistance to corrosion welded onto the inner and outer edges of the coil fig 4 The interturn insulation is made of 10 mm fiber glass plates covered with 0 2 mm of Mylar The 30 kA current is a com promise between the production capabilities of the aluminum supplier dimensions plates thickness flatness tolerances the investment in handling tools and manpower to manufacture the coil as well as the difficulties inherent to the transport of high currents The 168 turns coil is divided in 28 packages which are bolted together The mechanical rigidity of each package is insured by axial bolts All bolts are triple insulated The gaps between turns are closed with rubber joints to minimize heat transfer to the detector The packages rest on insulated bronze skates which follow the thermal expansion by moving on two rails embedded in the lower part of the magnet yoke The coil is fastened to one pole whereas at the other pole electrically in sulated air springs permit coil motion arising from thermal expansion An active thermal shield placed on the inside of the coil protects the detectors Table 2 Main parameters of the magnet Inside radius of the
144. tion Unlike the level 1 and level 2 triggers for which only the trigger data with coarse granularity and lower reso lution are available for the trigger decision the level 3 trigger has access to the complete digitized data with finer granularity and higher resolution A complicated algorithm written in FORTRAN and tested with the offline computer is executed The selection of good events is based on a the correlation of the energy deposited in the BGO and hadron calorimeters b the reconstruction of muon track in the Z chambers C the reconstruction of the vertex in the TEC chamber A steering program minimizes the processing time by properly arranging the sequence of the above calcula tions In case of a positive level 3 decision a service request SR signal is asserted on the output port and the FASTBUS computer interface CHI transfers the data from the emulator memory to the main data acquis on computer for tape writing 12 7 The online computers and software The online computer system of the L3 apparatus is sketched in fig 118 It is centered on a cluster of DEC VAX Three TA78 tape drives two TA90 cartridge drives and 10 GB of disks are connected via the 1 O 97 From FASTBUS DISK me SP v Spy MAIN Online 3 Mun Ch m Cat BGO TEC Trigger Host DISK Computer Computer FT ET TT TA man cont 1 T UE IBridge 1 i Term vw fh Pe Sery E Work Safety Detector Control Terminals idk Nim Stati
145. to maintain the preamplifier temperature be low 35 C and to minimize the gradients along the crystals To evacuate this heat each of the 320 24 chan nel boards is covered with a brass screen to which copper pipes have been soldered The pipes are con nected to cooling fluid circuits 35 These screens are positioned about 1 mm away from the electronics com ponents to achieve adequaie heat transfer A schematic layout of the preamplifier board and its thermal screen is shown in fig 91 To prevent heat transfer to the electromagnetic calorimeter from outside the calorime ter is surrounded by very thin active thermal shields connected to the fluid circuits The level 1 readout boxes each containing up to 32 readout electronics cards with 12 channels per card are installed on the end faces of the hadronic calorimeter barrel Each channel dissipates 2 W resulting in a total dissipation of 23 kW This heat has to be evacuated to maintain the electronics below 40 C A level 1 readout box 1s equipped with 17 built in cooiing screens sand wiched with readout cards Special screens were devel oped using the roll bond technique 36 Four boxes are connected to a cooling fluid circuit resulting in a total of two sets of four circuits A schematic view of a cooling box is given in fig 92 A lt xong fluid circuit 35 is basically composed of a vacuum pump a circulation pump a water cooled heat exchanger and a thermal valve to regulate th
146. tup rising above the Mb s range when LEP Phase I reaches its design luminosity These high speeds are primarily needed to support remote interactive terminal sessions rapid transmission of a minute fraction of the data file transfer for program development and physics analysis along with several other services Each physics group is expected to need network access to CERN in the 64 kb s range or higher This early analysis has since been confirmed by detailed analyses performed by the HEPNET Review Committee HRC for high energy physics as a whole 70 This has led the HRC to recommend provision of Mb s speeds on the ESNET backbones used by HEP by 1989 NSFNET also has planned links to Europe in this speed range aithough these links will be shared by B Adeva et al The construction of the L3 experiment 2 large broadly based scientific community In Europe several initiatives which seek to provide 2 Mb s links to CERN have been launched in response to the findings of study groups on the computing and networking needs for the LEP experiments 71 72 Although it is hoped that some of L3 s future high bandwidth networking needs will be satisfied by the expansion of general purpose networks such as ESNET or NSFNET in the US LEP3NET will con tinue to be L3 s primary network for the next few years Only LEP3NET will be able to provide the full range of higher level protocols rapid interactive access and guaranteed bandwidth
