Home

Design of the BGO-OD Tagging System and Test of a Detector

1. 4 4 POCal Plane si sori ei 4 5 Calculation of the Detector Geometry LL 4 5 1 Alignment of the Scintillator Bars 45 A 2 2 0 2 4 a xi ob Geor we e Se x 45 9 Complete Detector Layout 4 3 24 4 5 Roh Roo Boe db weh Roh 4 6 Simulation of the Energy Resolution 222 22 000 Final Design and Prototype Detector Sil PMI ASSeBIDIIGS 25 2 8 42 5 02 BE EB Er ILE AE Eres 9 2 DUES cee dodi ur dd ae BS A A O AE ede SIETE TETTO DI CHASSIS u a e med ian 5 4 The Complete Prototype Detector 0 0 13 17 17 17 18 19 20 20 21 21 25 26 21 21 28 3l 3l 97 32 33 33 34 37 37 38 39 43 43 45 48 49 Contents 7 6 Experimental Tests 59 6 1 Electronics Setup and Data Acquisition 2 CE a e 59 OL Components seu se ie EUR RAR a 59 6 1 2 Assembly of the Electronics 65 6 1 3 Readout and Data Acquisition lll 66 6 2 Test at the Crystal Barrel Experiment 0 67 6 2 1 Assembly of the Test Stand 2 00 02 000 67 0 2 2 Detector sees ya 8 0 heute a we td Geek Ge ea 68 6 2 3 First Experimental Data of the Test at the CB Experiment 69 0 5 Threshold Settings a ue uo u Geen ua cee BL GPS we se Bh ww Ge SS 1 12 6 4 Test at the BGO OD Experiment LL 73 6 4 1 Mechanical Construction and Electronics 73 6 4 2 Detector and Beam Settings 74 6 4 3 First
2. AI 5 g 90 80 70 60 50 40 30 20 10 O 30 20 10 0 10 20 30 40 50 60 70 AT nax C7g Figure 110 Probability that the FPGA recognizes a coincidence in dependence of the temporal distance channels 7 and 8 D FPGA Coincidences 145 At Sgo 30 20 10 0 10 20 30 40 50 60 70 At max Cg0 Figure 111 Probability that the FPGA recognizes a coincidence in dependence of the temporal distance channels 8 and 9
3. EV el a cs BEE dAjojoug 1ebbe FL HERE HERE HERE SC suJor Qny3sUl3 AMC T ke T T pw BI i ccu AA L909 uniumun v Je1JaJey i n juezuy E ZOO F6E zoo FLOL W 0 G IN 6 9 p c0 0 T9 1L 0 0 Frl 0l AN Figure 75 R ght s de of the top slide Appendix 120 sj U94UIY GNYISULS IUC Free fp wn Z0 0 T 1E0 IN AN Figure 76 Left s de of the bottom sl de 121 A Technical Drawings W 9 AAA AREA v SZ BUJOA qnu3sul3 AE T_T v Na H e c N g SS ai Da N J m D S o 02 ZN 0M H Ww AS g 1 e qp Medis 200 AE gt gt N Figure 77 Right side of the bottom slide 122 Appendix M3x0 5 6H Anzahl 3 Material Aluminium 6061 Datum Name __ freee 30 03 2010 siebke IT ead Ir Einschub Ruecken 5 E E E ul NN Anzahl 10 o Material Aluminium 6061 DS T_T Ls Rome SE ea ma T ki Halterung Scinti Tagger Prototyp 1 EY stats Ancerungen com fre ll US Figure 79 Clip used to fix the scintillator bars A Technical Drawings 123 Anzahl 9 Material Aluminium 6061 Datum __ eeesee 30032010 siebke EI I ha3 Hamamatsu Rohr 2 0 Spannhuelse Anzahl 9 Material Aluminium 6061 1L al HamamaTsu Dec
4. Then the time between the transit of the electron and the arrival of the ionisation electrons at the anode where most of the gas amplification takes place can differ by Ar 1mm 10cmus7 10ns gt Atpmr It is however possible to use a MWPC and scintillation counter together and measure position and time separately This method was used for the SAPHIR tagging system TOPAS II Bur96 3 Detectors making use of Cerenkov radiation are very fast since the light is emitted almost instantaneously when the electron traverses the material This light can be detected using PMTs The downside is that these detectors have to be rather big to maintain a sufficient light output which strongly affects the spatial resolution A lead glass Cerenkov detector is for example employed in the CB M ller Polarimeter Kam10 For the BGO OD tagging system the first method 1s chosen The use of a combined system of an MWPC and large scintillator bars limits the maximum electron rate which can be detected because each single PMT sees a substantial fraction of the total rate and the MWPC already saturates at small rates When using smaller scintillator bars the total rate can be increased At the same time the spatial resolution of the scintillator bars can be improved sufficiently so no additional position resolving detector is needed A positive side effect 1s the lower cost of a single detector compared to a combined system To further increase the resolution
5. 2 2 22 Co Emm 13 Coordinate system used in the simulation and dimensions of scintillator bars 14 Overview of the setting for the simulation I5 Calculation of the beam width LL L6 Simulated focal plate sia doses AA aa 17 Exemplary electron trajectories for equidistant energies and scintillator bars 18 Exemplary electron trajectories for equidistant energies and adjusted positions Of thescm llator Darts 202 2 due 2 02 5 Au RARA 19 Exemplary electron trajectories for equidistant energies and adjusted positions and widths of the scintillator bars 2 2 CC EEE nn 20 Possibilities for multiple electron events lcs 2 Staggering of the scintillator bars in multiple vertical planes 22 Calculated detector layout with constant and variable resolution 23 Resolution changeover in the vertical plane detector 24 Simulated energy distribution and resolution without radiator and with Cu ZOO MA ROCHE DOE onda eoi d ae Oy Bk BB ed 25 Exploded view of the PMT assembly 26 View of the back side of a Slide LL 27 Profile of the slides for the prototype detector 28 Chassis with one mounted PMT assembly 29 ESO Ude du ae o RO we n na 30 Assembly of the prototype detector 0 31 Block diagram of the electronics 22 2 noo 32 Simulated ADC spectrum of an ideal detector w
6. 5 trigger artefact 6 ADC signal For more details see the text 79 7 Data Analysis In the previous chapter the experimental setup as well as a first discussion of the raw spectra has been presented This chapter covers more complex analyses and shows important results about the usability of the prototype detector Except when otherwise quoted the data are taken from the dedicated test at the BGO OD site 7 1 Detection Efficiency of the Prototype An important property of a detector 1s its efficiency In the context of a tagging system different efficiency variations should be noted Here the tagging efficiency 1s defined as Py Al 40 Y N the ratio of the number of photons impinging on the hadronic target and the number of electrons detected in the tagging system Ideally both numbers would be equal to the total number of photons produced in the Bremsstrahlung radiator Ny Ne Ny total In reality they are reduced due to the following reasons 1 The photon beam is collimated after leaving the tagging magnet The amount of photons being removed in the collimator depends on the angular distribution of the Bremsstrah lung process and the dimensions of the collimator as well as the beam flaw 2 The hodoscope only covers a fixed energy range i e electrons outside of this range do not hit the detector in the first place This 1s a purely geometrical factor 3 The detector itself does not detect necessarily ea
7. Hadrons and Nuclei 28 139 148 2006 JACKSON J D Klassische Elektrodynamik Walter de Gruyter Berlin 4th edi tion 2006 KAMMER S Strahlpolarimetrie am CBELSA TAPS Experiment Ph D thesis Universit t Bonn 2010 References 109 Kle05 KLEINKNECHT K Detektoren f r Teilchenstrahlung B G Teubner Wiesbaden 4th edition 2005 KM59 KOCH H AND MOTZ J Bremsstrahlung Cross Section Formulas and Related Data Reviews of Modern Physics 31 4 920 955 1959 LD91 LYNCH G R AND DAHL O I Approximations to multiple Coulomb scattering Nuclear Instruments and Methods in Physics Research Section B Beam Interac tions with Materials and Atoms 58 1 6 10 1991 Leb02 LEBERIG M D Das COMPASS Triggersystem zur Messung des Gluonbeitrags AG zum Protonspin Ph D thesis Universitat Mainz 2002 LeC74 LeCROY research systems corporation CAMAC Model 2249A Technical Data 1974 Leo94 LEO W R Techniques for Nuclear and Particle Physics Experiments Springer Verlag Heidelberg 1994 lep10 LEPS Beamline Sept 2010 URL http www rcnp osaka u ac jp Divisions npi b lepsbl html Mes10 MESSI F In preparation Ph D thesis Universitat Bonn 2010 MKA 08 MCGEORGE J ET AL Upgrade of the Glasgow photon tagging spectrometer for Mainz MAMI C The European Physical Journal A Hadrons and Nuclei 37 129 137 2008 M 73 MULLER J W Dead time problems Nuclear Instruments amp Methods 112 1
8. Z Oo i DETECTEUR n laser I AIMANT NETTOYEUR Figure 6 Layout of the GRAAL beamline BAA 97 The method of internal tagging is e g used in the GRAAL experiment at the ESRF in Grenoble BAA 97 see Figure 6 An argon laser produces photons with wavelengths of 351 nm and 514nm The laser photons interact with the electron beam between two bending magnets over a distance of 6 5m During the backscattering on the Eg 6GeV electrons the photons acquire a maximum energy Of kmax 1 5GeV The scattered electrons are deflected by the bending magnet and are separated by at most 56mm from the electron beam The detector for the scattered electrons is located directly after the bending magnet at a minimum distance of 14mm to the beam Bremsstrahlung Tagging With Bremsstrahlung tagging the electron impinges on a thin about 100um radiator foil made of a high Z material e g copper The electrons emit Bremsstrahlung radiation with a certain probability when traversing this foil and are then guided into the spectrometer magnet Their deflection in the magnetic field depends on their energy loss during the Bremsstrahlung process By detecting the electrons spatially resolved in the tagging spectrometer their energy and thus the energy of the photons can be deduced There are three main differences of the photon spectra between the two methods 1 Itis apparent from Table 1 that the photon rates achieved with Bremsstrahlung taggin
9. The cross section for elastic n p scattering at 1 MeV is o 4b and de creases for higher energies The density of H atoms per volume in plastic is n 5 x 10 cm leading to a mean free path 1 L 5cm 28 no For a thickness of x 40mm as used for the SAPHIR tagging system TOPAS II Bur96 the probability for a reaction is then 1 exp x L 55 The recoil proton will produce scin tillation light and the neutron will be detected To reduce the neutron detection efficiency the scintillator bars of the BGO OD tagging system only have a thickness of x 5mm reducing the reaction probability to approximately 10 Together with the coincidence technique described below the detection efficiency is smaller than 1 To reduce the efficiency for charged particles not involved in the Bremsstrahlung process the scintillator bars are arranged at least half overlapping as in Figure 12 Electrons coming from the radiator will always hit two adjacent scintillator bars while this 1s only the case for a part of the background electrons Since electrons are detected with almost 100 efficiency 34 Requirements of the BGO OD Tagging System Cbg Cbrems Figure 12 Function of overlapping scintillator bars Bremsstrahlung electrons brems always hit the scintillator bars perpendicularly background electrons ep may come from different directions Neutrons n come out of the beam dump and go into the opposite direction onl
10. given on what is still to do to construct the complete tagging system Chapter 4 Detector Design To define the best positions for the single scintillation counters a simulation was set up which is able to compute the focal plane A further program has been created to calculate the best arrangement of the scintillator bars in the focal plane as well as in the vertical plane which is needed due to spatial limitations After setting the magnetic field the desired energy widths the size of the photomultiplier tubes and some other tuning parameters the complete layout can be generated without further input This way a design using three vertical planes with an energy width AE between 0 6 E 20MeV for Ey 3200MeV for the high energetic photons and 1 5 Eo 50MeV for the low energetic photons was generated Furthermore a modified simulation setup was used to estimate the resolution of the ho doscope In the focal plane no influence of the beam flaw can be observed Starting with the vertical plane the resolution becomes worse than the theoretical minimum of og AE V12 The worst simulation without including the radiator is og 0 56 E 18 MeV using a 200 um Cu radiator the resolution becomes og 0 63 Ep 20 MeV The best resolution for both sce narios is og 0 19 Ep 6MeV Chapter 5 Final Design and Prototype Detector After the desired positions of the scintillator bars have been calculated the mechanical construction for a
11. the focal point for more details see Gre00 The plane consisting of the focal points for different electron energies 1s the focal plane It will be calculated in Section 4 4 The Tagging Hodoscope Finally the scattered and deflected electrons enter the tagging hodoscope From the de tected position one obtains the scattered electron s energy and thus the Bremsstrahlung photon this number is calculated from the simulation see also Section 4 3 26 Basics of the Underlying Physical Processes energy given the primary electron energy It is obvious that the placement of the hodoscope into the focal plane of the tagging magnet increases the energy resolution There are different possible detectors to detect the deflected electrons 1 Scintillation counters using plastic scintillator and photomultiplier tubes offer a fast and precise measurement of the timing of incoming electrons Plastic scintillators can have a rise time of about 0 5 ns photomultiplier tubes have a transit time of some ns and a jitter of At 0 5ns Leo94 It is easily possible to manufacture plastic scintillator bars in the desired sizes down to certain limit given by the size of the PMTs and the required light output 2 In contrast to scintillation counters MWPCs offer a high spatial resolution of 100 um and smaller Gre00 However the timing resolution is not suitable to be used as refer ence Assume a wire spacing of 2mm and a drift velocity of 10cmus
12. 128 PE c T Ot Se a 133 D FPGA COmcmencCes suina 446464 645 KOR Dew A CRUS eS 141 List of Tables List of Tables X LR A Vi SI LU N m Properties of different photon tagging systems Properties of the Hamamatsu R7400U and the ET Enterprises 9111SB PMT Properties of the Saint Gobain BC 404 plastic scintillator 22 222 Beam spot size and angular divergence 2 00 05000 Probabilities for different multi hitevents 4 Settings for the test at the BGO OD site Efficiencies calculated from the coincidences Discriminator efficiencies uncorrected and corrected detector efficiencies List of Figures List of Figures I Overview of the BGO Open Dipole experiment 2 Overview of the Electron Stretcher Accelerator ELSA 3 Kinematics of the Bremsstrahlung process lll 00 4 4 Feynman graphs for Bremsstrahlung 5 Kinematics of the Compton backscattering process 6 Layoutof the GRAAL beamline rn 7 General scheme of a Bremsstrahlung tagging system 8 The Goniometer and the different radiators 00 4 9 Energy level diagram of an organic scintillator molecule 10 Construction of a photomultiplier tube 11 Side view of the available space for the tagging system 12 Function of overlapping scintillator bars
13. 140 160 180 200 channel Figure 45 Threshold curve ratio of the number of hits with entry in the TDC and all hits for channel 5 of the prototype detector 107 events The dashed dotted curve is the same as in Figure 39 and belongs to the right axis 6 4 Testat the BGO OD Experiment The second test was performed at the BGO OD beam line This test was dedicated to the prototype of the new tagging system so that effects on other experiments did not have to be considered This allowed for changing the extracted beam current and for interruptions of the beam to access the detector 6 4 1 Mechanical Construction and Electronics The existing holding structure of the old tagging system Bur96 could be modified easily using aluminium profiles which allow a placement of the prototype between the old hodoscope and 74 Experimental Tests Figure 44 View of the prototype detector mounted in the BGO OD area the tagging magnet see Figure 44 In contrast to the CB tagging system the tagging magnet of the BGO OD experiment is a vertical bend device The possibility to move the detector easily up and down allows to make measurements at different electron rates Figure 45 shows an overview of the complete tagging system as well as the employed electronics In addition to the modules needed for this test there are others mounted in the crates which are used for a different experiment For a detailed description see Section 6 1 2 To avoid th
14. 2 2 Bremsstrahlung 19 Z Z Figure 4 Feynman graphs for Bremsstrahlung Thus the simple approach is very close to the more exact quantum mechanical derivation Since the exact shape 1s not needed for the present work the energy distribution will mostly be approximated by dk do 10 Ok k 10 2 2 2 Angular Distribution The formula for the cross section which is differential in photon and electron emission angles is given in KM59 Z20 dk p dQ dQ p sin 0 d0x 6 0 6 TAU EUER A MR PT An kp q E pcos 0 MEO 4 D sin 00 yi 2 i 2pposin 0 sin 05 cos 4E Eo q Eg pocos 8o E pcos 0 Eo pocos 0 2 pe sin 0 p sin 00 2pposin 0 sin 05 cos E pcos 0 Eo pocos 0 11 q p po pp 2 pokcos 09 2pkcos 0 2pop cos 0 cos O J sinOsin0gcosQ 12 Using this as a starting point it can be derived BLP71 that the photon and the secondary electron move forwards in a narrow cone with an apex angle sai gt 13 also called the characteristic angle For a beam energy of Eg 3200 MeV this means c 0 16 mrad 14 20 Basics of the Underlying Physical Processes 2 2 3 Limitations of the Born Approximation The Born approximation requires that the kinetic energies of the initial and final electron are large enough to fulfil KM59 2TZO ZG 2TZO Bo b For fo P 1 and a radiator made of copper Z 26 27Za p 1 33 Consequently this ap
15. 3 To achieve the maximum efficiency for all channels the production of the light guides and the scintillator bars as well as the assembling has to be done with more care Especially air must not enter the glued joint An additional in crease of the efficiency is expected due to the new scintillator material which will be used for the final detector 7 3 Electron Rate Stability As mentioned in Chapter 3 3 it is desirable to use the tagging detector at rates as high as possible Due to the energy distribution of Bremsstrahlung photons dN dEy Ey the rate is not equally distributed over the complete hodoscope low energy photons being emitted more often than high energy photons If the tagging system covers photon energies of 10 90 Eo about 7 of all electrons hit the highest channel corresponding to the lowest photon energy assuming an energy width of 50MeV In this section the behaviour of the prototype for high rates will be investigated The principal reason for a decreased rate stability is the dead time of the involved components as explained in the next section In the subsequent sections different methods to investigate the behaviour of the detector at high rates are presented 7 2 1 The Effect of Dead Times on Observed Rates It is expected that at some point the rate seen with the prototype will become lower than the actual rate due to the dead time 7 of the used components One distinguishes two different kinds of dead t
16. 8 and 9 can be found in Appendix C The errors of the scaler rate are too small to be visible in this plot The line is fitted for n up to 4MHz No data exist for the gap at n 4 MHz 7 2 7 Dead Times In addition to the direct comparison of the real and the observed rates the extraction of the dead time of the scintillator photomultiplier tube combination 1s also tried It is not possible because of the behaviour of the discriminators When trying to fit Equation 59 or 60 to the observed rates the resulting dead times lie at about 25 ns This number 1s smaller than the dead time of the discriminator of 30ns and thus cannot be correct Probably the discriminator itself induces this incorrect result One possible origin could be pulses which are still over the threshold when the dead time ends making the discriminator immediately send the next signal The timing of this signal 1s incorrect and thus useless Nevertheless the dead time seems to be decreased because also hits shortly before the end of the dead time are counted This makes it Impossible to extract the dead time of the detector itself without further tests It 1s also not possible to use equation 96 Data Analysis 61 to separate the dead time of the discriminator from the dead time of the photomultiplier tube Without knowing the effective dead time of one of the components the rates would have to be measured up to even higher rates Only then the influence of the combined dead
17. Brems 2 0 10 6 90 Table I Properties of different photon tagging systems ny is the approximate photon rate See also FP09a for all entries except for MAX Lab 16 Introduction The in beam testing 1s presented in Chapter 6 Chapter 7 covers the analysis of the experimental data Finally a short summary is given in chapter 8 followed by a conclusion 17 2 Basics of the Underlying Physical Processes 2 1 System of Units and Symbols Throughout this work the natural system of units will be used which is defined by h c 1 1 Especially during theoretical calculations also Me 2 to further simplify complex expressions When using only the equivalence h c 1 energy momentum mass length time MeV units 3 When also using me 1 energy momentum mass length time 1 4 The following symbols will be used in this section Eo Po initial energy and momentum of the electron E p energy and momentum of the scattered electron k k energy and momentum of the emitted photon Bo D velocity of incident and scattered electron unless otherwise quoted Do D 1 05 0 angles of po and p with respect to k angle between the planes po k and p k dl element of solid angle sin 05d609d6 in the direction of k dQ element of solid angle sin d d in the direction of p q momentum transferred to the nucleus q po p K Outs RMS of the angle for multiple scattering pr
18. Eg 1 6GeV 2 4 GeV 2 6GeV 3 2GeV and 3 5 GeV As described in Section 40 Detector Design 2 4 2 the tagging magnet of the BGO OD experiment 1s of the same type as the CB tagging magnet but is driven with a smaller current To get the according field for the BGO OD tagging magnet the measured values have to be scaled For Ey 2 4 GeV the current in the CB tagging magnet is cp 669 62 A the current in the BGO OD tagging magnet is Ipgo op 579 90 A Fro10 So all measured values for the magnetic field have to be multiplied by IBGO OD CB 0 8660 29 to fit the BGO OD tagging magnet For higher beam energies the magnetic field does not increase linearly with the current in the magnet so that c will deviate from the ratio of the two currents The simulations in this chapter all base on the field map for Eo 3 2GeV without scaling Energies will always be quoted as fraction of Eo and as an absolute number valid for Ey 3 2GeV e g E 1096 Eg 320MeV To extract the electron trajectories from the simulation virtual sensitive planes are placed below and behind the tagging magnet When an electron hits one of these planes information about its momentum p and its position xo is stored Using this information the trajectory can be extrapolated easily for positions x at which the magnetic field can be neglected X Xo P 30 being an arbitrary number positive or negative Figure 14 shows the general setting of t
19. Experimental Data of the Test at the BGO OD Experiment 76 7 Data Analysis 79 7 1 Detection Efficiency of the Prototype 000084 79 7 1 1 Basic Idea of Efficiency Measurements and its Application to the Pro tOLV DE ac a a Bet G Sea oe an idee de 79 1 12 Observed Elnelensies uem 3 ERE AR SR a SI 7 1 3 Correction for Discriminator Thresholds 84 1 2 Electron Rate Stability i ade oe di o de Soe e em 88 7 2 1 The Effect of Dead Times on Observed Rates 88 12 2 Measurement Principle a uu icu orent wok ea ee 89 7 2 9 Election Beam Struchire ox a sewed ace hace e ee Bi 89 7 2 4 Scaler versus Primary Electron Current 90 Jo Dealer versus PDC i e oh Bees Botte Bide dde ned 02 1 2 0 Scaler versus Scaler c o u sw o t hore ne nn BH Hc Hew Ss os 95 25 Dead Times paella lA LA AI 95 7 3 FPGA Coincidence Matching 2222 CE Em 96 7 4 Comparison of Simulated and Measured Spectra 98 TAA Testat ine CBR Sie uo nce eee eos bob A 98 14 2 Testate BGO OD SIE 24 3 hill 99 7 4 3 The Usefulness of this Comparison 222 22 2 nn nn 101 8 Conclusion and Outlook 103 Sl Quilook ieri Bou oe vii BG Ss BS Beads She ds e itas es don 104 8 2 CONCUSSION gio ee AO a a aa A eo oou 105 References 107 9 Danksagung 111 Appendix 113 A Technical Drawinss e 2 due Qos Bat ia A Rn 113 Bi Triple Comcidences ucc eus a a a a
20. SAPHIR Detektor Ph D thesis Universitat Bonn 1996 CAE06 CAEN S p A User s Manual Mod V1190 VX1190 A B 128 64 Ch Multihit TDC 7th edition 2006 CER93 CERN Geneva GEANT Detector description and simulation tool 1993 CERN Program Library Long Write up W5013 che10 Chemical Search Engine Nov 2010 URL http www chemindustry com apps chemicals Chr04 CHRISTIANSEN J High Performance Time to Digital Converter CERN EP MIC 9 2004 Version 2 2 for HPTDC 1 3 CMA 09 CREDE V ET AL Photoproduction of n and n mesons off protons Phys Rev C 80 5 055202 Nov 2009 CRE96 CREATIVE ELECTRONIC SYSTEMS S A CBD 8210 CAMAC Branch Driver User s Manual 2nd edition 1996 DBB 00 D ANGELO A ET AL Generation of Compton backscattering y ray beams Nu clear Instruments and Methods in Physics Research Section A Accelerators Spec trometers Detectors and Associated Equipment 455 1 1 6 2000 108 EBB 09 Els07 els10a Els10b ET 09 Ewal0 FP09a FPO9b Fro10 Gen99 Gre00 HAB 03 Hamo04 Ham07 Ham10 Han10 H1106 Jac06 Kam10 References ELSNER D ET AL Linearly polarised photon beams at ELSA and measurement of the beam asymmetry in 7 photoproduction off the proton The European Phys ical Journal A Hadrons and Nuclei 39 373 381 2009 ELSNER D Untersuchung kleiner Partialwellenbeitr ge in der N he dominieren der R
21. The mean energy of each of these distributions is then associated to the cor responding c channel This way the c channels of the complete simulated detector are energy calibrated 50 Detector Design Figure 23 Resolution changeover from AE gt 0 6 Eo 20MeV to AE 0 9 Ep 30 MeV in the vertical plane detector The distance between the electrons is 0 3 E 10MeV Resolution Measurement In the second step realistic events which can consist of more than one electron are sim ulated Many electrons can hit the hodoscope in such a short time that they cannot be resolved temporally In this simulation they are simply counted as simultaneous The actual number of electrons per event is given by a Poissonian distribution see Section 4 5 2 The struck s channels of the different single electron events are then put together to form a multi electron event If one s channel is hit more than once this piece of information is lost It just looks like the channel was hit only once Starting from this pattern of s channels the corresponding c channels are reconstructed In some cases this can lead to a misidentification or a loss of an electron see Section 4 5 2 Again for each c channel an energy distribution of the real energies 1s created Due to the multi electron events this distribution will differ from the distri bution which was created for the calibration The energy resolution oz 1s defined as the standard deviation of this n
22. an input channel over a fixed time which is defined by a digital pulse on the gate input Because this is the same as the total charge going into this input channel this kind of ADC is also referred to as QDC After each measurement the ADC outputs a channel number x for each input channel which 1s related to the charge O as follows x Q c Xpedestal 36 where c is a conversion factor and xpedestal an Input channel dependant offset Figure 32 shows a simulated ADC spectrum for an ideal detector with two independent channels Either channel can release a trigger leading to a measurement of both detector chan nels The detector which released the trigger will then induce a charge on the ADC input while the other detector will induce no charge Looking at a fixed detector channel this leads to a narrow pedestal peak at xpedestal 100 and a Landau distributed pulse height spectrum well 33Low Voltage Differential Signal Charge to Digital Converter simulated by a short ROOT script this distribution describes the energy deposit of ionizing particles in thin material see Leo94 62 Experimental Tests 0 50 100 150 200 25 0 300 channel Figure 32 Simulated ADC spectrum of an ideal detector with two independent channels separated from this pedestal In reality both structures will be broadened due to different effects including a finite charge resolution of the ADC Pulse Splitter Since the analogue output signal of th
23. angle 1s not enough for all slides but each time the resolution decreases 1 e the scintillators strongly increase in size the angle has to be changed This way three different angles are needed for the vertical plane detector The slides at which the resolution changes are different to the other slides as they have different angles for the top and the bottom side and thus have the outline of a wedge Unfortunately the thicker part of this wedge lies at the back of the detector chassis implying that it cannot be pulled out without first removing one other slide This seems to be acceptable because the scintillator bars will be replaced much less often than the PMTs 26NBR Nitrile butadiene rubber 5 3 Chassis 55 o c qu a Ev E 2cm to magnet ax lcm Ocm Figure 27 Profile of the slides for the prototype detector to scale The numbered rectangles represent the scintillator bars The slides are made of three parts Figure 26 Two opposing parts hold the scintillators They are screwed to the third part which 1s located on the back of the system The back part contains two threads allowing the slide to be fixed in the chassis once it is moved into its final position The alternating placement of the phototubes 1s reflected in the design of the slides On one side a single mount for a scintillator and two lead throughs for the light guides are built The light guides are glued onto the scintillator bars On the oppos
24. constant normalisation factor In the following sections these methods will be explained in more detail and the results will be presented 7 2 3 Electron Beam Structure When analysing the experimental data with respect to electron rates one has to take care of the temporal structure of the beam extracted from ELSA When the stretcher ring 1s filled electrons are continuously extracted for about 4s After refilling the stretcher ring with electrons from the booster synchrotron for about 1 s the extraction starts again Hil06 One extraction period is called spill Figure 57 shows two selected complete spills taken from the whole measurement of about 2h where N is the number of hits in the scaler of channel 5 Most spills show a structure like the first complete one shown here The rate rises very fast and stays almost constant after a small drop In the end it drops to zero again Some spills differ from this structure with the rate changing during the complete spill time e g as the second complete spill in Figure 57 During this test roughly 10 20 of all spills deviated visibly from the ideal shape These structures imply two things when measuring rates 1 The rate must not be computed by simply summing the hit counts for a long time and dividing by this time This averaging would lead to an underestimation of the rate actually seen by the detector Therefore the rate is computed for each spill individually 43 Actually these n
25. from the measured data BGO OD 101 ErED board protolype dive a i 105 Back plane of the chassis 2 113 elt side Plane of hechassts nc deu a foie AA TET ce a DEL 114 Right side plane of the chassis 1 oo on 115 Bert side of the Middle slides s 24 2 4 0 24 A e 116 Right side of the middle slide 20 0 0000048 117 Lett side of the top shde as ee e homm Hw an 118 Right side or thetopshde ivi AA A A AAA 119 Leftside ofthe Bottom shde ci ssa Se SO GS ee ade 120 Right side of the bottom slide 2 2 nn 121 Back side of the slideS o 4 o 4 0 0 a o aa a Pr 122 Clip used to fix the scintillator bars LL 122 Cylinder of the PMT assembly llle 123 Cap Of the PMT assembly 4 4 4 Pa A ood ato a A E A 123 Part 1 of the cable lead through lll 124 Part 2 of the cable lead through 2 0 00 0008 124 54 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 List of Figures 11 Clip used to fix the PMT assembly on the chassis 125 E cornea es LEE 125 SemullatorD9ti zu gr rm Nas eS me SU es bs E Bee eI RE 126 Framework used to mount the prototype detector behind the CB tagging system 127 Exclusive coincidences of two channels andchannell 0 128 Exclusive coincidences of two channels and chamel2 129 Exclusive coincidences of two channels and ch
26. from the position the prototype detector was designed for one problem arose If the detector is mounted in its designed position and its back plane is parallel to the side of the magnet where the primary beam exits the electrons hit the scintillator bars perpendicular to their surface When arranged at another position the detector has to be rotated to accomplish This number is dimensionless The corresponding threshold voltage rises linearly with this value 6 2 Test at the Crystal Barrel Experiment 69 Figure 38 Top view of the CB tagging system The primary beam gets scattered on the bottom The produced photons leave the tagging magnet on the top The scattered electrons are deflected onto the tagging hodoscope The electronics of the prototype detector left side is located to the right of the tagging magnet this Hence the detector was positioned in a way that by visual judgement the scintillator bars of the prototype are aligned parallel to the scintillator bars of the CB tagging system Because this could not be done precisely two slightly different orientations were used at the position corresponding to high rates This should assure that at least for one measurement the scintillator bars are aligned almost correctly Afterwards the positions were measured relative to the magnet and used for a comparison with simulated data of the expected hit spectrum see Section 7 4 6 2 3 First Experimental Data of the Test at the CB Expe
27. in Figure 15 but it has to be taken into consideration that the figure only shows the projection of the scenario onto a plane parallel to x In fact the distance projected onto the x z plane the plane of the Bremsstrahlung electrons is used here instead of the spatial distance This way the focussing in y direction is not taken into account Now the beam width w lc is defined as w le JE alle au 42 Detector Design the RMS of all distances to the centre track The focal point Xfoca 1s then found easily by means of minimizing w with respect to ke Xfocal X0 c nia pc 32 Using this method the focal points for energies between E 9 4 E 300MeV and E 43 8 Eo 1400MeV with steps of AE 1 6 5MeV have been calculated For smaller energies electrons move into parts of the magnet without measured field values for higher energies the electrons already hit the beam dump before being focussed The simulated focal plane is indicated in Figure 16 pi X02 pc ES RU iic X0 c da lc NS lo P2 X0 1 Figure 15 Calculation of the beam width pc Xo c and p Xo are the momenta and the positions on the sensitive plane of the centre electron and the deflected electrons respectively le is the distance to the sensitive plane d l are the distances perpendicular to the centre track E 4496 Eg 1400MeV E 9 E focal plane 300MeV Figure 16 Simulated focal plane The red tracks indicate
28. new tagging system is developed in two steps First the optimum arrangement of the single scintillation counters without regarding the mechanical construction is calculated in this chapter The next chapter covers the actual mechanical design of a prototype detector made to test the design before the complete hodoscope is built Due to the spatial limitations and the two types of photomultiplier tubes the tagging ho doscope will consist of three different areas 1 The focal plane detector will be located almost parallel to the bottom side of the tagging magnet Aside from the calculation of the focal plane itself the construction of the ho doscope for this area will be easier than the design of the remaining detector due to the large space which 1s available for each channel This will be shown in section 4 4 where the focal plane is calculated The focal plane detector will be built using the ET 9111SB PMT 2 The vertical plane detector will be located to the front side of the beam dump The size of the scintillator bars 1s decreasing when going from the bottom in direction of the primary beam whereas the electron rate increases So for the lower part of the vertical plane detector the slightly larger ET 9111SB PMT will be used 3 The upper part of the vertical plane detector 1s exposed to the highest rates and needs small scintillator bars to achieve the desired resolution This makes the Hamamatsu R7400U best suitable for this
29. scattering It is caused by many small angle scattering processes mainly in the Coulomb field of the nuclei Ne glecting few large angle deflections the angular distribution may be approximated as Gaussian with an RMS value which 1s given by LD91 13 6MeV nel EL 0 038 In z 19 Po Xo Xo Oms is the RMS deflection angle of the scattering projected to a plane The RMS angle in the space is given by OU V20ys Here x Xo is the thickness of the scattering medium measured in radiation lengths 2 4 Principle of Photon Tagging 21 Y k e Eo B A A e E Y ko Figure 5 Kinematics of the Compton backscattering process 2 4 Principle of Photon Tagging As already pointed out in Section 1 there are mainly two different methods for producing highly energetic photon beams Bremsstrahlung tagging and Compton backscattering Both methods make use of a scattering process with accelerated electrons and for both the scattered electron is momentum analysed to infer the photon energy and the time of production 1 e tag the photon The two methods are presented next in general terms Then the method of Bremsstrahlung tagging is described in more detail 2 4 1 Methods of Photon Production Compton Backscattering It is possible to produce a beam of high energy photons by Compton scattering laser light against highly energetic electrons e g those produced in a storage ring BAA 97 BCD 90 When laser light with energy ko is inci
30. see section 6 1 1 TDC At about 300ns the spectrum is distorted by the ADC see Figure 49 Including these parts into the calculation would lead to a false reconstruction of the rate Because the prompt peak contributes a large amount of the total hits in the spectrum the used range starts immediately before the prompt peak and ends immediately before the distortion by the ADC 2 Not all hits of the selected range f f may be used as starting point for a distance mea surement When starting shortly after t in Figure 59 a longer distances to the next hit can be measured as when starting later b Starting at a later time shorter distances At are still measured whereas longer distances At are not counted leading to an overall suppression of long distances and an overestimation of the real rate To avoid this a maximum distance Afmax has to be defined Only distances up to Afmax may be counted and only if the first hit occurred before t Atmax c When the first hit arrives after t2 Afmax no distance is measured d This way distances up to Atmax are counted without distortion 7 2 Electron Rate Stability 93 At Ats lt Almax po I 1 oeo o oe a c AAA At At lt Atmax oe b d bd A PA ty to 1 Alma 12 Figure 59 Measurement of temporal distances The black dots represent entries n
31. tagging magnet e y c NS 3 K 3 thickness x a b Figure 13 a Coordinate system used in the simulation b Dimensions of scintillator bars 22a construct in OOP Object oriented Programming languages like C or Java 23Bourne Again SHell a command language interpreter 4 3 Simulation of the Magnetic Field of the Tagging Magnet 39 Scintillator Dimensions The dimensions of the scintillator bar will be called as in Figure 13 b The thickness 18 given by the dimension of the material The width and the length can be chosen in a wide range Beam Flaw In the best of cases all electrons enter the tagging magnet at the same point and their tracks are parallel to each other To begin with the electron beam hitting the Bremsstrahlung radiator 1s not perfect It shows a finite spot size and an angular divergence due to the magnetic optics in the external beam line and due the emittance The spot size is the geometrical size of the electron beam on the plane perpendicular to the direction of motion It 1s approximated by the Gaussian widths oP and o in vertical and horizontal direction The angular divergence is defined by the Gaussian widths ol and 00 Approximate values for these properties are taken from Els07 see there for more details When the electron beam hits the radiator it undergoes further angular deflections due to multiple scattering whereas the spot size 1s not influenced All these deviations from
32. the PMT assemblies fit These have a depth of 5mm and their inner diameter is made fit to the PMT assemblies Directly outside of the chassis the metal cylinder of the PMT assembly has a smaller outer diameter see Figure 25 Screwed mounting 56 Final Design and Prototype Detector s NS groove for slide Sd gt VOS front g IN mounting clip for PMT assembly slot Figure 28 Chassis with one mounted PMT assembly isometric scale 1 2 5 The electron beam comes from the front clips which fit into this recess are used to attach the assembly to the chassis By mounting the assemblies firmly into the slots no light comes from the outside into the chassis The slots have holes with the same diameter as the holes in the PMT assemblies and the slides Once the slides and the PMT assemblies are mounted the PMT is shifted through all three consecutive holes until it touches the light guide 5 4 The Complete Prototype Detector All parts of the prototype detector were manufactured in the mechanical workshop of the HISKP in Bonn Since the final scintillator material was not available at the time of pro duction a spare piece of unknown and unneeded scintillator material had to be used To match the cross section of the scintillator bars to the cross section of the PMT light guides as shown in Figure 29 are used The assembling procedure consists of different steps shown in Figure 30 1 7 1 The light gu
33. the described behaviour 6 3 Threshold Settings As shown in Section 6 2 3 the thresholds of the discriminator were set slightly too low There fore even reflections and small pulse fragments were counted as real hits Since this s as unwanted as throwing away real hits the adjustment of the thresholds has to be done with care Clearly the threshold should lie between the pedestal peak and the signal peak When these are too close together it is possible to increase the high voltage of the photomultiplier tubes to within a certain range which leads to higher signals and to a better separation of the peaks The reflected signals however are amplified too To get the position the threshold lies in the ADC spectrum the following method is ap plied The ADC always records the energy deposit for all channels even those where the energy is below the threshold of the discriminator This 1s because it suffices that only one channel has a signal above the threshold to start the readout If a signal is above the threshold it will not only have an entry in the ADC but also in the TDC However this is only true for those hits arriving at the ADC while the ADC gate 1s open hits arriving earlier or later are not counted by the ADC This time window can be read off in Figure 36 In other words if there 1s an entry in the TDC during this time the entry in the ADC corresponds to a signal above the threshold Selecting only those entries in the ADC s
34. the gates themselves and between the gates and the n and outputs can be re programmed externally enabling the user to build complex logical circuits without having to construct special hardware and to change the function of this module as one needs For this test one XILINX Spartan 3 XC3S1500 module houses the trigger logic the co incidence matching and the scalers It was programmed by D Hamman Ham10 The in and outputs are implemented on separate mezzanine modules of which up to three can be attached to the FPGA board 40 Field Programmable Gate Array 6 1 Electronics Setup and Data Acquisition 65 VME computer Aside from the VMEbus form factor the VME computer or VME CPU is an ordinary computer running a customized GNU Linux operating system Amongst others it possesses two Ethernet 100MB s and 1GBs controllers USB ports and a VME interface It per forms different tasks Most important it runs the DAQ software controlling and reading out the TDCs the ADCs and the scalers see Section 6 1 3 Furthermore it reprograms the FPGA and sets the thresholds of the discriminator via its serial interface Using the slow control Han10 a system which allows for the setting and monitoring of many different parameters related to the complete experiment e g temperatures voltages it is possible to set the thresholds remotely without accessing the experimental area directly HV supply The High Voltage HV supply use
35. the power supply In contrast to NIM the CAMAC and VMEbus systems feature a backplane allowing for communication between different modules in one crate VMEbus offering the higher data transmission rate up to 320MB s7 compared to about 8MBs for CAMAC In addition there are special modules which can connect a CAMAC crate to a VMEbus crate allowing for communication between them 28Nuclear Instrument Module 22Computer Automated Measurement And Control 30Versa Module Eurocard bus lin the CAMAC Crate a Joerger Crate Controller Type A 2 Model CCA 2 in the VME crate a CES CBD8210 Branch Highway CRE96 60 Experimental Tests 3 7 5 ma HV supply FE E aa TDC gie x CPU Sr i E lt Discriminator i 2 20ns iz gt mi us 2 i FPGA pecus Vidia 22 3 PMT scaler trigger va a controller 2 coincidence Ae a 50ns gt Un c INT 5 5 m lt gt D 152 5ns MM ZZ es Gee DON ADC Kiri ei y Crate controller analogue Geray lne digital LVDS control readout Figure 31 Block diagram of the electronics The values in brackets are used in the second test Trigger Logic The goal of the measurements with the BGO OD experiment is the investigation of hadronic states Compared to the electromagnetic background the cross section for the pro duction of an interesting hadronic state is very small To assure that enough of these states are produced the experiment will be run with very high ph
36. the scaler which are used for the calculat on of distances the grey dots are not used Without additional constraint short temporal distances are favoured over long distances a When introducing a maximum distance this issue no longer appears b For details see the text Since only few distances can be measured for each event one spill does not provide enough data to calculate the rate reliably Therefore the data are collected for all spills with a similar rate in the scaler with steps of 0 5 MHz For each step the real rate is then computed by fitting an exponential function to the spectrum of distances Still the statistical error of the real rate is quite large Figure 60 making the results barely usable at least for channels 1 4 The plots for channel 5 and 8 show an expected behaviour as the scaler rate rises linearly with the real rate up to about 4MHz and then starts to lose hits This is also the case for channel 9 but just as for channels 6 and 7 the rate measured with the TDC decreases with respect to the rate from the scaler starting at some point For channel 9 the TDC starts to lose hits starting at about r 10MHz This is not a failure of the prototype detector itself but rather a limitation of the TDC as will be explained below The HPTDC chips Chr04 on the TDC provide for four groups of eight input channels making 32 input channels in total For each group there is a single buffer collecting the data from all ch
37. this is an advantage since the measurements cannot influence each other On the other hand it is difficult to assign to each rate measurement the correct current measurement Both values are stored together with a timing information which can yet be different The 7 2 Electron Rate Stability 91 n MHz channel 1 n MHz channel 6 n MHz channel 9 O 100 200 300 400 500 000 700 800 900 1000 I pA Figure 56 Scaler rate of channels 1 6 and 9 vs extracted electron current The other channels can be found in Appendix C Shown are the statistical errors the errors of the scaler rate are too small to be visible in this plot The line 1s fitted for up to 600pA 92 Data Analysis current measurement is stored some seconds after the spill ended but the displacement is not constant making the measurement very error prone The statistical error of this ambiguous current measurement 1s estimated by the standard deviation of three consecutive measurements the middle one most probably belonging to the corresponding spill The uncertainty is quite large as can be seen in Figure 58 There the rate measured with the scaler is plotted against the extracted current A line is fitted for currents up to 500mA and shows roughly the range within which the rate increases linearly with the current It is conspicuous that the rate increases linearly with the current up to about 600 pA and then starts rising more slowly independently of the cha
38. time would differ significantly from the single dead times allowing the calculation of both dead times with a fit of Equation 61 to the observed curve To summarize this section it can be said that a rate of 4MHz 1s possible for each channel without a significant loss Scaling this number to the complete hodoscope leads to more than SOMHZ for the tagging rate Even higher rates are possible when small losses are accepted Therefore the results fully conform to the requirements 73 FPGA Coincidence Matching Another important object of investigation is the test whether the FPGA recognizes all coinci dences between two s channels channel corresponding to one PMT correctly This is crucial since this information is part of the trigger condition in the experiment Therefore the coinci dences are reconstructed from the individual TDC events and compared with the TDC events of the corresponding c channels here channel corresponding to the coincidence of two neigh bouring PMTs A priori it is not known which hit in one s channel belongs to which hit in a neighbouring s channel Of course hits belonging to a single electron should arrive within few ns but the FPGA is not aware of this So each combination of two hits from two neighbouring s channels sj and s is counted as a possible coincidence s here grey lines in Figure 62 Assuming that the hit in s is prior to the hits in s it does not suffice to take only the hit in s which fo