147. twork LEP3NET began in 1982 1983 as a pilot project involving links between Caltech DESY and CERN for the MARK J and L3 experiments The network was initially based on the public packet switching networks TELENET US DATEX P Federal Republic of Germany and TELEPAC Switzerland Public net work usage soon expanded to include links to Michigan Amsterdam Madrid and Aachen Dedicated links to Annecy Lyon Rome and Naples continued to be used with progressive migration wholly or in part to X25 The performance of international links using the public packet networks was a fraction of the maximum line speed typically 4 8 or 9 6 kb s in 1983 and the charges for the volume of data sent were quite high The very fact that charges were volume sensitive led to the unacceptable situation that monthly bills could rise to unpredictable levels These factors soon made leased lines the preferred solution Following a feasibility dem am m amp m d Il onsira on using a leased satellite link between Caitech and Princeton in 1985 1986 the network became oper ational on January 30 1986 when a transatlantic line was del vered by AT amp T and the Swiss PTT From then on the network lines and switches demonstrated mean time between failures measured in months and mean times to repair in hours in most cases The carrent topology of LEP3NET is shown in fig 119 LEP3NET is widely used by high energy physicists in the USA and in
148. uld be well below the 1 level 10 2 3 LEP beam backgrounds Minor contributions to the systematics come from background interactions mixed into the event sample Synchrotron radiation induced events are easily re moved by a suitable energy threshold in the trigger Off momentum electrons resulting from beam gas or beam wall interactions will be studied as part of the event sample They are identifiable as highly acoplanar events wh h violate the geometric trigger but are accepted by the energy trigger A study of these events will rermit a reliable backgrouna subtraction accurate to better than the 1 level 10 2 4 LEP beam parameters One of the most serious obstacles to keeping the luminosity systematics below the 1 level is the depen dence of the observed Bhabha rates on variations of the beam parameters at the interaction point The parame ters are predictions derived from single separated beam measurements Hence a major effort was undertaken to adequately eliminate the dependence of the Bhabha calibration on the precise values of the beam parame ters After consultation with the LEP instrumentation group the following parameters were picked as the most essential to control 43 IP position x gt y zy IP width lt a 6 o beam dispersion Co Y o X and angular beam offset x y gt A Monte Carlo study of the effects of beam parame ter values in combination with the trigger design and event se
149. uropean and USA HEP DECNET in a way that is transparent to most users In spite of the obvious value of a link between USA and European HEP and the great range of facilities offered by DECNET L3 itself continues to use the Coloured Books network software for most network traffic The Coloured Books software s greater immunity to network disturbances its more efficient use of the limited bandwidth and its availabil ity on non DEC machines make it the natural choice for most serious work Nevertheless LEP3NET does not impose a choice of network software Coloured Books and DECNET can coexist on the same machines using the same physical connections 13 4 Examples of LEP3NET usage Even before the official start of LEP3NET the Princeton group which is responsible for the BGO electronics and readout was making heavy use of a LEP3NET pilot project linking them to CERN via a Princeton Caltech satellite link and the Telenet gate way at Caltech The Princeton group has continued their heavy network usage which has permitted them to continue to contribute to the analysis of BGO test beam data Network access to the test beam data acquisition system has proved invaluable for ihe preparation and debugging of the system by experts who were usually not at CERN The muon chamber reconstruction software has been written almost exclusively by people resident in the USA but used the latest version of the SIGEL3 L3 Monte Carlo simulation program d