39. 0nm Gre00 The light output of a scintillation material 1 e the number of emitted photons is typically measured relative to the light output of anthracene an organic crystal In anthracene an elec tron loses in average ant 60eV per emitted photon The light output of plastic scintillators lies around 60 of anthracene so that amp 100eV Leo94 2 5 2 Photomultiplier Tubes A photomultiplier tube PMT is a device which is able to convert very faint light pulses down to single photons into an electric signal A simple layout is shown in Figure 10 After the photons pass through the input window faceplate they hit the photocathode Due to the pho toelectric effect photoelectrons are emitted The probability for a single photon to produce 0Multi Wire Proportional Chambers H PolyStyrene trichloro nitro methane CCI3NO che10 PPolyvinylIToluene 1 ethenyl 2 methylbenzene 1 ethenyl 3 methylbenzene 1 ethenyl 4 methylbenzene C27 H30 che 1 0 13 4 di phenyl benzene C1gH14 che10 142_phenyl 5 4 phenylphenyl 1 3 4 oxadiazole C70H14N20 che10 15 pyridin 3 ylbutan 1 on CoH NO che10 165_phenyl 2 4 5 phenyl 1 3 oxazol 2 yl phenyl 1 3 oxazole C24H16N20 che10 28 Basics of the Underlying Physical Processes FOCUSING ELECTRODE SECONDARY ELECTRON LAST DYNODE STEMPIN VACUUM DIRECTION OF LIGHT FACEPLATE ELECTRON MULTIPLIER DYNODES PHOTO
40. 2 47 57 1973 O R10 O RIELLY V The Near Threshold Pion Production Program at MAX lab In American Physical Society Division of Nuclear Physics Spring Meeting Washing ton DC USA 2010 Pus10 PUSCH T In preparation Ph D thesis Universit t Bonn 2010 SAA10 SCHMIDT C ET AL The explora analysis software in preparation CB Note 2010 Sa105 Saint Gobain Ceramics amp Plastics Inc BC 400 BC 404 BC 408 BC 412 BC 416 Premium Plastic Scintillators 2005 SBB 94 SCHWILLE W ET AL Design and construction of the SAPHIR detector Nuclear Instruments and Methods in Physics Research Section A Accelerators Spectrom eters Detectors and Associated Equipment 344 3 470 486 1994 Sch09 SCHEDLER M Optimierung von Hochfrequenz Intensit tsmonitoren am Elektro nenbeschleuniger ELSA Bachelor thesis Universitat Bonn 2009 Sie76 SIEBKE W Die Hom otypie bei den Hepatahydratsulfaten M ssbauereffektun tersuchungen an den Mischkristallsystemen Fe Me SO47H50 Me Mg Ni Zn Co Ph D thesis Universit t Erlangen N rnberg Erlangen 1976 110 References T1m69 TIMM U Coherent Bremsstrahlung of Electrons in Crystals Fortschritte der Physik 17 12 765 808 969 Wal10 WALTHER D CAD drawings 2010 111 9 Danksagung An dieser Stelle m chte ich mich bei all denen bedanken die mich bei der Anfertigung dieser Diplomarbeit unterstiitzt haben F r die Vergabe des interessanten und he
41. 4 and 5 neighbouring s channels see Table 5 The probability P that this reconstruction is in fact correct is P 33 In the present case p lies between p 0 1 and p 0 3 meaning 0 77 lt P lt 0 91 So in at most 23 of these two electron events one electron is not reconstructed correctly Instead it is associated to a neighbouring c channel Therefore an additional error of AE 1 6 50MeV is introduced for this second electron The electron with the higher energy is always identified correctly as its energy is defined by the first s channel which is not hit see Section 4 5 1 This effect 1s incorporated in the simulation of the energy resolution see Section 4 6 no of neighbouring possible ratio of probabilities most probable s channels origins origin 2 1 0 1 p l O t lt 1 3 2 150 i 1 p 4EQp p p 0 1 4 352 1 p p 3 for p 0 6 5 4 3 1 p 1 3 0 single electron 1 two electrons in the same c channel 2 two electrons in adja cent c channels 3 two electrons separated by one intermediate c channel 4 two electrons separated by two intermediate c channels Table 5 Probabilities for different multi hit events 48 Detector Design 4 5 3 Complete Detector Layout The focal plane has enough space to place all scintillator bars in an optimum position 1 e in the focal points while maintaining enough space to place the PMTs More challenging is the vertical plane detector If the sci
42. 9 or 2 and 4 respectively will only produce a hit in one detector If two of these particles are detected within a short time in the channels 7 and 9 they will be counted as coincident In this case it seems that channel 8 missed a particle reducing the observed efficiency 3 It has to be ensured that the discriminator does not miss suitable signals coming from the photomultiplier tubes This can happen if the thresholds are set above the beginning of the signal peak thereby removing real electron signals As will be seen later this is the case here Therefore it will be tried in Section 7 1 3 to correct for this In the following section the efficiencies of channels 2 to 8 will be calculated without regarding the mentioned problems Section 7 1 3 covers the correction of these values for a reduced discriminator efficiency 7 1 2 Observed Efficiencies Before actually calculating the efficiencies the coincident rates for two and three hits have to be determined To make sure that every hit of one particle in different channels 1s counted hits are considered as coincident if they are detected within Afcoine 10ns This time span should cover all delays of the different channels with respect to each other To reduce possible background events the coincidences are computed as exclusive This means that a coincidence of two channels implies that no other channel was hit within Afgoinc That way accidental coincidences are sorted
43. A Figure 63 shows the probability Pij At sij At ci id Maia 62 that hits separated by Ar s y axis are recognised as coincident between 30ns and Afmax Cij x axis see also Figure 62 b For example p 10ns 20ns 1 which means that it does not take more than 20ns for the FPGA to send a coincidence signal for all hits with a distance of 10ns The reason why negative values for Ar c are possible originates in different pulse recognition in the TDC and the FPGA In principle the coincidence signal from the FPGA should be delayed with respect to the single hits But when the TDC uses the trailing edge of the pulse of the discriminator and the FPGA uses the leading edge the timing information of the single hits is delayed by the signal length of the discriminator 20 ns That way the coincidence signal can be seen prior to the single hits The most interesting part of Figure 63 1s the red filled area in the lower right The value for pij is exactly one for Ar s lt 15ns and Atmax cij gt 3ns implying that all single hits with a distance of at most 15ns are seen as coincident after at most 3ns Also on the other channels coincidences are found with a probability of 100 For distances bigger than 15 ns the probability decreases probably due to a non constant signal length of the discriminator This is also the reason why not all coincidences are seen within a constant time Since the FPGA and the TDC do not use the
44. B experiment in July 2010 with the detector located behind the CB tagging system This provided a first check for the prototype and the read out The second test was dedicated to the prototype and took place in the BGO OD area close to the final position of the new tagging system Before describing the individual tests in detail the electronics and the data acquisition common to both tests are introduced 6 1 Electronics Setup and Data Acquisition The complete data collection can be split into a hardware and a software part On the one hand the electronics which transforms the analogue output of the photomultiplier tubes into digital signals incorporating information about signal height and timing On the other hand the data acquisition DAQ Ham10 a software which controls the electronic components reads out and stores all gathered data 6 1 1 Components Figure 31 shows a block diagram of the complete electronics setup For a better understanding of the complete system the components are presented individually in this section The next section explains the assembly of all these components Electronic Standards All electronic components are packed into modules which follow widely used standards in nuclear and high energy physics These are NIM CAMAC and VMEbus Common to all of them is that they provide a standardised mechanical and electrical interface Multiple modules of one type can be put into a crate which also provides
45. BASE SOC VIII BARS EROE EN na RUN DRS RENT DEI TEC ec DE BI REG EG ER SENT INT LAN Heel ENS ew REN HA BR CART IS NN TOO RIT CER SAO Powe DREI UN PREC SON FOO TEST BR TERN 500mY Ch2 500mY lt M 40 0ns A Ch3 X 480mV Ch3 500mVY 2 Ch4 500mV 2 uv 162 800ns Figure 36 Timing of the different signals The input signal red line activates the trigger signal cyan line and the gate signal blue line The delayed input green line reaches the ADC while the gate is open The input signal is a square signal generated by a function generator time is needed by the ADC and the VME computer to process the input data 3 Resetting the electronics to the initial state ready to record the next events The readout of the TDC and the scalers can be done directly through the VME backplane while the ADC cannot be read out directly To accomplish this the CAMAC crate controller is connected to a VME CAMAC interface sitting in the VME crate This way the VME computer has access to the ADC The data are finally stored in a ROOT file on a network drive easily accessible for the analysis 6 2 Test at the Crystal Barrel Experiment The first test of the prototype was done in parallel to a regular measurement of the Crystal Barrel experiment in July 2010 Because this beam time was not dedicated to the prototype it had to be mounted behind the tagging system of CB to avoid any influence on the ongoing experiment After describing the experimenta
46. CATHODE THBV3_0201EA Figure 10 Construction of a Photomultiplier Tubes Ham07 an electron is called the quantum efficiency The quantum efficiency depends strongly on its wavelength The maximum quantum efficiency is typically about 25 Next dynodes are con nected to different high voltages in a way that the voltage increases along the flight path of the electrons This way the electrons from the photocathode are accelerated until they hit the first dynode and produce more free electrons This is repeated several times until the electrons are collected at the anode The total gain or multiplication of the PMT is the number of output elec trons divided by the number of photons Gains of about 10 can be achieved There are other kinds of dynode layouts but the amplification principle is the same for all PMTs Because the gain depends strongly on the focussing of the electrons onto the dynodes already weak mag netic field can lead to a decrease of gain by distorting the flight path of the electrons To shield the PMT from external magnetic fields a layer of high permeable metal e g Mumetal can be wrapped about the tube This shielding should be longer as the PMT itself and exceed the photocathode by at least the radius of the shielding Ham07 The dynode voltages are usually obtained with a simple voltage divider circuit which is connected to a single high voltage source about 0 5kV 2kV The combination of the socket wh
47. D Electron Stretcher Accelerator ELSA l A f IN AAN c S E A Seui SS h P s Vm s N B _ BN mm Dipole horizontal extracti nsepa N A mm Dipole vertical hadron ye Ep XS x o r3 Quadrupole physi superconducting O H DORIS cavity Skew Quadrupole i 3 r Solenoid E x PETRA cavi mm Sextupole experiments SE cavity mm Combined Function Magnet B NN SNA Crystal Ba Tn NI m Solenoid M p M m E mm Radio Frequency tune jump Na lt n quadrupole N 2 N N x Et gr Ut PN AS rne stretcher ring WU PS Ne EN 0 5 3 5 GeV b DIS Na WEE NS XL a Y NN x m e xm gin ME if l N ti Sa injection septa EN IN ZK o 9 VAL XOX QD Sa AX tune jump e 5 o AVS quadrupole J y gt LINAC 1 x Ne EN half cell of stretcher ring O
48. Design of the BGO OD Tagging System and Test of a Detector Prototype von Georg Siebke Diplomarbeit in Physik angefertigt im Physikalischen Institut vorgelegt der Mathematisch Naturwissenschaftlichen Fakultat der Rheinischen Friedrich Wilhelms Universit t Bonn Bonn November 2010 Das Bild auf der Titelseite zeigt ein Photo des Elektronenstrahls hinter dem Magneten der Photonenmarkierungsanlage Siehe Kapitel 6 4 2 Ich versichere dass ich diese Arbeit selbstindig verfasst und keine anderen als die angegebenen Quellen und Hilfsmittel benutzt sowie die Zitate kenntlich gemacht habe Georg Siebke Referent Prof Dr Hartmut Schmieden Koreferent Prof Dr Kai Thomas Brinkmann Zusammenfassung Auch wenn das Verhalten der kleinsten bekannten Materiebausteine der Quarks bei hohen Energien sehr gut verstanden ist so gibt es noch immer ungel ste Fragen auf der Ebene der Ha dronen mit Protonen und Neutronen als prominentesten Vertretern Um deren Struktur weiter zu erforschen wird zur Zeit das BGO OD Experiment am Elektronenbeschleuniger ELSA in Bonn aufgebaut Ziel des Experimentes ist die Anregung von Nukleonen z B in einem Fl s sigwasserstofftarget mittels hochenergetischer Photonen Die bei dem Zerfall des angeregten Nukleons entstehenden Teilchen werden zum einen im zentralen BGO Ball nachgewiesen der sensitiv auf geladene und ungeladene Teilchen ist Die Spuren von nahe der Strahlrichtung emit tierten geladenen Tei
49. N 20 MeV P a VAI P V E EN JA 8 E N upol s NN 4 9 quadr RY gt S 74 ce electr PS 4 ES a u laboratory E 1 T L r nei Il j Figure 2 Overview of the Electron Stretcher Accelerator ELSA els10a Some components of the BGO OD experiment are missing in this picture energy photon rate and the tagged range of the photon energy The concept of photon tagging will be described in detail in Chapter 2 4 This thesis covers the development of the tagging hodoscope This part of the tagging system detects the electrons which were scattered during the Bremsstrahlung process The focus of the study is primarily on the part which detects high energetic electrons and is exposed to the highest rates The readout electronics is developed in Mes10 The Bremsstrahlung target is part of Bel10 After describing the basics in Chapter 2 the requirements for the new tagging system are defined in Chapter 3 Based on the requirements the general design for the detector is developed in Chapter 4 The building of a small prototype is described in chapter 5 Experiment Method Ey max GeV ny s MeV Ey Ey max CLAS JLab FP09a Brems 6 0 104 20 95 SAPHIR ELSA SBB 94 Brems 28 10 32 93 CB ELSA CMA 09 Brems 3 2 104 9 9 LEPS SPring 8 lep10 Compton 2 4 10 60 100 GRAAL ESRF BAA 97 Compton 1 7 105 33 100 A2 MAMI C MKAt08 Brems 1 5 10 5 93 MAX Lab O R10 Brul0
50. N 1000 1 2 3 A 5 6 7 8 9 10 channel Figure 88 Exclusive coincidences of two channels and channel 1 600 500 400 300 200 100 channel channel B Triple Coincidences 1 2 3 A 5 6 7 8 9 10 channel Figure 89 Exclusive coincidences of two channels and channel 2 1 2 3 A 5 6 7 8 9 10 channel Figure 90 Exclusive coincidences of two channels and channel 3 600 500 400 300 200 100 600 500 400 300 200 100 129 130 channel channel Appendix 1 2 3 4 5 6 7 8 9 10 channel Figure 91 Exclusive coincidences of two channels and channel 4 1 2 3 4 5 6 7 8 9 10 channel Figure 92 Exclusive coincidences of two channels and channel 5 600 500 400 300 200 100 900 800 700 600 500 400 300 200 100 channel B Triple Coincidences D z e ES o 1 1 2 3 4 5 6 7 8 9 10 channel Figure 93 Exclusive coincidences of two channels and channel 6 N 1000 1400 1200 1000 800 600 400 200 8 9 10 o channel Figure 94 Exclusive coincidences of two channels and channel 7 131 132 channel Appendix N 1000 E 1400 e ES o 1200 1000 800 600 400 200 2 3 4 5 6 7 8 9 10 i channel Figure 95 Exclusive coincidences of two channels and channel 8 N 1000 1000 800 600 400 200 8 9 10 o channel Figure 96 Exclusive coincidences of two cha
51. a flawed or non polished surface of the scintillator and the light guide as well as air between the different components e g in the glue film leads to additional losses which potentially lead to an efficiency smaller than 100 Moreover electrons which hit only an edge of the scintillator will produce less photons in the first place and are detected with a lower efficiency 30 Basics of the Underlying Physical Processes 31 3 Requirements of the BGO OD Tagging System Several aspects have to be considered when designing the tagging system for the BGO OD experiment The experiment itself makes demands on the energy resolution and the precision of the timing An additional emphasis is placed on a straightforward and easily maintainable system as the tagging system has to be always completely ready for operation The largest constraint for the detector design is the spatial situation Only a limited amount of space is available between the tagging magnet and the beam dump 3 1 Spatial Restrictions The arrangement of the tagging magnet and the beam dump could only be changed by a major rebuilding of the experimental site and therefore provides a fixed restriction for the design of the tagging system Figure 11 shows a drawing of the tagging magnet and the beam dump the latter constituting the main spatial restriction The magnet is oriented in a way such that the electrons entering from the left are deflected towards the ground As expla
52. amel3 129 Exclusive coincidences of two channels and chamel4 130 Exclusive coincidences of two channels and chamel35 130 Exclusive coincidences of two channels and chamel6 131 Exclusive coincidences of two channels and channel 7 131 Exclusive coincidences of two channels and chamel8 132 Exclusive coincidences of two channels and chamel9 132 Scaler rate vs current in ELSA chamell3 134 Scaler rate vs current in ELSA channel 4 6 135 Scaler rate vs current in ELSA channel 7 9 2 oo oo en 136 Scaler rate vs reconstructed rate from the TDC channels 13 137 Scaler rate vs reconstructed rate from the TDC channels 46 138 Scaler rate vs reconstructed rate from the TDC channels 7 9 139 Scaler rate vs scaler rate from the lowest channel channels 7 9 140 Probability that the FPGA recognizes a coincidence channels 1 and 2 141 Probability that the FPGA recognizes a coincidence channels 2 and 3 142 Probability that the FPGA recognizes a coincidence channels 3 and 4 142 Probability that the FPGA recognizes a coincidence channels 4 and 5 143 Probability that the FPGA recognizes a coincidence channels 5 and 6 143 Probability that the FPGA recognizes a coincidence channels 6 and 7 144 Probability that the FPGA recognizes a coinc
53. ane b The procedure is repeated until the next scintillator bar fits into the first vertical plane c The complete staggering procedure is then iterated until the full energy range is covered The energy width AE and the energy range as well as the minimum distance between the edges of two scintillator bars can be adjusted The latter has to be chosen appropriately to leave enough space for the placement of the photomultiplier tube Furthermore the minimum overlap between three consecutive scintillator bars and the increment in the scintillator width can be set see Section 4 5 1 When starting with a energy width of AE 0 6 E 20MeV for low energies the to tal number of planes increases up to six see Figure 22 a This is clearly too high as a large number of planes complicates the mechanical construction and increases the impact of multiple scattering in the scintillator To reduce this number the energy width is enlarged at two points leading to a jump in the width of the scintillator bars This way the number of planes can be reduced to three see Figure 22 b The complete layout is shown in Figure 22 c Starting from AE gt 0 6 Eo 20MeV for Ey 3200MeV the energy width is in qu Y JE jd Y Figure 21 Staggering of the scintillator bars in multiple vertical planes For a description of the procedure see the text 4 6 Simulation of the Energy Resolution 49 a b c Figure 22 Calculated detector layout wit
54. angement of the scintillator bars 1s depicted in figure 19 One possible problem arises by this layout Multiple electrons which are close in energy and time may be identified falsely or lost This will be examined in the next section 4 5 2 Multiple Hits The layout using an overlap of three scintillator bars can produce ambiguous patterns of struck s channels if two or more electrons hit the hodoscope If the temporal distance between two electrons is smaller than the time which the detectors and the electronics can resolve the elec trons cannot be distinguished This is not a problem as long as these electrons hit distant s channels but if two or more electrons hit nearby s channels there is a certain probability that these electrons cannot be identified correctly Only the case of two electrons will be looked at here because the probability for more than two electrons in near s channels 1s sufficiently low as will be shown later There are several possibilities for such misidentifications which depend on the chosen detector layout When using a layout with exactly half overlapping scintillator bars only electrons whose energy corresponds to the same c channel lead to an error as only one of them will be detected If two electrons hit neighbouring c channels they can be distin guished correctly It is different 1f the scintillator bars are more than half overlapping since now a single electron can hit either two or three s channels see Fig
55. annels of this group This buffer 1s read out with a clock rate of 40 MHz thus limiting the total rate for one group to this value However this rate can be achieved only if all hits arrive with the same temporal distance which is not the case here If the rate becomes too high for the TDC hits will be lost first in the last channels of one group explaining why channels 6 7 and 9 are affected the most Channels 1 7 belong to one group and channel 8 and 9 to another see also Appendix C for the plots for all channels To circumvent these limitations the channels of the detector can be mapped in a non triv ial way to the channels of the TDC The final tagging hodoscope will have about 100 channels Only a few of them will be exposed to rates critical for the TDC By grouping one of these together with seven low rate channels the total rate of one group can be limited The maxi mum rate for a single channel is 10 MHz Chr04 making rates up to 4MHz possible even for randomly distributed hits with a relative loss of less then 1073 Chr04 The second issue to consider is the correlation between the channels in one TDC group If the channels are highly correlated which is the case for geometrically neighbouring channels the maximum rate is further decreased as multiple hits will then occur at the same time So no neighbouring detec tor channels should be put in one group in the TDC The mapping between TDC channel and detector channel has then
56. araus ein Signal f r den Trigger gene riert Der Prototyp Detektor erf llt die Designziele hervorragend und kann als Grundlage f r den Bau des gesamten Hodoskops dienen Contents Contents Zusammenfassung List of Tables List of Figures 1 2 Introduction Basics of the Underlying Physical Processes 2 1 System of Units and Symbols LL 2 2 AAA AAA 222 L Enere y Distibullon 6m dl zu A A A 2 2 2 Angular Distribution e 2 2 3 Limitations of the Born Approximation 2 3 Multiple Seattenne sa a qo Be Xp a SO A Oe d d 2 4 Principle of Photon Tagging LL 2 4 1 Methods of Photon Production 02 00 2 4 2 Elements of a Bremsstrahlung Tagging System 2 Detector COMPOnentis ru een out Boe A S RE ae 252 SCMUlaLore 2 4 34 peri i i ae ee a 2 25 2 Photomultiplier Tubes 2 24246 eae PeR 2 5 9 Light Collection and Efficiency 2 4 2 ace wwe Hee ned Se chs Requirements of the BGO OD Tagging System o spatial RESTICUONS saa m uten Eu Red nen nennen tet 32 Energy Range and Resolution 222 2 CE m mm 3 3 Rate Stability and Timing 2 2222 oo oo 334 Maintenance 25 4 ii dt eh he Be es ne 3 9 Backetound 4 signora eue we ERA Ra 3 0 Selected PM Isand Scintillator so a s usia asia e ee ee Detector Design Al SOL Ware Tool es iaa iaa X AAA ei ar 4 GoncralRenaiks rad hh 4 3 Simulation of the Magnetic Field of the Tagging Magnet
57. at the low photon energy limit one can think of an additional scintillating fibre detector as used for the CB tagging system FP09a 2 5 Detector Components The functionality of plastic scintillator and photomultiplier tubes PMTs is explained in more detail in this section 2 5 Detector Components 24 S m internal p S E absorption To fluoresecence triplet states So singlet states Figure 9 Energy level diagram of an organic scintillator molecule Leo94 2 5 1 Scintillators A plastic scintillator is actually an organic scintillator dissolved in a plastic solvent Common solvents are PS and PVT A charged particle traversing through plastic scintillation material deposits ionisation en ergy in the solvent This energy is transferred very quickly to the actual scintillator e g p Terphenyl PBD and PBO The scintillator gets excited to a triplet state T T or to a singlet state S S see Figure 9 These states all decay to the S state via internal degradation without emitting radiation The S state decays radiatively with a high probability to a vibrational state of So Since the energy of the radiated photon is smaller than the distance between So and S the scintillator is transparent to its own radiation Usually a secondary scintillator like POPOP is added to shift the wavelength of the radiation to a more suitable value in the visible range about 42
58. ay in Mes10 At the moment there are tests of the FrED board which amplifies and discriminates the analogue signal see Figure 68 The future plan is to amplify and split actively the analogue output of the PMT with the AFA board as close as possible to the hodoscope The active splitting prevents crosstalk between the two outputs Multiple channels are collected on the B FrED board This board will take care of the discrimination and the connection to the slow control Using the new electronics further tests can be made providing information about the precise timing of the signals To increase the resolution for low photon energies one could possibly use an additional detector using scintillating fibres with a small diameter of a few mm Such a device 1s employed in the tagging system of the Crystal Barrel experiment FP09a However the minimum resolution depends on the shape of the electron beam see Section 4 6 For the 200um Cu radiator the worst resolution is og 20MeV for Eg 3200MeV The minimum resolution given by the width of the scintillator bars is Just Omin 14MeV Thus less than o 6 MeV corresponding to an energy width AE 17 MeV cannot be achieved 8 2 Conclusion A very promising design for the tagging hodoscope has been developed and tested during this thesis incorporating the demands of Chapter 3 All tests and analyses which were performed conform to the expectations By translating the mechanical design to the co
59. ces 141 D FPGA Coincidences See Section 7 3 for a description of these graphs At s 100 90 80 70 60 50 40 30 20 10 0 20 10 0 10 20 30 40 350 00 70 Af nih Cia Figure 104 Probability that the FPGA recognizes a coincidence in dependence of the temporal distance channels 1 and 2 142 Appendix P 100 90 80 70 60 50 40 30 20 10 10 0 10 20 30 40 50 60 70 30 20 At max C gt 3 At s Figure 105 Probability that the FPGA recognizes a coincidence in dependence of the temporal distance channels 2 and 3 P 34 100 90 80 70 60 50 40 30 20 10 0 30 20 10 0 10 20 30 40 50 60 70 At max C34 1 At S34 0 9 0 8 0 7 0 6 0 5 0 4 0 3 0 2 0 1 0 Figure 106 Probability that the FPGA recognizes a coincidence in dependence of the temporal distance channels 3 and 4 D FPGA Coincidences 143 30 20 10 0 10 20 30 40 50 60 70 At te o Figure 107 Probability that the FPGA recognizes a coincidence in dependence of the temporal distance channels 4 and 5 AI S5 0 30 20 10 0 10 20 30 40 50 60 70 At ax Cg Figure 106 Probability that the FPGA recognizes a coincidence in dependence of the temporal distance channels 5 and 6 144 Appendix 30 20 10 0 10 20 30 40 50 60 70 At ax C67 Figure 109 Probability that the FPGA recognizes a coincidence in dependence of the temporal distance channels 6 and 7 100
60. ch electron hitting it Since the photon rate was not measured number 1 is not accessible within this experiment Number 2 can be calculated using the energy range of the detector and the precise Brems strahlung cross section Number 3 is directly accessible within this experimental test The detector efficiency again can be decomposed into two factors The probability that a tempo rally isolated electron hitting a scintillator bar leads to a detectable signal from the phototube Edetect and a factor which arises from the finite dead time of the complete setup including the electronics Edead E Edetect dead 41 The latter will be discussed in Section 7 2 the examination of Egetect follows in this section Here Edetect Will be simply called e 7 1 1 Basic Idea of Efficiency Measurements and its Application to the Prototype To measure the efficiency of a detector for ionizing radiation the method depicted in Figure 50 can be used Electrons leaving some source e g a radioactive source or an electron accelerator 80 Data Analysis Figure 50 Simple efficiency measurement Electrons leaving a source penetrate three consec utive detector channels If a signal is detected in the first and the last channel there must be a signal in the middle channel too penetrate three consecutive detectors If one electron has been detected in the first and the last channel it necessarily also went through the intermediate channe
61. ching the shown spectrum If two signals follow in a short time it is more likely that the first one releases the trigger than the second one This way the count rate for small distances prior to the trigger 1s suppressed For large distances the count rate asymptotically reaches the same rate as after the prompt peak Scaler A scaler counts the number of signals between two trigger incidents even during the time when the remaining electronics is being read out This way the scaler sees all hits and not only those during a small window around the trigger time like the TDC does Each detector channel has its own scaler An additional scaler is connected to a fixed clock of 1MHz By dividing the number of entries of one scaler by the number of entries in this 1 MHz scaler one directly gets the mean rate in MHz since the last trigger As well as the trigger logic the scalers are implemented on an FPGA board see below Coincidence Matching The trigger will only be released when two neighbouring channels were hit The aim of using adjacent channels is to reduce the background see Section 3 5 The logic which computes these spatial and temporal coincidences is also implemented on an FPGA board but still has to be tested see Section 7 3 FPGA Roughly speaking an FPGA is an integrated circuit which consists of many logical gates and different in and outputs which are by default not connected to each other The connections between
62. count electrons or other ionizing particles can come basically from any direction The dotted lines in Figure 51 b indicate the regions which are accessible by particles going through channels 7 and 9 or 4 and 2 respectively Nearly all particles seen by channels 7 and 9 also hit channel 8 making it possible to determine the efficiency g of this channel Nevertheless only a small range which is covered by channels 4 and 2 is covered by channel 3 This implies that only a fraction of all particles seen by channels 4 and 2 can be detected by channel 3 When using the coincident count rates 754 and n234 to get the efficiency ez of channel 3 as in equation 42 it will be underestimated 2 Independently of the geometry x ray photons leaving the beam pipe or neutrons coming out of the beam dump can decrease the observed efficiency The probability for a neutron to interact with the scintillator 1s so small that it will interact with at most one of the bars The photons in contrast will be absorbed completely so that neutrons as well as photons 7 1 Detection Efficiency of the Prototype 81 Ey My LET a b Figure 51 Possible trajectories of electrons in the detector a Side view of the scintillator bars of the prototype detector The solid lines show trajectories of electrons coming from the radiator b Two sections of the detector The dotted lines indicate the regions which are accessible for electrons traversing channels 7 and
63. cture This effect was even more prominent when first tests were made during the build up of the experiment The pedestal peak showed more spikes and had a width of more than 20 channels The situation could be improved by doing the following 1 Earthing all electronic components by using a braided copper wire connected to the hous ings of the crates and the pulse splitter The splitter box was electrically insulated from the metal construction on which the old hodoscope and the prototype is mounted 2 Exchanging the power packs of all crates for different ones 3 Connecting the complete electronic setup to another power point using another phase of the electric power supply All these actions resulted in narrowing the pedestal peak The reason for this is probably an interference prone electric power supply and high frequency noise induced by other electronics 6 4 Test at the BGO OD Experiment TI PMT High Voltage V Threshold Electron Rate 1 800 50 lowest 2 800 45 3 800 50 4 800 50 5 800 40 6 800 45 7 800 45 8 850 45 9 850 40 highest Table 6 Settings for the test at the BGO OD site 300 N 1000 250 200 150 100 50 too 170 140 160 180 200 channel Figure 47 Measured ADC spectrum using channel 5 of the prototype detector during the second test 107 events The dotted line is suppressed by a factor of 10 sitting nearby However the removal of electronic components from the rack which were not use
64. d for the photomultiplier tubes is a LeCroy System 1440 which can be loaded with up to 16 modules The PMTs are connected to a 1443N module providing 16 outputs of up to 2500 V negative high voltage The HV system is connected to the slow control and can be programmed remotely 6 1 2 Assembly of the Electronics The block diagram and a photograph of all electronic components which were used for the test of the prototype detector are shown in figures 31 and 35 respectively Apart from different cable lengths and the mechanical construction it 1s the same for the test at the CB experiment and the test at the BGO OD experiment In the beginning see left part of Figure 31 the signal of the phototubes is split and then used for analogue and digital processing as described in the next two paragraphs Digital Signal Chain One output of the splitter is carried to the discriminator to generate digital pulses for suitable signal heights The discriminator itself has two outputs per channel one being directly connected to the TDC the other to the FPGA The FPGA generates the coincidence signals and the trigger for the TDC as well as the gate signal for the ADC In addition the input signal 1s redirected to the scaler implemented on the same board Analogue Signal Chain The other output of the splitter is carried to the ADC The signal has to be delayed since the ADC has to receive the beginning of the gate from the FPGA before the analogue
65. d here did not yield a better signal After the complete test was finished the entire electrical installation in the BGO OD area has been renewed A further test will show if this solves this problem Figures 48 and 49 show the TDC spectrum of the same detector channel as in the previous test Figures 40 and 41 The reflections and the artefact resulting from the thresholds Figure 41 2 3 and 4 have been removed through the described measures The first peak in structure 6 is almost vanished whereas the size of the dip has increased This supports the justification in Section 6 2 3 as increasing the thresholds decreases the chance that a pulse accidentally is above the threshold but increases the chance that a pulse falls below this threshold when lowered by the signal coming from the ADC The falloff directly behind the prompt peak which has become visible results from the finite dead time of the discriminator Between two hits there is at least a distance of 30ns 78 Experimental Tests gt 10 10 10 10 10 2000 1500 1000 500 0 500 1000 t ns Figure 48 Measured TDC spectrum using channel 5 of the prototype detector during the second test 107 events The y axis is logarithmic 10 10 10 10 200 100 O 100 200 300 400 500 600 t ns Figure 49 Measured TDC spectrum using channel 5 of the prototype detector during the second test 10 events detail The y axis is logarithmic 1 prompt peak
66. dent on the electron beam at an angle of about 4 180 it is scattered backwards close to the direction of the incoming electrons Using 9 as the angle of the scattered photon with respect to the incoming photon beam and as the angle of the scattered photons with respect to the electron beam see Figure 5 the energy k of the scattered photon is DBB 00 D k o lit M 20 1 B cos 9 ko Eo 1 cos 97 In the extreme relativistic case D 1 Ey gt 1 04 05 180 9 lt 1 equation 20 can be approximated as AE ko J 21 1 4Eoko E00 The energy of the scattered photon is highly dependent on the emission angle When collimat ing the photon beam it is still necessary to use a tagging method to obtain the photon energy exactly For Compton backscattered photons two tagging methods exist internal and external For internal tagging the scattered electrons are momentum analysed by the magnets of the stor age ring The detectors are located very close to the main orbit of the storage ring For external tagging the scattered electrons are removed from the storage ring by an additional magnetic field and are analysed by an external tagging spectrometer similar to the Bremsstrahlung tag ging 22 Basics of the Underlying Physical Processes DE FAISCEAU dipole MONITEUR Y DE a magnet FAISCEAU CBLE CRYOGENIQUE tagging interaction detector ZONE gt Cn
67. e of 10 90 of Eo this implies a total rate of more than ntota 50 MHz for the complete hodoscope entirely fulfilling the requirements The later experiment will use the coincidence of two s channels as the trigger condition of the tagging system Therefore the FPGA which generates the coincidence signal for the trigger was tested Each single coincidence was recognised during the experimental test To check the correct work of the simulation program simulated c channel spectra were compared to the measured spectra The Crystal Barrel test showed that the simulation can be used to align the detector precisely at least for the high photon energy range The comparison for the BGO OD experiments has to be repeated with better statistics to confirm this 8 1 Outlook Further tests can be made using the already existing prototype 1 2 3 An energy calibration would show if the simulation predicts the electron trajectories cor rectly During this calibration Ey is varied in a large range and the current in the tagging magnet is swept to redirect the primary beam without radiator directly into the hodoscope During this test the beam intensity is strongly decreased see also FP09a The influence of an additional shielding against the magnetic field of the tagging magnet can be investigated The design of the prototype provides enough room to wrap two layers of Mumetal foil around the PMTs The precise timing of the proto
68. e beam direction horizontal and vertical The top stage rotates the plate around the beam axis the other two stages rotate it perpendicular to the beam axis The radiator plate is mounted back to back onto the goniometer The Radiator First the electron beam hits the radiator During their transit through the material the electrons undergo Bremsstrahlung with a certain probability resulting in a specific energetic and angular distribution see Section 2 2 For the BGO OD experiment multiple different radiators and parts for beam diagnostics are mounted on a round plate sitting on a goniometer A goniometer is an instrument consisting of different motorised stages allowing for a precise positioning and alignment of the radiator plate in multiple dimensions The high precision is mainly needed for the alignment of a diamond which is used for coherent Bremsstrahlung Currently a new goniometer Figure 8 a consisting of two linear and three rotation stages 1s installed Bel10 Figure 8 b shows the plate with the different radiators The indicated beam direction corresponds to the use of the diamond radiator Otherwise one of the other radiators can be moved into the beam by the bottom linear stage and the top rotation stage Three different copper radiators SOum 100um and 200um will be used to gener ate incoherent Bremsstrahlung Their thickness measured in radiation lengths is x Xo 3 5 x 1073 7 0 x 1073 and 14 0 x 1073 Horizonta
69. e cross section formula describing the charge distribution of the nucleus The inelastic scattering of electrons off the nucleus showed that the nucleus can itself be excited and that it consists of nucleons protons and neutrons It did not take long to discover that the nucleons also possess excited states like the A resonance and thus are not pointlike Eventually the nucleons were found to be made of two different quark flavours the up and the down quark today four more quark flavours are known charm strange top and bottom Beside nucleons other baryons are known all made of three quarks In addition to baryons there are the mesons consisting of one quark and one anti quark The simplest mesons made of up and down quarks are the pions All quarks come in three different colour charges which are charges of the strong inter action This interaction 1s responsible for the binding of the nucleus too as it consists only of positively charged protons and electrical neutral neutrons Without the attractive force of the strong interaction between nucleons to counterbalance the electromagnetic interaction stable nuclei could not exist The strong interaction however differs from the electromagnetic inter action by an important fact While the coupling strength e of the electromagnetic interaction decreases for larger distances the coupling strength s of the strong interaction increases This implies two phenomena When looking at sma
70. e field map of the CB tagging magnet has been measured for five different energies of the primary electron beam Bal10 Since the BGO OD tagging magnet is operated with a lower magnetic field for the same energies the field map has to be scaled accordingly using the ratio of the currents in the two magnets This is possible at least for beam energies up to Ey 2400 MeV because the currents are then smaller than 700 A This is discussed in more detail in Section 4 3 Another important feature of a dipole magnet like the tagging magnet is its focussing ability Electrons which do not enter the magnet at a central axis the z axis are deflected towards this axis Hence the magnet acts like a lens on the electron beam One reason for this focussing 1s the fringe field of the magnet The magnetic field inside of the magnet is almost constant in a certain range Outside of this range it ramps down to zero over a characteristic distance In the absence of sources electric currents the following equations holds VxB 0 V B 0 23 Hence the change of one component of the magnetic field induces also a change of the other components If the field Bo inside of the magnet points into the y direction there are also finite contributions to the x and z components at the exits of the magnet Gre00 B xy 02 By 24 B y 0 B 25 The fringe fields 5 and B vanish at the centre axis and focus electrons of the same energy into a small spot
71. e input channels relative to the trigger signal The trigger signal has a dedicated input port on the TDC Until a trigger occurs the TDC continuously measures the times of all input signals It then continues measuring for a defined time and stores all measured timing information between a fixed time before the trigger and a fixed time after the trigger In the continuous storage mode the TDC continuously stores the times of the input signals into an output buffer until it is reset This can lead to an overflow of the output buffer when the TDC cannot be read out sufficiently fast The CAEN V1190A is based on the CERN EPC MIC HPTDC chip Chr04 It provides 128 channels and a timing resolution of 100ns The range of the time window can be pro grammed and is set to 1800ns before and 1000ns after the trigger The trigger matching is based on an internal clock of 40MHz which means that the measured times can be displaced up to 25ns 1 40MHz relative to the trigger For this reason the trigger signal is not only used to trigger the TDC but is additionally put on one of the input channels By subtracting the time which is measured in this way from each measured time one gets all the times relative to the trigger with the highest possible resolution Figure 34 shows a simulated TDC spectrum for an ideal detector with a single channel and a mean rate of n 1MHz The temporal distance between two hits follows an exponen tial distribution im
72. e phototubes will be used as an input for the ADC as well as the discriminator the signal has to split somehow This can be achieved by using a passive pulse splitter as shown in Figure 33 One downside of this simple circuit is that 1t halves the size of the signal In addition resistors which have a slightly wrong size can lead to reflections and the outputs can influence each other as will be shown in Section 6 2 3 The situation can be improved by using an active pulse splitter which amplifies the signals and prevents cross talk This is planned for the final setup of the tagging system Mes10 Outs Figure 33 Passive pulse splitter R Z 3 Z being the cable impedance Leo94 Time to Digital Converter To obtain the timing information of hits in the detector a multi hit Time to Digital Con verter TDC is used The employed CAEN V1190A TDC CAE06 provides for two different operation modes the trigger matching mode and the continuous storage mode the former be ing the one employed and described here In this mode the TDC stores the time of the leading 6 1 Electronics Setup and Data Acquisition 63 10 10 10 10 10 10 1 1500 1000 500 0 500 1000 t ns Figure 34 Simulated TDC spectrum of an ideal detector with one channel and a mean rate of 1MHz The x axis shows the time of a hit relative to the trigger time with a bin width of Atpin Ins The y axis is logarithmic edge of digital signals on th
73. e problems due to reflected pulses as seen in Section 6 4 3 the cable length between the PMTs and the splitter was minimized In return longer cables had to be used between splitter and discriminator and ADC respectively see also Figure 31 values in brackets Besides these differences the electronic setup was the same as in the previous test 6 4 2 Detector and Beam Settings The detector was mounted at two different positions along the aluminium profiles one near its designed position see Section 4 5 3 the other at the topmost position possible without being hit by the primary electron beam to reach the maximum possible rate To identify this position and to avoid hitting the detector frame with the primary beam the detector was first mounted at a position which lies definitely below that point Then a Polaroid film was attached to the back of the prototype the upper part being exposed to the primary beam After switching on 6 4 Test at the BGO OD Experiment 75 b Figure 45 a Overview of the location for the BGO OD tagging system The old hodoscope was part of the SAPHIR tagging system b Electronics rack From top to bottom VME crate NIM crate unused module CAMAC crate delay lines the electron beam for 1 s 2s and processing the film the position of the secondary beam could be clearly identified see Figure 46 The beam photo also shows electrons which underwent Bremsstrahlung below the secondary beam The thi
74. e same if the detector is moved some cm To minimize the effects of background radiation the spectrum of single hits is not used but instead the actual double and triple coincidences are reconstructed as in Section 4 5 2 A quantitative measure for the agreement between simulation and experiment is then computed as follows 1 The mean rate for all channels is normalized to 1 for the real spectrum as well as for the simulated one see Figure 65 a b d 2 The rate for each channel of the real spectrum is divided by the corresponding rate of the simulated spectrum see Figure 65 c e 3 The standard deviation o of all eight ratios describes the quality of simulated spectrum Ideally the normalized simulated spectrum is equal to the normalized measured spectrum Just in this case Ni real i Vi 63 Ni sim gt c Q 64 Each deviation of the simulated spectrum leads to o gt 0 This quantity is then calculated for a between 15 0 and 25 0 with steps of 0 1 and 40000 events for each step Figure 66 shows the o dependence of amp for the range where o s reasonably small The smallest value for o is encountered at an angle of a 21 5 Omin 0 21 5 0 046 65 The Figures 65 b and c shows the simulated spectrum as well as the ratio for this particular angle 7 4 2 Test at the BGO OD Site This time the prototype was mounted very close to the primary electron beam leaving the tag ging magne
75. e small groups of them easily The complete construction can be roughly split into three different parts The overall chassis the PMT assemblies and the slides holding the scintillator bars To have more space available for each PMT assembly they are put alternately on the left or the right side of the Bremsstrahlung plane In the following the single components will be described The design is made using Autodesk Inventor 20097 all shown pictures are extracted from this software Technical drawings for all parts can be found in Appendix A 5 1 PMT Assemblies Base cable lead through bavonet cap O ring cvlinder PMT 3cm 2cm lcm Ocm Figure 25 Exploded view of the PMT assembly to scale The PMT assemblies consist of one metal cylinder into which the PMT together with its socket assembly fits exactly and is closed with a bayonet cap Figure 25 The signal and high voltage cables are guided through a pair of plastic pieces made of black POM which fit into a notch on the bayonet cap This way the light incidence is minimised and a safe mount for the cables is provided Between the cap and the socket assembly a metal spring 1s placed to assure 242 3D mechanical solid modeling design software for creating 3D digital prototypes 25POM Polyoxymethylene 54 Final Design and Prototype Detector a constant pressure of the phototube onto the scintillator bar This pressure is needed because the scintillator bars are not glu