150. va et al The construction of the L3 experiment 10 mm thick uranium plates two 5 mm plates put together There are thus 77 chamber layers in the HC1 container The HC2 and HC3 containers with a depth of 27 and 23 chamber layers respectively are subdi vided into two compartments each again via 16 mm thick stainless steel walls All compartments are equipped with 10 mm thick absorber plates As for HC1 shielding considerations led to the replacement of the first 15 mm of uranium in HC2 by the same thickness of steel Table 6 summarizes the dimensions and mechanical properties of the HC1 HC2 and HC3 containers 6 2 2 Proportional chambers Within a half ring a chamber layer consists of four chambers each covering an inte val A 45 U layer The wires are stretched azimuthally to measure the polar angle O directly Even numbered chamber layers V Layers are rotated by A 22 5 with respect to the odd numbered ones This stereo angle between successive layers allows measurement of the coordinate orthogonal to O and the gaps between chambers do not coincide in successive layers Chambers whose wires would have crossed the boundary between two half rings are split into two halves Every second chamber layer is thus comprised of three full sized chambers and two half sized ones Details of the construction of the proportional tube chambers are illustrated in fig 66 The individual brass tubes have an inner cross section of 5
151. velength nm Fig 83 Optical transmission of a few crystals through their iull length as a function of the wavelength The crosses repre sent the minimum transmission values accepted cording all dimensions of a crystal by comparison with an accurate steel standard with the help of 39 inductive position sensors The accuracy of each measured point was better than 10 pm After the initial batches the crystal rejection rate for bad aspect dents scratches etc low transparency or incorrect dimensions was 0 2 The light collected at the large end face of a tapered crystal with its six faces polished increa es strongly with the distance from the light source to the large end face This is shown in fig 84 for a crystal viewed by a photomultiplier and illuminated by a Cs source Good linearity and energy resolution require a nearly uniform light collection efficiency as indicated by test measure ments and Monte Carlo simulation By coating the polished crystals with a 40 to 50 um thick layer of high reflectivity NES60 paint one obtains a nearly flat light collection efficiency curve fig 84 with a light output comparable to that reached with the best wrappings tested 160 270 um thick 800 N 600 500 400 300 Light output Channel number 200 100 0 V 2L 15 g 0 Distance to PM cm Fig 84 Light collection curves measured with a collimated 137Cs source running along the crystal main d
152. x 0 08 16 2 4 pulse charge pC Fig 72 Pulse height distribution of U noise signals About 10 of the gate openings capture a uranium signal Electro magnetic pickup noise is small as demonstrated by the dashed distribution observed while lowering the high voltage by 300 V from the standard operation point pushed further out This fact is used to correct for any changes of the detector gain due to environmental changes gas temperature pressure HV etc Spectra f are taken at regular intervals during run startup or end and compared to a reference spectrum r The actual gain g is determined from the minimum of the x function sk E sen Er 7 F oo oo k gt J 10 dx f r x dx which is a measure of the similarity between the test spectrum f and the reference spectrum r By compari son to directly measured spectra from Fe we have established that this calibration method works reliably for gain changes witiiin a factor of five fig 73 The variance of the gain g determined from the x minimization is inversely proportional to the number of uranium signals observed fig 74 The gain test spec trum information is recorded at full readout speed 395 precision run corresponding to 3000 entries takes approximately 150 s The iimited knowiedge of the reference spectrum determines the ultimate precision of this gain test method With one day of running refer ence spectra for 0 3 accurate gain tests