76. east nai 10MHz over the complete energy range without significant losses This is roughly the rate which could be achieved with other tagging systems at ELSA Since only a small fraction of the photons produced in the radiator leads to an interesting interaction in the target even higher rates of nf 50MHz or more are desirable to further improve the situation For example the total cross section for the reaction yp gt X K is at most Otot 0 5 ub for a photon energy around Ey 1400MeV Ewal0 About 5 of all tagged photons have an energy between 1300 MeV and 1500 MeV assuming that all photons between 10 E and 90 Eo are tagged Using a liquid hydrogen target of x 2cm length p 0 07 gcm gt the reaction rate is N nc Ow P ke 5 ng 20 028 27 where Na is the Avogadro constant and A 1 g moll is the atomic weight of hydrogen Hence at a tagging rate of 50 MHz there will be only roughly one reaction per minute in the specified energy range justifying the need for as high tagging rates as possible 3 4 Maintenance 33 Due to the geometrical composition of the hodoscope the differences in the time of flight for electrons of different energies are small and can be precisely predicted as they all move approximately at the speed of light If the transit time in the scintillator and the PMT fluctu ates only by a small amount it is possible to calibrate the timing of the tagging system very accurately This turns the tagging sys
77. ed energy distribution and resolution without radiator a and b and with Cu 200 um radiator c and d The vertical lines denote the start of the vertical plane The horizontal lines denote the theoretical minimum resolution due to the energy width For details see the text 52 53 5 Final Design and Prototype Detector Using the arrangement of the scintillator bars described in the last chapter as a starting point the design of the mechanical construction for the vertical plane detector is presented in the following sections Since the design has to be tested the prototype is created only for nine channels The energy range of this prototype lies at the second step in the energy width see Figure 22 b This range is chosen to test the capability of high electron rates and to observe the influence of the step in the energy width For this part of the tagging system the Hamamatsu PMTs are chosen since these have very small dimensions paired with a high rate capability as described in Section 3 6 One desirable feature of the tagging system 1s easy maintenance see Section 3 4 It should be possible to replace single photomultiplier tubes without dismounting the complete detector and possibly affecting the alignment of the scintillator bars and thereby its energy calibration This is achieved by putting everything except for the PMTs into one big chassis In addition the scintillator bars are mounted in a way which makes it possible to remov
78. ed to the PMTs instead a piece of transparent silicone 1s used for the coupling see Section 5 4 The springs used here apply a force of F 15 N which is more than actually needed This value was chosen because no other suitable normative springs with the correct dimensions were available To make sure that no light goes beyond the PMT base an NBR O ring with a slightly higher diameter than the inside of the metal cylinder is placed between the spring and the base 5 2 Slides mounting clip light guide 10cm 9cm cm 7cm 6cm 5cm 4cm 3cm 2cm lcm Ocm scintillator light guide Figure 26 View of the back side of a slide isometric scale 1 2 Three consecutive scintillator bars are each placed onto a single slide Figure 26 which can be moved individually into the chassis Thereby all scintillator bars are accessible easily If they were mounted directly into the chassis only the first plane would be accessible directly That means to remove a scintillator bar in the middle or back plane several other scintillator bars would have to be removed individually The usage of the slides implies that scintillator bars from one slide must not touch scintillator bars from another slide during movement To ac complish this the slides must be assembled in a certain angle relative to the chassis see Figure 27 For each slide a maximum and minimum angle fulfilling this prerequisite is calculated As it turns out one single
79. electrons The blue line indicates the position of the focal plane which 1s calculated using the simulation 4 5 Calculation of the Detector Geometry 43 4 5 Calculation of the Detector Geometry As already mentioned before and shown in Figure 16 only a part of the complete detector can be placed into the focal plane Therefore the high energy range of the detector must be placed in a vertical plane Because the spatial distance between two electrons of energy E and E E becomes smaller when going to higher electron energies see Figure 14 the space which is available for the physical scintillation counters also becomes smaller This makes it difficult to place the scintillator bars in the vertical plane detector where the widths of the bars could become smaller than the diameter of the photomultiplier tubes The next two subsections cover this in more detail First the outline and the alignment of the scintillator bars relative to each other will be discussed then the composition of the complete detector will be presented 4 5 1 Alignment of the Scintillator Bars The width of the scintillator bars defines the energy resolution of the detector Up to a limit which arises from the beam flaw a smaller width leads to a better resolution see also Section 4 6 The easiest way to build the detector would be to use scintillator bars of a fixed width over a certain energy range Using a simple setup which either uses no overlap of neighbouring sc
80. es the threshold and causes a false entry in the spectrum By minimizing the cable length between the PMT and the splitter the time between the first and the second appearance of the signal can be reduced far below 30ns so that the reflection is simply overseen due to the finite double pulse resolution of the discriminator This structure arises when the DAQ opens the trigger just between the real hit and a second hit as in 2 or 3 The second hits defines the zero point and the real hit shows up at smaller times reflecting the structure of 2 and 3 The digital output signal of the discriminator has a length of 20ns The TDC always sees and records the same edge leading or trailing of this signal If the DAQ resets the trigger during this short time the FPGA sees an event and immediately causes a trigger which then can be displaced by up to 20ns to the real event time Since the trigger defines the zero point the real hit time is moved to an earlier time A hit during this time window of 20ns cannot release the trigger which would lead to the exponential structure as explained is Section 6 1 1 As the hit probability during this time 1s the same as after the trigger the height of the spectrum up to 20ns before the prompt peak is the same as after the prompt peak This artefact 1s caused by the ADC Every time the gate signal reaches the ADC it emits a small negative unwanted pulse on its signal inputs and a higher positive pulse w
81. esonanzzust nde des Protons mit linear polarisierten Photonen Ph D thes s Universit t Bonn 2007 Electron Strechter Accelerator ELSA Nov 2010 URL http www elsa physik uni bonn de Beschleuniger bilder elsaplan_en pdf ELSNER D Private communication 2010 ET Enterprises Ltd 9 B series data sheet 2009 EWALD R Untersuchung der X K Photoproduktion am Proton mit dem CBELSA TAPS Experiment Ph D thesis Universit t Bonn 2010 FORNET PONSE K Die Photonenmarkierungsanlage f r das Crystal Barrel TAPS Experiment an ELSA Ph D thesis Universit t Bonn Bonn 2009 FORNET PONSE K Private communication 2009 FROMMBERGER F Private communication 2010 GENTNER M Pr paration von Teilchenstrahlen f r Experimente der Hadronen physik langsame Extraktion an ELFE DESY und ELSA sowie Strahlkiihlung an HERA Ph D thesis Universitat Bonn 1999 GREEN D The Physics of Particle Detectors Cambridge University Press New York 2000 HRIVNACOVA I ET AL The Virtual Monte Carlo In Computing in High Energy and Nuclear Physics La Jolla California 2003 Hamamatsu Photonics K K R7400U series 2004 Hamamatsu Photonics K K Photomultiplier Tubes Basics and Application 3rd edition 2007 HAMMAN D In preparation Ph D thesis Universit t Bonn 2010 HANNAPEL J Private communication 2010 HILLERT W The Bonn Electron Stretcher Accelerator ELSA Past and future The European Physical Journal A
82. etection Efficiency of the Prototype 85 Landau Gauss 200 Landau Gauss 150 100 50 channel Figure 55 ADC spectrum with fitted functions Each fit includes an exponential function to describe the background noise Neither the Landau nor the Gauss function fit the observed spectrum For the convolution Y NDF 2 The shape of the signal peak in the ADC spectrum is expected to follow a Landau distri bution Kle05 Therefore an attempt to fit a Landau function to the ADC spectrum as shown in Figure 55 including an exponential function to match the background is made Obviously this does not work out which becomes evident when looking at the pedestal peak This peak corresponds to a single charge Q 0 and is expected to be only at x Q 0 Xpedestal SEE Equation 36 If the ADC spectrum for Q 0 is broadened it is obvious that all other charge deposits are broadened too This leads to the assumption that the signal s x appears to the ADC as the convolution s d x where d x has the shape of the pedestal peak Since it is easier to accomplish the distortion d x is approximated as a Gaussian d x ve 20 48 Thus the function to be fitted 1s f x ci s d x coe 3 ci s n d x 49 where n x is the undistorted background c nxd x co exp cax The convoluted func tion fits the observed spectrum quite well with y NDF 2 for the shown region For high x the spect
83. etector the BGO ball 1s made of 480 bismuth germanate BGO crystals It can detect charged and uncharged particles The forward spectrometer consists of different detectors for charged particles and the spectrometer magnet the OD open dipole It is used to measure the tracks and the momenta of charged particles emitted in forward direction The photons are produced in the tagging system using the high energetic electron beam of ELSA Figure 2 shows an overview of the electron accelera tor Unpolarised and polarised electrons are produced in the LINAC and LINAC2 respectively They are then accelerated in the booster synchroton and the subsequent stretcher ring to a max imum energy of Eo 3 5GeV The beam can then be extracted to the BGO OD or Crystal Barrel CB experiment Among different experiments studying similar questions two different tagging methods are used the Bremsstrahlung tagging and the Compton backscattering technique For the BGO OD experiment Bremsstrahlung tagging is used By shooting electrons onto a thin about 100 um radiator they are scattered and lose energy in the form of photons The energy of the photons can be inferred through the detection of the electrons in a magnetic spectrometer Table 1 shows an overview of different similar experiments their tagging method maximum photon BGO Bismuth germanate OD Open Dipole Deutsche ForschungsGemeinschaft German Research Foundation Bi4Ge3012 15 BGO O
84. etector during the BECOMES PLC 11 Measured TDC spectrum using channel 5 of the prototype detector during the BECOME a aang PA Boe amp ee A e A is a 78 Measured TDC spectrum using channel 5 of the prototype detector during the second Test detail a esa om ca EEE 78 Simple efficiency measurement CC mn 80 Possible trajectories of electrons in the detector 81 Effect of the dead time on coincidence counting 82 Exclusive coincidences of each combination of two channels 83 Exclusive coincidences of each combination of two channels and channel 35 83 ADC spectrum with fitted functions 0 000002 ea 85 Pulse distortion in the ADC and the discriminator 86 Spill structure of the electron beam 0 0002 eee 90 Scaler rate of channels 1 6 and 9 vs extracted electron current 91 Measurement of temporal distances 0 2 00000 eu 93 Scaler rate of channels 1 6 and 9 vs reconstructed rate from the TDC 94 Scaler rate of channel 9 vs scaler rate of channel l 95 Counting of coincidences and timing lll 96 Probability that the FPGA recognizes a coincidence 97 Different types of accidental coincidences rn 98 Comparison of simulated and measured spectrum 100 Deviation of the simulated data from the measured data CB 101 Deviation of the simulated data
85. ew distribution 34 where E are the real energies corresponding to the specific c channel E is the mean of all E In case of the ideal rectangular distribution Og 35 defines the smallest value which is possible for og The distributions and the resulting res olutions are simulated without a radiator and with the Cu 200 um radiator see Figure 24 In the left pictures the distribution of the differences between the detected and the real energies is shown for each c channel Using the differences instead of the real energies makes the dia eram clearer but does not change the standard deviation og which is shown in the right pic tures The values of Og E ea1 can be interpreted as the resolution the energy Erga is measured 4 6 Simulation of the Energy Resolution SI with The vertical lines in Figure 24 denote the position the hodoscope goes into the vertical plane The horizontal lines denote the theoretical minimum resolution og AE V12 This shows the strong influence of the placement out of the focal plane as the resolution clearly differs from the minimum value starting with the vertical plane The worst resolution without including the radiator is og 0 56 Eo 18MeV using a 200um Cu radiator the resolu tion becomes Og 0 63 Eo 20MeV This resolution is obtained for the highest electron energies where AE 1 56 Eo 50MeV The deviation from the theoretical minimum of Osomev 0 45 Eo 14 MeV thus is Omi
86. g are at the present state much higher 10 s MeV than the rates achieved with Compton backscattering 10 s MeV 2 With Compton backscattering is it easily possible to produce highly polarised photon beams When using linear or circularly polarised laser light the backscattered photon are also linear or circularly polarised The degree of polarisation can be up to 100 for the maximum photon energy The maximum polarisation is 1n principle only limited by the polarisation of the laser beam BAA 97 To produce polarized photons with a Bremsstrahlung tagging system coherent Brems strahlung is used Instead of an amorphous radiator like copper a crystal e g diamond GRenoble Anneau Acc l rateur Laser European Synchrotron Radiation Facility 2 4 Principle of Photon Tagging 23 has to be used and precisely aligned with respect to the beam direction EBB 09 For present experiments the maximum degree of polarisation that can be reached is about 80 3 The energy spectrum of Compton backscattered photons is rather flat compared to the dN dE E shape of the Bremsstrahlung spectrum By collimating the photon beam low energy photons can be removed resulting in a high energy photon beam For the BGO OD experiment the Bremsstrahlung method will be used This method proved to work fine for all other experiments which are were run at ELSA e g CB FP09a and SAPHIR Bur96 and provides the highest photon rates In orde
87. gnet Very low energetic electrons are deflected so strongly that they do not leave the magnet and cannot be detected At the high electron energy end the range is limited due to the primary beam The primary beam must not hit the hodoscope under any circumstances but has to fly into the beam dump If it hits parts of the detector it will illuminate the complete system due to the large amount of multiple scattering simply because of the huge intensity compared to the electrons which underwent Bremsstrahlung To maximise the range towards this end the mechanical construction should not exceed the detector channel which is closest to the primary beam Besides other factors e g the condition of the primary beam the energy resolution of the hodoscope is limited by the physical width of the scintillator bars The smaller the bars are the better is the spatial resolution and thus the energy resolution The energy width AE is defined as the span which is covered by one detector channel so that photons between Ey AE 2 cannot be distinguished For a beam energy of Ey 3200MeV an energy width between 20 MeV 0 696 Eo and 50MeV 1 5 E0 is targeted The actual resolution OE is different from AE and does not only depend on geometrical factors This 1s explained in more detail in Section 4 6 3 3 Rate Stability and Timing To provide enough statistics for the BGO OD experiment the tagging system has to be able to tag photons with a rate of at l
88. h combination of two channels The colour code and the number in each cell show the number of coincidences for the according combination of channels The numbers in the cells have to be multiplied by 1000 900 channel 800 700 600 500 400 300 200 100 1 2 3 4 5 6 7 8 9 10 channel Figure 54 Exclusive coincidences of each combination of two channels and channel 5 The colour code and the number in each cell show the number of coincidences for the according combination of channels The numbers in the cells have to be multiplied by 1000 84 Data Analysis channel coincident channels E Oz 2 1 3 0 9706 0 0002 3 2 4 0 9102 0 0003 4 3 5 0 9507 0 0002 5 4 6 0 9918 0 0001 6 5 7 0 9227 0 0003 7 6 8 0 9280 0 0002 8 7 9 0 9580 0 0002 Ss Table 7 Efficiencies calculated from the coincidences The results from using equations 43 and 44 are summarized in Table 7 The efficiencies for channels 2 and 5 are quite close to one the others show deviations of 5 and more As described in Section 7 1 1 this is at least expected for channels 3 4 6 and 7 due to geometrical considerations The observed values depend on the actual amount of background radiation This implies at least for channel 8 that either the discriminator throws away many signals or the efficiency is actually comparably low This will be checked in the next section 7 1 3 Correction for Discriminator Thresholds The efficiencies computed in the last secti
89. h constant a and variable b c resolution The bent red tracks are electrons with energies between 9 Ep 300MeV and 97 E 3100MeV with steps of 6 Eo 200MeV creased at 60 6 Eo 1940MeV to AE 0 9 96Eg 30MeV and at 74 4 Eg 2380MeV to AE 1 6 Eo 50MeV Figure 23 shows the changeover from AE gt 0 6 Eo 20MeV to AE 0 9 Ep 30MeV 4 6 Simulation of the Energy Resolution To investigate the influence of the beam flaw on the energy resolution the virtual detector 1s exposed to simulated electrons For this sensitive planes with the calculated shapes of the scintillator bars as shown in the last section are used for the simulated detector Now single electrons with a random energy following the Bremsstrahlung cross section are shot into the tagging system The information about the struck s channels and the energy of the incoming electron is stored for 10 events The gathered data set is then used for the next two steps Energy Calibration By analysing the struck s channels of each event the corresponding c channel is recon structed using the method described in Section 4 5 2 Since the energy of the incoming electron is known an energy distribution for each c channel can be created by doing this for all events In the ideal case 1 e when there is no beam flaw this distribution has a rectangular shape in fact it can be slightly curved since the Bremsstrahlung cross section increases for higher electron energies
90. he simulation The magnet itself the beam dump and the ground are only displayed for a better understanding they are not implemented as real matter but as vacuum Only the magnetic field and the two detector planes are considered in the simulation The magnetic field 1s implemented by an already existing plugin for Explora To model the beam properties an additional plugin had to be created This plugin can be configured to fit the particular needs by adjusting the following parameters startpoint defines the point at which the electron is created It 1s set to the position of the radiator phi theta define the direction of the electron in polar coordinates dphi dtheta define the beam divergence To apply theses parameters the vector d which de fines the electron s direction is first rotated about v d x e by an angle a into the x y plane e is the unit vector in z direction For this to work theta must be different from 0 by an amount which can yet be negligible d is then modified by adding random num bers to and 9 The random numbers follow a Gaussian distribution with o d phi and O dtheta respectively After this transformation d is rotated back by about v dx dy define the size of the beam spot in x and y direction n the number of electrons brems uniform Boolean values which define if the energy of the electrons is distributed ac cording the Bremsstrahlung cross section do dEy Ey or if it is uniform
91. he corresponding s channels are called s s5 The c channel c is defined by the overlap of the s channels s and s2 while there is no overlap with s3 The energy of c lies between E and 3 p The other extreme and probably the best solution would be to use a different width for each scintillator bar calculated to exactly match a constant energy width of the c channels The problems of this method arise from the manufacturing process First each scintillator had to be produced individually with a high precision Secondly one needs spare parts of the detector for a fast repair without waiting for the production of new parts This would imply that at least twice the number of scintillator bars had to be produced since each of them is different The solution chosen for the BGO OD tagging system lies in between As mentioned above a design using coincidences of two neighbouring detectors will be used Starting with the second example every scintillator bar would have another width if it exactly fits the same energy width for all energies The trick performed here is to match only the bottom side to low energies of each scintillator bar s to a fixed energy E y see Figure 18 The width of the scintillator bar has then to be at least large enough that also electrons with E E py AE hit it The energy of a c channel is now not only defined by the struck s channels but also by the next s channel which was not struck If e g an electron h
92. he event peak it will cut further into higher energies while lower energies will only be affected slightly since these are barely present below the threshold see also Figure 43 In total the number of events expected to be seen by the discriminator 1s reduced leading to an underestimation of Egisc Correcting the detector efficiency Having calculated s x and t x Equations 45 47 are now used to determine qisc assum ing d x 6 x t x e t x The results are summarized in Table 8 is the corrected detector efficiency pe 55 Edisc The value greater than one implies an underestimation of the discriminator efficiency Edisc as the probability to detect a particle clearly cannot be greater than 100 When gjsc 18 measured exactly the real efficiency Egetect 18 simply given by Edetect E 56 If Edisc is underestimated the only conclusion which can be made is ES detect lt E 57 The maximum observed uncorrected efficiency 18 Emax 0 992 the maximum corrected effi ciency is 4 1 005 These values show that the design of the detector in principle allows for a detection efficiency of 0 992 lt Egetect lt 1 000 58 Possible reasons for the smaller efficiency of the other channels can be the geometrical arrange ment Section 7 1 1 or flaws in the assembling of the detector like air in the glued joint between 88 Data Analysis the scintillator and the light guide see Section 2 5
93. hen the gate signal ends This pulses travel back through the delay cables and reach the discriminator at some point increasing or decreasing the signal height which lies on its input The negative pulse increases the chance that some noise or very weak signal is increased above the threshold leading to an excess of hits at this time Vice versa the positive pulse decreases the height of all signals at this time leading to a decreased count 72 Experimental Tests rate The long delay between the prompt peak and this structure can be explained by the way the gate signal travels to the discriminator The FPGA needs some time to generate the signal which then goes through a first delay to the ADC Between the ADC and the splitter there 1s another long delay leading to a delay of 240ns only due to the length of the used cables Including the delays of the discriminator and the FPGA a total delay of about 300ns seems realistic Nevertheless this effect has almost no practical consequence since it is separated by a large time from the prompt peak which is the only information that will be used in the later experiment The only effect is the possible increase or decrease of the entry in the corresponding scaler by 1 This 1s negligible at least for high rates as the time between two events is larger than 100us see Section 6 1 3 which implies a large number of hits in the scaler In the later experiment a different kind of ADC will be used without
94. ich holds the PMT and the voltage divider is called a socket assembly It has at least two connections the high voltage input and the signal output 2 5 3 Light Collection and Efficiency Because the shape of the scintillator generally differs from the shape of the PMT window they cannot be connected together directly Instead a light guide often made of PMMA is put between them If properly designed the light 1s totally reflected inside of the light guide with the result that the light is efficiently transferred from the scintillator to the PMT The critical angle for total reflection has to be kept in mind when designing the shape of such a light guide If a kink exist that has a smaller angle than 0 some photons will escape the light guide 7a nickel iron alloy with a very high magnetic permeability u gt 50000 I5Poly methyl methacrylate e g Plexiglas 2 5 Detector Components 29 The efficiency of a complete scintillation counter PMT and scintillator bar is determined by the number of electrons that finally reach the anode of the PMT The efficiency depends on different parameters of all three components This is illustrated in the following example The density of a plastic scintillator is roughly p 1 g cm For a scintillator thickness of x 0 5cm the mean energy deposit of a minimum ionising particle MIP is AE 2MeV g cm xp 1 MeV corresponding to 10 scintillation photons e 100eV Because the light is e
95. idence channels 7 and 8 144 Probability that the FPGA recognizes a coincidence channels 8 and 9 145 12 List of Figures 13 1 Introduction Measure what is measurable and make measurable what is not so Galileo Galilei 1564 1642 100 years ago in 1910 Thomson proposed his atomic model in which the atom consisted of an equally distributed mass and positive charge within which the electrons moved around as particles The charge of these electrons was shown to be opposite equal to the charge of a singly ionised atom The prior year 1909 Geiger and Marsden had determined that o particles impinging on a gold foil are scattered with angles larger than 90 In 1911 Rutherford showed that the observed rate of large angle scattering of particles is inconsistent with Thomson s model Instead the mass of the atom has to be concentrated in a pointlike hard nucleus leading to the cross section do sin 0 2 where is the scattering angle Only two years later in 1913 Bohr developed his model of the dynamics of the atom incorporating quantum theory Using this model it was possible to predict discrete excited electron energy states which were observed in the spectroscopy of hydrogen About 50 years later experiments done by Hofstadter showed that the cross section for the elastic scattering of electrons off gold is smaller than predicted for a pointlike nucleus This led to the introduction of a form factor into th
96. ides are glued on the scintillator bars THelmholtz Institut f r Strahlen und Kernphysik 5 4 The Complete Prototype Detector 37 2 The scintillator bars are mounted onto the slides a and b 3 The metal tubes of the PMT assemblies are mounted onto the chassis The slides are moved into the chassis considering the right order due the single wedge shaped slide and fixed by the screws on the back 4 To prevent crosstalk of light between different channels and to improve the light collection in the scintillator reflective plastic foil is wrapped around each scintillator bar 5 The complete device is covered with black plastic foil 6 The remaining parts of the PMT assemblies are put into the cylinders and locked by turn ing the bayonet cap a The coupling between the light guide and the PMT is achieved through a cookie a roughly 2mm thick round piece of transparent silicone b p 4cm 1 3cm 2cm E lem E Ocm Figure 29 Light guide isometric to scale The size is adapted to the width of the scintillator bar 58 Final Design and Prototype Detector 2a 3 5 light guide 6a Figure 30 Assembly of the prototype detector For description see the text 59 6 Experimental Tests In this chapter the experimental setup used for the prototype tests and the collected data will be presented The first test was performed in parallel to the Crystal Barrel C
97. ility pni that a pattern of n neighbouring s channels originates in a specific combination i of c channels can be calculated Instead of showing the complete derivation which is rather long but simple only the results are shown here They are summarized in table 5 For each number of neighbouring s channels which are hit at the same time within the time resolution the different possible origins are given For example three neighbouring s channels can be hit by a single electron Figure 20 a by two electrons of the same c channel or by two electrons from neighbouring c channels Figure 20 b In place of the complete expression for each probability pni the ratio of the probabilities for a single pattern of neighbouring s channels is given This makes it easy to read off the most probable origin Clearly an event with two or three neighbouring s channels belongs to a single electron with almost unit probability If the additional overlap p of the scintillator bars is not larger than approximately 0 6 four and five neighbouring hits correspond to the same combination of c channels This information can be used when reconstructing the c channels of real events as the energy of the electrons is not known a priori One interesting quantity is the probability that an event is indeed reconstructed correctly As t lt 1 this will be calculated only for the case that two electrons are reconstructed as sep arated by one intermediate c channel for
98. ime extendible and non extendible dead time If the dead time is extendible a second hit during this period will lead to an extension of the dead time by T beginning at the time of the second hit Photomultiplier tubes exhibit such an extendible dead time A second hit arriving before the signal has fallen down just adds up to the signal This way the signal height is increased but the second hit cannot be separated from the first one Components with a non extendible dead time are simply blind after one hit occurred A second hit within the dead time will have no effect This is the case for the discriminators used here which have a dead time of 7 30ns The effect of the dead time on the observed rates is given in Equations 59 and 60 for the non extendible and the extendible case respectively M 73 r n l rt r 59 hi 60 exp rT where n 1s the observed rate and r 1s the real rate If however a combined system 1s looked at the situation becomes more complicated In the present case an extendible dead time Tpmr the effect of the PMT is followed by a non extendible dead time Taise the effect of the discrimina tor This leads to the following behaviour M 73 r DEEP GEEK 61 r M es TPMT exp r TPMT n Of course this only holds as long as Tgisc gt Tpmr If the first dead time is longer than the second one the second one has no effect After determining n r in the next sections the attemp
99. ined in Section 4 4 its focal plane is almost parallel to the bottom side and lies closely below it That implies that the focal points for high energetic electrons lie within the beam dump or even beyond so that only a part of the tagging hodoscope can be placed into the focal plane The remaining part has to be located in front of the beam dump above the focal plane Electrons which lost only a small amount of energy during the Bremsstrahlung process will be very close to the primary beam at this distance to the magnet as both the scattered electrons and the primary beam are deflected by nearly the same angle tagging magnet beam dump goniometer 1270 quadrupole magnet Figure 11 Side view of the available space for the tagging system The electron beam enters from the left Distances are given in mm scale 1 50 Based on Wal10 32 Requirements of the BGO OD Tagging System 3 2 Energy Range and Resolution Ideally the tagged energy range should be as large as possible to cover a maximum photon energy range for a single energy of the primary beam It is still possible to deactivate single channels when a higher amount of high energetic photons is needed and the extracted electron current is increased The channels for the low energetic photons may then saturate due to the larger rate do dE E and therefore are not used in this case At the small electron energy end the range is limited by the dimensions of the ma
100. ing side there are two mounts and one lead through The mount for one scintillator bar consists of one notch in the inner side of the slide and a clip which is mounted on top of the scintillator to fix it The bottom of the notch has to be manufactured with a high precision as its position defines the energy of the corresponding c channel Using the same precision the rails on the side of the slide are made For the prototype a production tolerance of Ad 0 1 mm 0 2mm is used In the energy range of the prototype the energy width of AE 20 MeV corresponds to width of the scintillator bars of about d 5mm Therefore when a slide is removed and mounted again it can be realigned with an energy tolerance of AE Ad d 0 8 MeV As described in Chapter 2 5 2 the magnetic shielding of a phototube should exceed the photocathode by at least the radius of the shielding This way the shielding does not only cover the PMT but also a part of the light guide which therefore has to be elongated To keep the light guide still confined in the slide the side parts have to be made thick enough to provide for sufficient space 5 3 Chassis The chassis consists of three 2cm thick aluminium plates which are held together by screws Figure 28 The front top and bottom side are left free and will be covered by a black plastic foil On the inside the chassis has grooves which take the slides with the scintillator bars On the outside there are slots into which
101. intillator bars or a strict overlap of always exactly two scintillator bars Figure 17 would lead to a non constant energy width AZ for each c channel since the spatial distance between two electrons with a constant energy range the dispersion becomes smaller for higher electron energies This arrangement leads to two issues Due to the Bremsstrahlung cross section do dE Eo E E is the energy of the Bremsstrahlung electron the electron rate increases when the energy becomes larger The rate seen by a single scintillator bar increases even more due to the fact that the dispersion becomes smaller at the same time When using a small scintillator width which would work at high energies the total number of channels would be too large The second problem of this simple approach is that each c channel would have a different energy width AE making it more difficult to analyse the data which will be measured later Electron energy s channels c channels AE AE g AE Figure 17 Exemplary electron trajectories for equidistant energies distance E and scintil lator bars The c channels c are defined by the overlap of two scintillator bars 44 Detector Design Electron energy s channels c channels S1 2 3 S4 S5 Es Elib 4 E fs Flo t 30k te AE AF E3 p E1p 26 cu 2 A2 1 C AE Figure 18 Exemplary electron trajectories for equidistant energies and adjusted positions of the scintillator bars T
102. ints avoiding errors due to the coincidence time Afcoine and the dead time Afgeag The data sample 107 events taken with the lowest rate is used to avoid too high a loss due to dead time effects Because of the exclusive coincidence counting equation 42 has to be slightly modified N123 l gt 43 Nia Ni23 1 72 E N13 and N 23 now being the number of exclusive coincidences extracted from the data The statistical error Og on 2 is calculated assuming Poisson distributed numbers zu Sai 44 Figure 53 shows the number of coincidences for each combination of two channels As one would expect the coincidences between neighbouring channels clearly dominate Coincidences of two hits in the same channel do not appear since Afcoinc lt Afdead Similarly Figure 54 shows coincidences of three channels one of them being fixed to 5 Triple coincidences for the other channels can be found in Appendix B The coincidences of non neighbouring channels can originate in a detection efficiency for electrons which is smaller than one and background radiation E g neutrons are detected inherently with a small probability e 10 see Section 3 5 making is possible to penetrate multiple scintillator bars without depositing charge in the middle one 7 1 Detection Efficiency of the Prototype 83 N 1000 5000 channel 4000 3000 2000 1000 1 2 3 4 5 6 7 8 9 10 channel Figure 53 Exclusive coincidences of eac
103. is best This point could not be found for the second test which is probably due to the low statistics of the simulated data When increasing the number of simulated events the area of possible positions should be reduced This procedure can be used to measure the position of the detector with respect to the electron beam With this piece of information the detector can then be aligned precisely After doing this a complete energy calibration of the detector should be performed to ensure that the simulation reflects the actual reality This can not be checked by only minimizing the disagree ment of between the simulated and the measured data Figure 66 Deviation of the simulated data from the measured data CB o measures the size of the deviation for different orientations of the simulated detector For details see the text Figure 67 Deviation of the simulated data from the measured data BGO OD o measures the size of the deviation for different orientations of the simulated detector For details see the text 102 Data Analysis 103 8 Conclusion and Outlook In the previous chapters a complete design for a part of the vertical plane hodoscope has been described starting from scratch by defining the requirements of the new tagging system right up to the in beam test of a nine channel prototype detector The results are now summarised briefly by chapter and compared to the requirements of Chapter 3 Finally an outlook will be
104. itating the real behaviour of independent electrons entering the detector Nevents 10 individual events were simulated and stored in the shown histogram Clearly the number of hits for t 0 trigger 18 equal to Nevents 10 This peak can be broadened for a detector with multiple channels and is called prompt peak After the trigger signal arrives hits continuously occur with the same rate leading to a mean entry of Nevents 1 Albin 10 The hit distribution prior to the prompt peak is a bit more complicated but can be derived as follows At one random point in time fo the electronics becomes ready and the trigger is reset The first 37Tp principle also the trailing edge or both can be measured 38in contrast to a single hit TDC which can only do one measurement per channel until it is reset 32the TDC is simulated by a short ROOT script mimicking the trigger matching mode 64 Experimental Tests hit occurring after this point will then release the trigger The time distance between this hit at trieger and to follows again an exponential distribution P tuigger fo dt ne tiger 10 37 The probability for a hit occurring between t lt and t lt dt lt to is P t dt ndr 38 To observe a hit with a distance Ar before the prompt peak on the left side trigger fo has to be smaller than At This leads to At P At dt P trigger to lt At P t lt dt n dile ndr n 1 e dr 39 0 perfectly mat
105. itative approach will be used see e g Gre00 more details in Jac06 2 2 1 Energy Distribution Instead of viewing the electrons as incident on some material they will be considered at rest while the nuclei of the target material are considered to be moving with high velocity in the direction of the electrons The electromagnetic field of the moving nuclei can be handled as a distribution of low energy photons given by the Weizs cker Williams distribution Jac06 me 1 as in all following calculations dNy k 2a 1 1 n 1 123 F A dk x B2k k 3 5 The nuclei have Z protons Since the photons are soft their phase does not change significantly within the size of the nuclei Therefore the amplitudes for each proton can be added coherently leading to factor of Z for the total cross section The cross section for the scattering of a single soft photon off the electron is the Thomson cross section ST Or 06 6 3 The Bremsstrahlung cross section is then the product of the photon distribution and the Thom son cross section do pi dk 7 k dk T 9 16_ 3dk 2 1 123 EB B e a 8 OE al k gt 8 The quantum mechanical approach in the Born approximation uses the Feynman diagrams of Figure 4 It results in the following for the cross section differential in the photon energy extreme relativistic case Eo E k gt gt 1 KM59 2E Eo 1 l 9 EN DE kei eo Ey 3 Eo do szo
106. ith two independent channels a gt Passive pulse spliter aa ch d padane a ui eue de de ii de de 34 Simulated TDC spectrum of an ideal detector with one channel 35 View of the electronics setup used for the first test 36 Timing of the different signals 37 View of the framework in front of the CB tagging system 38 Top view of the CB tagging system 2 aon aoa a 39 Measured ADC spectrum using channel 5 of the prototype detector during the HESUICSE x en rd ee gic ee ete Beck te Bebe Sion de ee He By OE OA Un B UR A 40 Measured TDC spectrum using channel 5 of the prototype detector during the MISPLESE Aeree ini a od ca a ee E AN 10 41 42 43 44 45 46 47 48 49 50 51 22 53 54 55 56 57 58 59 60 6l 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 ay 78 19 50 81 82 53 List of Figures Measured TDC spectrum using channel 5 of the prototype detector during the TUS LCSW WCE tado ogni i n 71 Measured ADC spectrum using channel 5 of the prototype detector with entry in DEC SPEC ei acia e eee a D ee ec 73 Threshold curve for channel 5 of the prototype detector 73 View of the prototype detector mounted in the BGO OD area 74 Overview of the location for the BGO OD tagging system and the electronics rack 75 Photograph of the secondary electron beam taken with a Polaroid film 76 Measured ADC spectrum using channel 5 of the prototype d
107. its s and s2 but not s3 the energy of this electron lies between 2 y and Ez p If also s3 is hit the energy lies between 3 y and 4 Using this technique leads to a bit more complicated assignment of the s channels to the c channels but with the benefit of a constant energy width AE OE So far one problem remains When all scintillator bars have the same width more than two or three scintillator bars will overlap for high energies see Figure 18 To circumvent this the width of the scintillator bars is adjusted roughly to match AE Multiple consecutive scintillator bars have the same width until the overlap becomes too large Then a new group of scintillator bars with a smaller profile is used This is accomplished by increasing the width in steps of 1 mm To make sure that each electron hits at least two s channels a minimum spatial overlap of 10 between three scintillator bars 1s used Using this method one benefits of the advantages of both ways On the one hand multiple scintillators of the same size can be used simplifying the manufacturing on the other hand 4 5 Calculation of the Detector Geometry 45 S1 SI 3 4 Figure 19 Exemplary electron trajectories for equidistant energies and adjusted positions and widths of the scintillator bars The corresponding s channels are called s s4 s and s2 have the same width as well as s3 and su a constant energy width can be maintained For now the layout of the arr
108. ke iE Figure 61 Cap of the PMT assembly 124 Appendix EN Anzahl 10 Material Acetal ee schwarz OT o al l E LL Tem I _ hamamatsu_inlay2 i I mm Hi AA al A la meer rote Tagger eG a 1 eE Anzahl 10 Material Acetal Kunststoff schwarz AO OS AO O AS O ETS L LL Frere 30 03 2010 see A RA sr Hamamatsu Inlay Lie AAA e Prototyp 1 alici BUCHE frane ee ee M L IL Figure 83 Part 2 of the cable lead through A Technical Drawings 125 3 4 DURCH R98 Anzahl 20 e Material Aluminium 6061 NN Anzahl 9 SS Material PMMA ASTE Teaser I n 777 MR PRI 17 Lichtleiter Be Tagger Prototyp 1 YO stats Ancerungen com name PT Figure 85 Light guide 126 Appendix Anzahl 9 Material Szintillator A Dt ENSEM __ Feces 30 03 2010 sibke sei s1 Szintillator 1 a 4 Tagger Prototyp 1 eee ee O II Figure 86 Scintillator bar A Technical Drawings 127 Framework for the CB Tagging System See Section 6 2 1 for a description of this construction Figure 87 Framework used to mount the prototype detector behind the CB tagging system All distances are quoted in mm 128 Appendix B Triple Coincidences See Section 7 1 for a description of these graphs channel
109. krecht zur Fokalebene angeordnet Dies limitiert die durch die Granularit t des Hodoskops beschrankte Energieaufl sung der Photonenmarkierung weiter Bedingt durch die geringer werdende Dispersion muss dar ber hinaus an zwei Stellen in der vertikalen Ebene die Energieaufl sung verschlechtert werden Mit der Simulation dieser Detektoranordnung wird der Einfluss der Platzierung auBerhalb der Fokalabene untersucht Der im Rahmen der Arbeit aufgebaute Prototyp umfasst neun Kan le aus dem Vertikalteil des Hodoskops im Bereich eines Sprungs der Aufl sung Dieser Bereich wurde gew hlt da sich hier die mechanische Konstruktion am schwierigsten darstellt Weiterhin erm glicht die Wahl des Bereiches hoher Elektronenenergien eine berpr fung der Ratenfestigkeit des Detektors die wesentlich f r das BGO OD Experiment ist Die mechanische Konstruktion des Prototypen erlaubt es einzelne Photomultiplier und Szintillatorstreifen auszutauschen ohne dabei die Ener giekalibration des Hodoskops zu beeinflussen Der Prototyp wurde w hrend zweier Tests hinter den Tagging Magneten des CB Experiments und des BGO OD Experiments untersucht Da bei wurde gezeigt dass eine Detektionseffizienz von 99 und mehr erreicht werden kann und eine Rate von 50MHZz hochgerechnet auf den gesamten Detektor ohne signifikante Verluste m glich ist Des Weiteren wurde die Funktion eines FPGA Moduls getestet das Koinzidenzen zwischen benachbarten Szintillatorstreifen erkennt und d
110. l The efficiency of this intermediate channel is then calculated as 123 E 42 113 n13 and n123 being the count rates of coincidences in the particular channels If the electrons are leaving the source in a straight line as illustrated and background radiation 1s negligible also the efficiencies of channel 1 and 3 can be measured in this way All electrons inducing a signal on a certain channel i traverse necessarily also the other two channels j and k The efficiency for channel j is then n njj amp being computed analogously However this method cannot be simply transferred to the prototype detector for several reasons 1 The detector was not designed to measure efficiencies 1 e the arrangement of the scintil lator bars differs from the one shown in Figure 50 If the amount of background radiation see Section 3 5 could be neglected the efficiency could still be measured for all chan nels except for the first and the last one Figure 51 a shows a sketch of the detector geometry as well as two exemplary electron trajectories coming from the Bremsstrahlung radiator It is clear that within the assumption of all electrons coming in this way the efficiency 3 of channel 3 can be obtained Although it is not sandwiched between two other detectors all electrons penetrating channel 4 and 2 go through channel 3 Only for channels 1 and 9 no suitable coincidences can be made When taking background radiation into ac
111. l and vertical wires are used to measure the profile of the electron beam By moving them through the beam and measuring the rate of Bremsstrahlung electrons the beam structure can be inferred With the aid of the luminescent Chromox screen the electron beam can be directly observed In the centre hole the diamond will be mounted Tused for beam scans Chromox screen for an optical inspection of the beam 2 4 Principle of Photon Tagging 25 The Tagging Magnet The scattered electrons are vertically deflected in the magnetic field of the tagging magnet The BGO OD experiment uses a magnet identical to the one used in the CB experiment It is a dipole magnet from Brown Bovery Switzerland type MC It can be operated with currents up to 1500 A corresponding to a maximum field value of B 2 0T For each beam energy the current in the magnet is adjusted in a way that the primary electron beam is always deflected by the same angle to enter the beam dump For BGO OD this angle is Otag 7 8 for CB this angle is Qtag 9 0 FPO9a The effect of a constant magnetic field on relativistic particles is given by dv B ON X 22 7 vav E 22 This expression depends only on the ratio B E As long as the magnetic field increases linearly with the current the required current is proportional to the energy of the primary electron beam For currents of up to 800 A the deviation from the linear behaviour is smaller than 1 FP09a Th
112. l setup first data measured during this test will be presented 6 2 1 Assembly of the Test Stand With respect to the BGO OD tagging magnet the CB tagging magnet is rotated by 90 around the beam axis Therefore it is a horizontal bend device This opens the possibility to construct a frame located independently behind the CB tagging system which is capable of holding the prototype see Figure 37 The framework is made of aluminium profiles partially from RK ROSE KRIEGER It can be used to mount the prototype or other detectors at different posi tions behind the CB tagging system and in this way expose them to different electron rates for technical drawings see Appendix A An additional option is the positioning above or under neath the plane of Bremsstrahlung 68 Experimental Tests Figure 38 shows an overview of the complete setup The high voltage and the power supply is taken from the BGO OD area y GI LA Figure 37 View of the framework in front of the CB tagging system 6 2 2 Detector Settings During the test all photomultiplier tubes were operated at the same nominal voltage of 800 V and the same discriminator threshold value of 30 Using this threshold the complete signal peak could be observed in the ADC spectra The detector was mounted at two different positions one of them corresponding to the highest possible rate which could be reached within the spatial restrictions Since these positions were different