153. x 10 mm and a wall thickness of 0 3 mm The chamber is shielded on both sides against uranium irradiation by 0 7 mm thick brass plates To save the space required by decoupling capacitors high voltage is applied to the body of the tubes rather than to the wires A 0 2 mm thick poly carbonate foil glued between the tubes and the outer shielding plates provides the necessary insulation End regions are additionally protected against high voltage breakdowns by Kapton foil Each tube contains a 50 um diameter gold plated tungsten wire siretched with a 200 g tension and crimped on a gold plated Cu Be piece fixed inside the plastic chamber endpiece On one side of the chamber printed circuit boards soldered to pins on the Cu Be wire fixation pieces serve to trans mit the wire signals to the outside world Within a chamber serial circulation of the gas from one tube to the next is achieved via gas throughputs milled inside the plastic endpieces of 25 24 and 19 tubes respectively Overall the HCEC detector contains 2284 individual chambers with a tota of 54140 wires table 6 6 2 3 Absorber The azimuthal segmentation of the absorber is twice that of the chambers fig 67 shows the arrangement of 67 IT mm 2i b brass plate polycarbonatetoit brass tube gas channel endpiece brass tube polycarbonatetoil Cu Te wireclip PCB Fig 66 Hadron calorimeter forward backward system a Shows schematically
154. y 10 mm 40 mm slits The 10 mm slit accommodates the support of the LEP beam pipe while the 40 mm one allows for passage of the vacuum tube of an RFQ device for BGO crystal calibration 6 2 Technical descriptiox of the hadron calorimeter end caps The endcaps consist of stainless steel containers filled with alternating layers of brass tube proportional cham bers inner tube dimensions 5 mm X 10 mm and 5 mm and 10 mm thick absorber plates of depleted uranium fine and coarse sampling part respectively Over the end cap region the amount of material traversed by a U Cosmics Arbitrary units 0 20 40 60 80 100 120 Amplitude Fig 61 Calorimeter response to cosmic ray muons 65 70 60 NUMBER OF EVENTS A un o o CAI o 20 T e 4 0 11 0 18 0 250 320 39 0 ENERGY GEV Fig 62 Calorimeter response to 20 GeV pions particle originating at the interaction point varies be tween 6 and 7 nuclear absorption lengths 6 2 1 Containers The HC1 containers are subdivided into four com partments by 16 mm thick stainless steel walls provid ing structural rigidity Their thickness is chosen to ap 30 o Resolution 9 5 O Pion momentum GeV c Fig 63 Hadron calorimeter energy resolution The line is drawn through the points obtained with the BGO in front full circles and corresponds to o E 55 VE 50 66 B Adeva et al The construction of the L3 experiment n m
155. y of response fig 53 This measurement was performed using both y rays from the natural radioactivity of uranium fig 54 and cosmic rays Two independent tests were carried out to check the longevity of the chambers in a radioactive environment In the first test the chambers were irradiated with a 10 times more intense radioactive source over several days at the work ing high voltage In the second test the chambers were operated in the normai uranium plate chamber stack at 10 times higher proportional gas amplification fig 55 In both cases the chambers proved to be very stable and no aging effects were observed 5 3 Module design Both long and short modules have basically the same design The inner part of the module resembles a tower mounted on a 15 mm thick stainless steel base fig 56 60 B Adeva et al The construction of the L3 experiment 108 107 1 Pedestal EFERBE LLLLLII TII c uw e 4 Number of events e w 102 10 0 40 80 120 160 200 240 Amplitude arbitrary units Fig 54 Spectrum of signals caused by uranium radioactivity 1 normal self trigger 2 random gate tigger The base aud the top 15 mm thick stainless steel platc connected with four spacer bars constitute a supporting structure for the absorber chamber stack The spacers he along the parallel faces of the modules The 5 mm thick depleted uranium absorber plates are plated for safety reason
156. z 1 1 c il i 4 5 i V 2 UTERINE DE RENS RON RO M PIT CODE Ky IER ers ue nho S XSSC NE qS SSS pr P o og 5 10 15 20 25 39 Coil Package number Fig 30 Coil alignment Fig 29 The support tube entering the magnet 49 over several wires to obtain the final tracking accuracy Multiple sampling improves the resolution by a factor vn over the single wire resolution MI MM and MO sample the muon track n 16 24 and 16 times respec tively The resulting measurement error is s 6 2 a with e and e defined in fig 35 We use thin aluminum honeycomb with an average of 0 956 of a radiation length per two layers io enclose the middle chambers Using this design a multiple scattering induced sagitta error of lt 30 m at 50 GeV was reached Regarding pcint c above with these small sagitta values muons more energetic than 3 GeV will be con X20 Ye0 1s30338A EE kGausel pru EE ECC e ee ee ag ucc d M C ae dl i 5 pee k 4 8 in 4 i ne e 1 2 a 4 Distance along Z axis imi Fig 31 Measured field 50 B Adeva et al The construction of the L3 experiment Fig 32 End view a and side view b of the L3 detector Tracks in the muon chambers are those of a computer simu lated event of the type e e gt Z Higgs with Z gt nt n and Higgs bb fragmentation fined to one octant Therefore alignment is only critical

THE CONSTRUCTION OF THE L3 EXPERIMENT

Contents

Download Pdf Manuals

Related Search

Related Contents