113. lchen k nnen im Vorw rtsspektrometer gemessen werden dessen zentrale Komponente ein offener Dipolmagnet ist Dieser erm glicht die Bestimmung von Ladung und Impuls der Zerfallsprodukte Zur Erzeugung hochenergetischen Photonen wird der aus ELSA extrahierte Elektronenstrahl auf einen Radiator z B aus Kupfer gelenkt wobei manche der Elektronen Energie in Form von Bremsstrahlung verlieren Uber die Messung der Elektronen energie in einem speziellen Magnetspektrometer wird indirekt die Energie der Photonen be stimmt Die Kombination aus Radiator Magnet und dem Hodoskop das die Elektronen im Spektrometer ortsaufgel st nachweist hei t Photonenmarkierungsanlage Tagg ng System Thema dieser Arbeit war die Konzeption des Hodoskops sowie die Konstruktion und der experimentelle Test eines Prototyps Realisiert wurde das Hodoskop mit berlappenden Szintil latorstreifen ausgelesen durch Photomultiplier Die Grundlage f r den Entwurf bildete eine Si mulation zur Vorhersage der Bahnen der im Radiator gestreuten Elektronen im Magnetfeld Mit hilfe dieser Simulation ist es m glich die Fokalebene des Magneten zu bestimmen Im Idealfall wird ein Detektor in dieser Ebene installiert da dort die Energiebestimmung der Elektronen unabh ngig vom Eintrittswinkel in den Magneten ist Aufgrund der r umlichen Gegebenheiten kann allerdings nur ein Teil des Hodoskops in der Fokalebene platziert werden Der andere Teil wird stattdessen vertikal ann hernd sen
114. lf but it can access the GEANT4 and the VMC Monte Carlo engines VMC Virtual Monte Carlo HAB 03 is an approach to make it easy to use different simulation engines without changing the complete source code At the moment it supports GEANT3 CER93 and GEANT4 AAA 03 and provides a unified interface for defining particles detector layout and all other parameters To swap the Monte Carlo engine one has only to change one parameter The implementation makes strong use of the ROOT framework BR97 which simplifies the sharing of complex data between the VMC configuration and other programs using ROOT All programs which are used in this section are written in C ROOT or BASH script For all steps of the design in this section ROOT scripts were coded These can be used to repeat the calculations and simulations with different parameters e g a different energy width Only very few things have to be done manually 4 2 General Remarks To simplify the subsequent considerations some general definitions and remarks are made in the following paragraphs Coordinate System Independent of the orientation of the tagging magnet which 1s different for CB and BGO OD the z axis 1s defined by the direction of the incoming electron beam The x axis lies in the plane in which the electrons get deflected by the tagging magnet perpendicular to the z axis The y axis 1s chosen accordingly to get a right handed coordinate system see figure 13 a
115. ll distances corresponding to a large momen tum transfer Q the quarks inside the nucleons are quasi free since A lt lt 1 This behaviour is called asymptotic freedom In this region the interaction of quarks is well understood and described within perturbative QCD the gauge theory of the colour interaction For distances about the size of the nucleons small Q a gt 1 the quarks are confined making it impossible to describe the excitation spectra of the nucleons within perturbative QCD Various models have been developed to describe the excitation spectra Not all questions have been answered E g the models predict that the number of predicted excited states 1s much larger than the number of the observed states 14 Introduction BGO y l y I om Sing System Figure I Overview of the BGO Open Dipole experiment The shown tagging detector belongs to the old SAPHIR tagging system Based on Wal 10 To further examine the excitation spectra of the nucleons the BGO OD experiment Fig ure 1 is currently set up at the electron stretcher accelerator ELSA in Bonn It is funded by the DFG within the Transregional Collaborative Research Centre 16 Subnuclear Structure of Matter To excite the nucleons real photons of an energy of up to about 3 GeV are shot onto a liquid hydrogen or deuterium target The decay products of the excited states are detected in a spectrometer almost covering 47 of solid angle The central d
116. llows s first This first could just as well be part of another coincidence s belonging to the neighbouring s channel s and the s channel s next to it black line To see if the FPGA generates the according coincidence signals and how long it takes to do so the time span Af c from the second hit to the next hit in the corresponding c channel c 1s measured figure 62 b The coincidence is implemented on the FPGA simply as the logical AND between two neighbouring s channels As the signal which is output by the discriminator has a width of Si Si aa ars Sk i a pc b Figure 62 Counting of coincidences a and timing b The black dots represent hits on an s channel s a The two dots connected by the black line belong to the same electron the other two dots belong to another electron For details see the text b Ar s is the time span between two hits on neighbouring s channels Af c is the time span from the second hit to the coincidence signal 7 3 FPGA Coincidence Matching 97 At S 30 20 10 0 10 20 30 40 50 60 70 At max Csg Figure 63 Probability that the FPGA recognizes a coincidence in dependence of the tempo ral distance channels 5 and 6 the other channels can be found in appendix D The red area indicates an accurate detection of coincidences For the explanation of the different areas see text about 20ns all hits with a smaller distance should be recognized as coincidence by the FPG
117. ly distributed interval defines the energy range of the simulated electrons If neither brems nor uniform 1s set the energy for all electrons is set to the lower edge of the interval 4 4 Focal Plane 41 ingoming e y direction tagging magnet field map E 3 D O sensitive planes Figure 14 Overview of the setting for the simulation The bent red tracks represent electrons with energies between 6 Ep 200MeV and 100 Eo 3200MeV with steps of 9 Eo 300MeV The coloured area on the right picture shows the measured magnetic field red 1 6T for Ey 3200 MeV purple OT To simulate an electron moving in z direction phi 0 and theta 1 x 107 5 is chosen With these settings d phi has the meaning of a horizontal divergence whereas dtheta has the meaning of a vertical divergence 4 4 Focal Plane As explained in Chapter 2 4 2 the tagging magnet focusses electrons of the same energy into a small spot To calculate the focal point for a single electron energy E an electron beam of energy E with the properties given in Table 4 no radiator 1s simulated This test beam consists of n electrons e whose angle and parallel offset is distributed according to the beam properties and a single centre electron e without beam flaw see Figure 15 For each electron ej the distance d l to the centre electron track x Xo c pe is calculated depending on le The distance is measured perpendicular to x as
118. mitted isotropically only a fraction of the photons are emitted in a direction that is totally reflected This fraction is AQ 27 90 6 1 i d d sin 1 sin 0 2 26 AT 0 0 2 for plastic with 6 39 Further photons are lost in the light guide if the cross section of the scintillator bar A is bigger than the cross section of the area A which is coupled to the PMT Then at most A A photons are transmitted Leo94 For a scintillator width of 2cm the thick ness is 0 5 cm and a diameter of the photo cathode of 8 mm the ratio A A is approximately 0 5 and about 10 0 2 0 5 1000 photons will reach the PMT With a mean quantum efficiency of 10 about 100 electrons will be released in the photocathode This number fluctuates statistically but the probability that none or only a few electrons are produced 1s close to zero Hence in most cases a detectable electric signal will be generated implying an efficiency of the scintillation counter close to 100 In this example the loss of light in the coupling between light guide and scintillator and PMT respectively was neglected For wavelengths larger than 350nm the transmission of differ ent cyanoacrylate glues and silicone is close to 100 so in most cases no light is lost Leb02 More photons can however be lost if the emission spectrum of the scintillator and the trans mission spectrum of light guide and the window of the PMT do not match up Furthermore
119. mplete detector a tagging system is obtained which fulfils the experimental needs completely and which in addition is easy to maintain 4Front End Discriminator 4 Analog Fanout and Amplifier 106 Conclusion and Outlook References 107 References AAAT03 AGOSTINELLI S ET AL GEANT4 a simulation toolkit Nuclear Instruments and Methods in Physics Research Section A Accelerators Spectrometers Detectors and Associated Equipment 506 3 250 303 2003 BAA 97 BOCQUET J P ET AL GRAAL a polarized y ray beam at ESRF Nuclear Physics A 622 1 2 c124 c129 1997 Bal10 BALLING A In preparation Ph D thesis Universit t Bonn 2010 BCD 90 BABUSCI D ET AL Polarized gamma ray beams by inverse compton scattering Progress in Particle and Nuclear Physics 24 119 139 1990 Bel10 BELLA A In preparation Diploma thesis Universitat Bonn 2010 BLP71 BERESTETSKI V ET AL Relativistic Quantum Theory Pergamon Press Oxford 1971 BR97 BRUN R AND RADEMAKERS F ROOT An object oriented data analysis framework Nuclear Instruments and Methods in Physics Research Section A Accelerators Spectrometers Detectors and Associated Equipment 389 1 2 81 86 1997 Brul0 BRUDVIK J The Near Threshold Pion Photoproduction Program at MAX lab In Svenskt K rnfysikerm te XXIX Orebro Sweden 2010 Bur96 BURGWINKEL R Aufbau und Test und Eichung des hochaufl senden Tagging Systems TOPAS IT am Bonner
120. n 0 19 Eo 6MeV This number constitutes the best resolution which could be achieved with arbitrarily small scintillator bars The best resolu tion for both scenarios is og 0 19 E 6MeV for the lowest electron energies In this chapter the arrangement for all scintillator bars in the focal plane as well as in the vertical plane was computed Starting from the measured magnetic field map the electron trajectories were simulated and the focal plane was calculated Using this information the scintillator bars could be placed Finally the resolution which is expected for this design was calculated For all simulations the 3200 MeV field map was used without the scaling for the smaller current in the BGO OD tagging magnet Since all steps in this chapter can be repeated fully automated using the created ROOT programs the calculations for the final design of the tagging system can be redone using the reduced magnetic field Also the mechanical design which is made in the next section can be adapted easily to the final layout without a large effort 1800 1600 1400 16 1200 14 1000 12 800 600 400 200 0 MeV N E det E rea MeV pn a b OD A DA oco 500 1000 1500 2000 2500 500 1000 1500 2000 2500 E MeV E MeV rea N 1800 1600 1400 1200 1000 800 600 400 200 0 MeV Detector Design 8 6 d 4 2 0 500 1000 1500 2000 2500 500 1000 1500 2000 2500 E MeV E MeV rea Figure 24 Simulat
121. n line above the spot originates from syn chrotron radiation of the electrons which are deflected in the magnetic field For this measurement the thresholds and the high voltage were roughly adjusted using the method shown in Section 6 3 After taking first data the ADC spectrum for all channels was compared to the same spectrum while requiring a corresponding entry in the TDC The threshold was then adjusted to fit between the pedestal peak and the signal peak To make this possible the high voltage had to be increased for two channels The resulting values are summarized in Table 6 For further investigations on the rate stability an additional test was performed While the detector was mounted at the topmost position the rate of electrons leaving the accelerator was varied in a wide range from below 1 MHz to up to about 15 MHz per channel This way measurements for many different rates were acquired without changing the detector position The results of this test will be discussed in Section 7 2 76 Experimental Tests 12cm llem 10cm Ocm Figure 46 Photograph of the secondary electron beam taken with a Polaroid film to scale For details see the text 6 4 3 First Experimental Data of the Test at the BGO OD Experiment Figure 47 shows an example of the ADC spectrum of one of the channels of the prototype detector With respect to the previous test see Figure 39 the pedestal peak is broadened and exhibits a spiky stru
122. nnel This is in contrast to the ex pectation that this point lies at about the same rate for each channel This behaviour can be understood at least qualitatively When a higher current is extracted from ELSA the beam position may be shifted Gen99 If the beam is shifted upwards with increasing current the distance between the tagging hodoscope and the electron beam becomes larger implying that the electron energies seen by a fixed detector channel decrease Because of the Bremsstrahlung cross section do dEy Ey dEe Eo Ee this also leads to a decrease of the rate in the detector explaining the observed discrepancy in Figure 58 The correlation between beam shift and current changes with the exact adjustments of ELSA As these were changed during the data taking to allow for higher currents no exact prediction can be made about the electron rate hitting the detector nor can the maximum possible rate be determined 7 2 5 Scaler versus TDC The second method makes use of the exponential distribution of the temporal distances of hits in one detector channel The shape of this distribution is independent of the dead time T since this only cuts out small time spans while leaving the remain unaffected When calculating distances from the times in the TDC two things have to be considered 1 Only a part of the complete time span measured by the TDC is suitable for this Imme diately before the prompt peak short distances are suppressed
123. nnels and channel 9 C Rates C Rates See Section 7 2 for a description of these graphs 133 134 n MHZz channel 1 n MHz channel 2 n MHz channel 3 Appendix 100 200 300 400 500 600 700 800 2900 1000 I pA Figure 97 Scaler rate vs current in ELSA channel 1 3 nIMHz channel 4 n MHz channel 5 n MHz channel 6 C Rates 100 200 300 400 500 600 700 800 2900 1000 I pA Figure 98 Scaler rate vs current in ELSA channel 4 6 135 136 nIMHz channel 7 n MHz channel 8 n MHz channel 9 Appendix 100 200 300 400 500 600 700 800 2900 1000 I pA Figure 99 Scaler rate vs current in ELSA channel 7 9 C Rates 137 10 nIMHz Ht 10 n MHz channel 2 10 n MHz channel 3 r MHz Figure 100 Scaler rate vs reconstructed rate from the TDC channels 1 3 138 Appendix I 10 2 8 6 E 3 5 2 0 N I n D Ka O I 10 2 S 8 6 NO 4 D 3 S 2 0 0 2 4 6 8 10 12 Figure 101 Scaler rate vs reconstructed rate from the TDC channels 4 6 C Rates nIMHz channel 7 n MHz channel 8 n MHz channel 9 r MHz Figure 102 Scaler rate vs reconstructed rate from the TDC channels 7 9 139 140 Appendix nIMHz channel 7 n MHz channel 8 n MHz channel 9 Figure 103 Scaler rate vs scaler rate from the lowest channel channels 7 9 D FPGA Coinciden
124. ntillator bars were simply placed one above the other they would physically overlap even without the enlarged width discussed before For this reason they have to be staggered in more than one vertical plane A higher number of planes allows for a higher number of s channels in the same energy range because the spatial distance between the single channels decreases thus leading to a better energy resolution On the other hand as the distance in flight direction also increases multiple scattering in the first struck scintillator bar can change the electrons directions and increases the possibility that they hit a scintillator bar which does not lie in the original direction For the alignment of all scintillator bars in the focal plane as well as in the vertical plane a ROOT program was created which automates this process First it places as many scintillator bars as possible into the focal plane When the distance to the beam dump gets too small the program starts to build the vertical plane After placing the first scintillator bar it is checked if there is enough space to place the second bar directly above it Figure 21 a the dotted bar If this is does not work the second scintillator bar is placed in a second vertical plane behind the first one closer to the beam dump Now the same check is done for the third scintillator bar If there is enough space it is placed above the first one otherwise it is placed in a third vertical pl
125. ojected onto a plane Xo radiation length for copper Xy 1 42cm amp Fine structure constant a 1 137 2 2 Bremsstrahlung The process which is responsible for the emission of photons when electrons travel through material is called Bremsstrahlung When an electron of momentum po traverses the Coulomb field of a nucleus there is a certain chance for it to be scattered leading to the radiation of a photon of momentum k see Figure 3 The nucleus is needed to take the recoil momentum q Otherwise this process would be kinematically impossible due to momentum and energy conservation Only the incoherent Bremsstrahlung will be discussed here In the coherent Bremsstrahlung process the electrons are scattered in a crystal The recoil momentum is then absorbed by the lattice just as in the M bauer effect see e g Sie76 The process of coherent 18 Basics of the Underlying Physical Processes Po p Figure 3 Kinematics of the Bremsstrahlung process The incoming electron 1s scattered in the electric field of the nucleus During the scattering process a Bremsstrahlung photon is emitted Bremsstrahlung strongly depends on the orientation of the momentum transfer q with respect to the reciprocal lattice of the crystal This technique can be used to produce linear polarised photons for more details see e g EBB 09 Tim69 Bel10 It is not useful to derive the complete quantum mechanical cross section here A more qual
126. on differ apart from geometrical and background effects significantly from one which is not expected for an electron detector It is highly unlikely that electrons do not lose energy when penetrating a scintillator bar see Section 2 5 3 However it is possible that the thresholds of the discriminator are set so high that signals which belong to a real hit are not being detected In this section the attempt is made to estimate the fraction of the event signals which do not go above the threshold called discriminator efficiency Edisc To do this two things are needed The threshold curves of the discriminator t x for each detector channel computed as in Section 6 3 and the expected energy spectrum s x for electron signals in the ADC t x can be interpreted as the probability that a signal corresponding to the signal height x in the ADC is counted by the discriminator s x is the number of entries in ADC channel x which is expected for pulses originating in electrons hitting the corresponding scintillator bar The real spectrum always includes some noise n x which has to be removed before doing further calculations The total number of electrons can then be calculated as Xmax Meotal y s x 45 X JXmin The total number of particles expected to be registered by the discriminator 1s Xmax Nas Y s x t x 46 X Xmin If these quantities are known the discriminator efficiency is computed as follows Edisc 4T 7 1 D
127. oton rates of up to 50 MHz which also implies a total rate of events which is too large to store each event For this reason it has to be decided which events should be stored and which should be rejected This is the purpose of the trigger logic Only when a defined set of conditions is fulfilled a trigger signal is released lead ing to a readout These conditions are chosen to select hadronic events via their decay signature One condition will be coincident hits of neighbouring channels in the tagging system Other conditions can e g be a minimum energy deposit and a defined number of clusters in the BGO ball or a number of tracks in the forward spectrometer The coincidence in the tagging system is chosen because only events with a known photon energy can be analysed If the scattered Bremsstrahlung electron is not detected or the reaction was not caused by a Bremsstrahlung photon at all this parameter is unknown Using several trigger conditions the events to be stored are partly preselected and the resulting readout rate is strongly reduced After a trigger in this chapter s channels are simply referred to as channels i e the combination of one scintillator bar and one PMT 6 1 Electronics Setup and Data Acquisition 61 is released the trigger logic pauses until the data readout is finished and the electronics is ready again For these tests of the tagging system each single hit in one channel releases a trigger sig nal Thi
128. out For instance two background particles 82 Data Analysis Aldead Aldead Si e Sf 9 Si Sj gt S ae Sk SS Sk gt e L L AT A I a Alcoine Atclean Alcoinc Atclean a b Figure 52 Effect of the dead time on coincidence counting The dots denote hits in the neigh bouring channels s x The black dots are detected hits The grey dot is not detected due to the prior hit on the same channel within the dead time hitting channels 4 and 6 without hitting channel 5 also go through other channels with a high probability With the introduced constraint such events are filtered out Another aspect to be regarded is the dead time of the discriminator Afgeag 30ns If prior to a coincident hit one of the detector channels was already struck it is possible that it is still dead and cannot observe the important electron see figure 52 a This would lead to a reduced count rate of double and triple coincidences the latter suffering from the higher relative loss This can be avoided by demanding that during a time Afgean 40 ns gt Afdead before a coincidence no channel was busy To allow for an easier calculation the same is applied after the coincidence see Figure 52 b Now the number of coincidences can be extracted from the TDC data by scanning each event and searching for patterns which match the previously mentioned constra
129. part In the next section the software tools which are needed for the computation of the focal plane and other calculations are presented Section 4 1 After introducing some general defi nitions Section 4 2 the setup of the employed simulation is explained Section 4 3 After the calculation of the focal plane in Section 4 4 the focus is laid on the design of the upper part of the focal plane detector which is the most challenging problem For this area it 1s especially important to provide an easy maintenance because the probability for a failure increases with the electron rate impinging on the detector For the remaining parts of the detector the design can probably be adapted and simplified 4 1 Software Tools A lot of the calculations in this chapter rely on simulations to predict electron trajectories in the magnetic field of the tagging magnet For all the simulations the Explora package SAA10 is used The software provides functions to analyse experimental and simulated data and is able to simulate physical scenarios The instructions for Explora are written in XML files One can access the various functions by using different XML tags Each XML tag is based on one 21 Extensible Markup Language 38 Detector Design class in the source code of Explora which is written in C This design makes it easy to extend Explora by additional modules see Section 4 3 The underlying simulation is not provided by Explora itse
130. pectrum with a corresponding entry in the TDC spectrum leads to a distribution as seen in Figure 42 By dividing the number of entries with an entry in the TDC spectrum by the total number of entries for each channel in the ADC one gets the probability of a signal with a certain energy getting through the discriminator see Figure 43 Ideally this threshold curve s zero below the set threshold value and one above As can be seen this is not the case here instead the step is broadened Reasons for this may be the behaviour of the discriminator itself or an imprecise measurement of the ADC In the present case there is a certain chance that very low energetic signals are discriminated but also that real event signals are not counted To avoid this the high voltage of the PMT can be adjusted to increase the gap between pedestal peak and signal peak The artefacts seen in Figure 41 are not observed here because they are correlated to a real signal so that their entry in the ADC lies within the signal peak Nevertheless by adjusting the thresholds some of these artefacts can be avoided see Section 6 4 3 6 4 Test at the BGO OD Experiment 73 350 300 250 200 150 100 50 N 1000 to 120 140 160 180 200 channel Figure 42 Measured ADC spectrum using channel 5 of the prototype detector demanding a corresponding entry in the TDC spectrum 107 events 350 81 lt 300 E 08 250 0 6 200 150 0 4 100 0 2 00 120
131. prototype detector using nine Hamamatsu R7400U PMTs was designed and built The design is extendible for the complete segment of hodoscope which uses this PMT A focus was laid on easy maintenance of the hodoscope in terms of the replacement of single PMTs and scintillator bars The experimental tests showed that it 1s in fact possible to replace single PMTs within several minutes The replacement of the scintillator bars 1s possible without affecting the energy calibration of the complete setup Chapter 6 Experimental Tests A first test at the Crystal Barrel experiment was performed to ensure the correct func tioning of the test setup including the electronics The second dedicated test at the BGO OD experiment was then used to collect various data During the second test the position of the prototype was changed as well as the electron current extracted from ELSA Chapter 7 Data Analysis The data from the experimental tests could be used to examine several important properties of the tagging hodoscope 104 1 2 3 4 Conclusion and Outlook By including effects which arise due to the discriminator it was shown that a detecting efficiency between 99 2 and e 100 0 can be achieved This coincides with the expectation of Section 2 5 3 An analysis of the electron rates in single scintillator bars showed that rates of up n 4 MHz are possible for a single channel without significant losses Assuming a coverag
132. proximation can be expected to deviate from the exact behaviour by a small amount lt 1 15 For extreme relativistic energies the screening of the field of the nucleus by the electrons of the atomic shell has to be taken into account Using the atomic form factor Fg 2 on 52 Par 16 Ze qr where p r is the electron charge distribution the cross section formulas 9 and 11 can be cor rected by simply multiplying do by 1 F le Using a Thomas Fermi model for the atom the amount of screening can be expressed in terms of y defined as 100k y E 17 EyEZ3 This number is close to the ratio of the radius of the atom ra 1 az 3 and the maximum impact parameter which for relativistic energies is rmax Jn Po p k 2E0E k If the maximum impact parameter is much larger than the radius of the atom Y 0 the charge of the nucleus is completely screened If it it close to the radius of the nucleus y gt 0 the complete charge Ze is seen by the electron Assuming an incident electron energy of Ey 3200MeV and 5 Ep lt k E lt 95 Eo it follows that 3 x 1074 lt y lt 0 01 corresponding to almost complete screening In this case the cross section may be approximated by KM59 2 E 2 E amp E PE a 18327 Bess 18 3 Ey n l 18 dk do 4Z7 a A le 9 Ep 2 3 Multiple Scattering The main process responsible for deflections of incident electrons is multiple
133. r to switch to Compton backscattering the acceleration facility would have to be modified which would raise the expenses by an unacceptable amount 2 4 2 Elements of a Bremsstrahlung Tagging System The complete tagging system consists of three distinct parts the radiator the tagging magnet and the tagging hodoscope A schematic of such a tagging system 1s shown in figure 7 The primary electron beam enters from the left and hits the radiator Some electrons will undergo Bremsstrahlung and lose a varying amount of energy which depends on the cross section see Section 2 2 The scattered electrons as well as the remaining primary beam are then deflected by the tagging magnet into the tagging hodoscope and the beam dump respectively Usually the tagging magnet is simply a dipole magnet The beam dump does not belong directly to the tagging system but 1s needed to stop the primary beam For more information on the beam dump see e g Els07 radiator tagging magnet SA NB N Re ET Figure 7 General scheme of a Bremsstrahlung tagging system For a description see the text from this point when referring to tagging system it is always meant a Bremsstrahlung tagging system 24 Basics of the Underlying Physical Processes Cu 100um B al Cu 200 pmi NU c Kapton 125 um a b Figure 8 The Goniometer a and the different radiators b The bottom and the middle stage move perpendicular to th
134. rausfordernden Themas sowie die Unterst tzung w hrend der Arbeit bedanke ich mich bei Herrn Prof Dr Schmieden Besonderer Dank geb hrt Herrn Dr Elsner f r seine Betreuung w hrend des letzten Jahres sowie f r seine gewissenhafte Korrektur der Diplomarbeit Danken m chte ich Dieter Walther f r seine Hilfe bei der Erstel lung der technischen Zeichnungen sowie der feinmechanischen Werkstatt des HISKP f r die Realisierung der mechanischen Bauteile Au erdem m chte ich mich bei Dr Russell Johnstone Dr Claire Chapin meinen Eltern und meiner Freundin Gesine f r das Korrekturlesen bedanken Zu guter Letzt bedanke ich mich bei der ganzen Arbeitsgruppe f r das gute Arbeitsklima sowie f r die hilfreichen Diskussionen 112 Danksagung 113 Appendix A Technical Drawings Prototype Detector Technical drawings of all parts of prototype detector Left and right are as seen when looking from the front while the detector is mounted at the BGO OD experiment All distances are quoted in mm See Section 5 for a description M5x0 8 6H 1 3 4 DURCH a DIN 74 86 3 X 90 0 Anzahl 1 i Material Aluminium 6061 E gl Gehause Ruecken 3 i fo i a Tagger Prototyp 1 a Figure 69 Back plane of the chass s Appendix 114 EV epu dAjojoug 1ebbe ae E DEE O EEE IS _ _D ua quiy asnaeuar 726 TTT omm pr AAA HE 1909 UNIUIUN V e1194ey yezu
135. riment Figure 39 shows an example of the ADC spectrum of one of the photomultiplier tubes The broad peak originates from electron hits in the scintillator The energy distribution in this peaks reflects the varying energy loss described by the Landau distribution During most events this particular channel does not detect a signal because all nine channels can provoke a trigger This leads to a high pedestal peak which is clearly separated from the signal peak Furthermore some background is visible between the two peaks In the ideal case the count rate should be zero in this region The ADC spectrum will be analysed in more detail in Section 7 1 70 Experimental Tests 350 300 250 200 150 100 50 N 1000 too 10 710 160 180 200 channel Figure 39 Measured ADC spectrum using channel 5 of the prototype detector during the first test 107 events The dotted line is suppressed by a factor of 10 Z 10 10 10 10 10 10 15000 1500 1000 500 0 500 1000 t ns Figure 40 Measured TDC spectrum using channel 5 of the prototype detector during the first test 107 events The y axis is logarithmic Figures 40 and 41 show the TDC spectrum of one channel of the detector In comparison to the expected form Figure 34 there appear additional structures The different features in Figure 41 are explained as follows 1 The prompt peak is broadened on the right side later times and has a sharp edge to earlier
136. rum is overestimated due to the Gaussian approximation as the pedestal peak falls faster to zero than the Gaussian 86 Data Analysis Figure 56 Pulse distortion in the ADC and the discriminator s x signal generated by the PMT d x signal distortion common to ADC and discriminator d x additional distortion of the ADC t x threshold curve of the discriminator The terms on the lines indicate the current signal shape Until now it was assumed that both the discriminator and the ADC see the same distor tion The ADC however can introduce another distortion or uncertainty d x due to the charge measurement The complete signal seen by the ADC is then SApc x s d d x 50 This does not change the overall shape of the signal since the Gaussian distribution 1s invariant under a convolution The signal seen by the discriminator is still sgise x s d x When se lecting only those hits lying above the discriminator threshold f x has to be applied to sqisc x Yet when looking at this spectrum in the ADC it is distorted by d x ne x s d 1 d 2 51 The complete path of the signal from the PMT to the ADC is summarized in Figure 56 Now two cases can be considered depending on the impact of d x d x negligible d x 6 x The complete spectrum s x n x is seen in the same way by the ADC and the discrim inator By dividing the spectrum with hits above the threshold by the complete spectr
137. s arising from the dead time and the signal length In summary the generation of the coincidence signal was shown to work reliably During 10 events the FPGA did not lose a single coincidence event as long as the time span between the single hits was short enough 7 4 Comparison of Simulated and Measured Spectra An interesting test is the comparison between simulated data and really measured data To make this possible the position of the detector prototype has been measured relative to the tagging magnet for both experimental tests This piece of information was then used to generate simulated spectra as in Section 4 6 The general procedure of the comparison will be first explained for the test at the CB site and then extended for the second test 7 4 1 Test at the CB Site For the test at the CB site the measurement of the position and the angle a between detector and magnet was not very precise That is why the simulation was made for different angles For one angle close to the measured angle the agreement between simulation and measured data should be maximal The simulated position was not changed for this test since the impact of a 7 4 Comparison of Simulated and Measured Spectra 99 small displacement is expected to be very small The detector was mounted at a position corre sponding to low electron energies so that the rate 1s approximately constant over the dimension of the prototype This implies that the spectrum looks th
138. s minimum bias condition also called tagger OR a logical or of all tagging channels is chosen to provide more options for the data analysis as will be shown in Section 7 The trigger logic is implemented on an FPGA board see below Discriminator The purpose of a discriminator is to detect event signals and to distinguish them from electronic noise In the simplest case this is achieved by only accepting signals which reach an adjustable threshold When the discriminator detects such a signal it generates a logical pulse Nothing happens if the signal is below the threshold To assure that this method works properly the signal has to be clearly larger than the noise typically a few mV Then the threshold can be set between the noise and the signal level When a discriminator detects a suitable signal a second signal coming shortly after will be ignored The shortest time without rejecting the second one is called dead time If the dead time is too short one long pulse could activate the discriminator multiple times In case it 1s too long the maximum count rate will be significantly reduced The discriminator used here features 32 channels NIM inputs and LVDS outputs The dead time is 30ns Analogue to Digital Converter An Analogue to Digital Converter ADC converts an analogue signal into a digital signal In this case an LRS 2249A 12 channel ADC with a relative resolution of 0 1 is used LeC74 It integrates the current in
139. same edge of the signal there can be a jitter between their times due to the signal length 98 Data Analysis Si e9 Si _ ___ _ _ _ Sj eoo Sj eoe Spa Sk Cij e a Cij O a b Figure 64 Different types of accidental coincidences For a description see the text The green area originates from incorrectly reconstructed coincidences Assuming a real coincidence of channels s and s a later hit on s will also be regarded as coincident with the hit on s If the later hit originates from an electron out of the Bremsstrahlung target there are two possibilities e s was hit a second time too Figure 64 a Then there is a hit on c at around this time e s was not hit Instead the other neighbour sy of s was hit figure 64 b There is no hit on cj at this time Since the overlap between two scintillator bars is more than 50 the probability for at least one of the neighbouring channels being hit too is bigger than 50 With the same probability a hit on c will be found at the time of the second hit on s The gap between the red and the green area arises due to the dead time of the discriminator It 1s washed out as the distance between two hits of one coincidence is not constant The purple area belongs to accidental coincidences of electrons or background radiation The diagonal cut in the lower left can be observed due to timing constraint
140. signal arrives The delay of At 150ns was chosen for practical reasons The cable for the gate signal was prolonged accordingly to match the right timing The correct timing was checked using a oscilloscope with a pulsed signal as input as shown in Figure 36 The pulse clearly arrives at la GE VME 7807RC 66 Experimental Tests Figure 35 View of the electronics setup used for the first test The components are the same for both tests the ADC green line while the gate blue line 1s open The delays between the input and the ADC discriminator are reduced by equal amounts with respect to the setup on Figure 31 The gate signal jitters within 10ns because it starts and ends exactly at a clock cycle of the FPGA which runs at 100 MHz 6 1 3 Readout and Data Acquisition To start taking experimental data one remotely connects to the VME computer and starts the DAQ Data AcQuisition which is coded by D Hamman Ham 10 This program first initializes the TDC and the ADC and programs the FPGA Then it starts to record events until a predefined maximum number is reached One event cycle consists of three phases 1 Waiting until a trigger occurs 2 Reading out the electronics ADC TDC scalers This takes more than 100us so that events occurring within this time are only counted by the scalers The long 6 2 Test at the Crystal Barrel Experiment 67 Tek Run input je trigger TDC 2 gate ADC gt input ADC 3 HE RET CES RIDI
141. t This implies a big change of the rate when moving the detector up or down Hence this time the spectrum is simulated for different angles amp and for different vertical dis placements Ax of the prototype relative to the measured position Due to this second dimension and the limited amount of time the granularity as well as the number of simulated events had to be decreased for this task amp is varied between 4 0 and 7 0 with steps of 0 5 Ax between 7 0cm and 7 0cm with steps of 0 5cm 4000 events are simulated per point The resulting values for o are shown in Figure 67 Because of the low statistics no single minimum can be seen here However it seems probable that Ax 0 and a gt 2 0 which is realistic as the position could be measured with a higher accuracy than the incline 100 Data Analysis E 2 gt 1 5 1 0 5 0 12 3 4 5 6 7 8 9 channel a g 2 g 2 lt 1 5 3 1 5 1 1 0 5 0 5 0 0 12 3 4 5 6 7 8 9 12 3 4 5 6 7 8 9 channel channel b c g 2 g 2 gt z 1 5 3 1 5 1 1 0 5 0 5 VS 5 6 7 800 d LI GS wp 20 39 channel channel d e Figure 65 Measured spectrum a Simulated spectrum for 21 5 b and 24 0 d Ratio of real and measured spectrum for 21 5 c and 24 0 e 7 4 Comparison of Simulated and Measured Spectra 101 7 4 3 The Usefulness of this Comparison The first test shows that there s a single position where the agreement between simulation and measurement
142. t is made to extract the dead time by using Equations 59 to 61 7 2 Electron Rate Stability 89 7 2 2 Measurement Principle For the following analysis data were taken at different rates between 0 5 MHz and 12 MHz per channel The rate observed directly with the prototype detector can be extracted from the scalers see Section 6 1 1 To obtain the real rate three different methods are used 1 During the experiment the electron current leaving ELSA and entering the experimental hall was measured with a high frequency resonator Pus10 Sch09 and monitored by the slow control As this current is proportional to the rate of electrons hitting the tagging system it can be used as measure of the rate of Bremsstrahlung electrons The absolute value of the rate however 1s unknown 2 The second possibility is to further investigate the TDC spectrum Since the distances between two hits follow an exponential distribution the shape of the distribution of tem poral distances can be used to extract the rate This rate is indeed the real rate as the only effect of the dead time is the non existence of small distances Larger temporal distances are not affected 3 One can take advantage of the fact that the different channels of the detector are exposed to different rates Since low rates are effected less by the dead time than high rates the rate of the lowest channel in electron energy can be used as an estimate of the real rate apart from a
143. tem into a suitable candidate for the timing reference of the complete experiment since the time when the Bremsstrahlung photon hits the target 1s calculated easily 3 4 Maintenance The different components of the tagging system undergo an ageing process when they are irra diated After some years of operation scintillators lose their ability to produce light and PMTs lose gain and become defective If this happens a certain energy range will not be tagged and a significant part of the total rate will be lost Hence it is important to assure a reliable opera tion of the tagging system and to exchange the particular part This should be possible without changing the position of any other part and destroying the energy calibration FP09a of large parts of the complete system Therefore the hodoscope is designed in a way that enables to eas ily exchange single photomultiplier tubes and scintillator bars easily without disturbing other parts 3 5 Background The function of the tagging hodoscope is to detect the scattered electrons from the Brems strahlung process in the radiator but not any other particles like for example neutrons which are backscattered from the beam dump Els10b and electrons which are scattered in the beam pipe To see the influence of backscattered neutrons their detection efficiency is estimated Since neutrons are not charged their detection requires a hadronic interaction mainly elastic scattering off protons
144. the perfect beam will be summarized as beam flaw Since also the angles for multiple scattering are approximated Gaussian see equation 19 the resulting angular divergence is given by the quadratic sum of the beam divergence before hitting the radiator and the multiple scattering RMS angle The beam properties calculated for the different radiators are shown in Table 4 radiator 0 mm 0 mm ot mrad o mrad Oys mrad no radiator 1 0 1 5 0 08 0 30 0 0 Cu 50 um 1 0 1 5 0 26 0 39 0 25 Cu 100 um 1 0 1 5 0 37 0 47 0 36 Cu 200 um 1 0 1 5 0 51 0 58 0 50 Table 4 Beam spot size and angular divergence Channel The term channel is used in two different meanings On the one hand a channel corre sponds to a single scintillator bar and PMT On the other hand a channel refers to a temporal coincidence of two or three neighbouring scintillator bars corresponding to a single electron energy To avoid misunderstandings the former will be called s channel single the latter will be called c channel coincident 4 3 Simulation of the Magnetic Field of the Tagging Magnet To calculate the focal plane and to site the different s channels the electron trajectories for different energies have to be known These can be calculated using a simulation of the tagging magnet The simulation requires the measured field components of the used dipole magnet Such a measurement was done for the CB tagging magnet Bal10 for different primary beam energies
145. times This is due to the effect that not only this specific channel can cause a trigger If one electron hits two overlapping scintillator bars one of them must be the first one This one starts the trigger and will always be at the same time in the TDC spectrum it defines the zero point except for a constant delay Due to different cable lengths or a fluctuating transit time in the PMT the second hit can then be displaced by a small time leading to a broadening towards later times 2 An additional peak appears 30ns after the prompt peak which is the dead time of the discriminator This is caused by the signal of the PMT sometimes showing high spikes 6 2 Test at the Crystal Barrel Experiment 71 200 100 O 100 200 300 400 500 600 t ns Figure 41 Measured TDC spectrum using channel 5 of the prototype detector during the first test 107 events detail The y axis 1s logarithmic 1 prompt peak 2 afterpulses of the PMT 3 reflections 4 5 trigger artefacts 6 ADC signal For details see the text 3 4 5 6 after this time When these are above the threshold they lead to a second hit By adjusting the just roughly set threshold levels this effect can be avoided A second narrow peak appears about 50ns after the prompt peak It is caused by a reflec tion of the signal pulse at the splitter which could be verified by varying the cable length between PMT and splitter This reflection sometimes reach
146. to be shifted to the data analysis 94 Data Analysis 10 lt 8 6 Gi d Ss E A gl S O 0 10 8 6 o 4 t e c3 a 5 0 N T lt ON t c3 Ka O r MHz Figure 60 Scaler rate of channels 1 6 and 9 vs reconstructed rate from the TDC The other channels can be found in Appendix C Shown are the statistical errors The line 1s defined by Lt 7 2 Electron Rate Stability 95 7 2 6 Scaler versus Scaler The third method is to approximate the real rate r by cjnj the rate n of the lowest channel which is much smaller than the rates of the highest channels times a normalisation factor c Firstly this is because of the dE E run of the cross section and secondly the width of the scintillator bars jumps from 11mm to 17mm in the middle of the detector To assure that the decrease of the rate for one channel can be observed while the rate of the lowest is only slightly affected only channels 7 9 will be used for this comparison with channel 1 since these are exposed to the highest rate Figure 61 shows the scaler rate ng of channel 9 against the scaler rate n of channel I There is no significant loss of rate for up to 4MHz in channel 9 At the same time this justifies the approximation since for ng 4 0 MHz n r 1 3 MHz and no deviation from the linear behaviour has to be expected n MHz channel 9 Figure 61 Scaler rate of channel 9 vs scaler rate of channel 1 The other channels
147. trons which do not come from adjacent c channels To decide which c channels have to be reconstructed from a certain number of adjacent s channels the probabilities are estimated that a certain reconstruction is valid At the same time this shows how many electrons are expected to be lost due to false reconstruction For this estimation the additional overlap in addition to the half overlap divided by the width d of the scintillator bar is assumed to be the same for all bars p l d const The other parameter is the probability t that in addition to the first electron a second electron comes in one specific c channel e g in the same c channel as the first electron The estimation works as follows The total rate of electrons between 63 E 2000MeV and 94 Eo 3000 MeV is assumed to be about 10 MHz and constant for all energies neglecting the real shape of the Bremsstrahlung cross section The temporal resolution of the electronics is estimated by 10ns which is rather pessimistic This leads to a mean of 0 1 electrons per time span of 10ns called event throughout this section Therefore the probabilities for one and two electrons in one event given by a Poisson distribution P N 0 1 N exp 0 1 are 0 090 and 0 005 re 4 5 Calculation of the Detector Geometry 47 spectively Assuming an energy width of AE 1 6 50MeV according to 20 c channels t is given by t 0 005 0 090 20 0 003 Using these parameters the probab
148. tter timing However the substantial larger costs of the BC 418 did not justify this small benefit equivalent to EJ 204 and NE 104 0 equivalent to EJ 228 and Pilot U 3 6 Selected PMTs and Scintillator 35 quantity R7400U 9111SB outline diameter mm 15 9 26 5 length mm w o connector 11 5 43 spectral response range nm 300 650 280 630 peak wavelength nm 420 350 photocathode material bialkali bialkali quantum efficiency at peak n a 28 active diameter mm 8 22 window material borosilicate glass borosilicate glass dynodes structure metal channel circular focussed number of stages 8 10 maximum ratings anode to cathode voltage V 1000 1500 average anode current mA 0 1 0 1 typical nominal characteristics voltage V 800 800 gain 7x 10 7x 10 anode sensitivity Alm 50 50 timing rise time ns 0 78 1 8 transit time ns 5 4 15 transit time spread ns 0 23 1 2 Table 2 Properties of the Hamamatsu R7400U and the ET Enterprises 9111SB PMT Ham04 ET 09 quantity BC 404 general base polyvinyltoluene density gcm 1 032 refractive index 1 58 scintillation properties light output anthracene 68 rise time ns 0 7 decay time ns 1 8 pulse width fwhm ns 22 wavelength of max emission nm 408 light attenuation length cm 140 Table 3 Properties of the Saint Gobain BC 404 plastic scintillator Sa105 36 Requirements of the BGO OD Tagging System 37 4 Detector Design The design of the
149. type was not tested This depends on the scintillator mate rial but for the prototype only old scintillator material could be used By further testing with the new scintillator see Section 3 6 the timing resolution of the final detector can be investigated To finish the complete tagging system some building blocks are still missing 1 The mechanical layout for the lower part of the vertical plane and for the focal plane has to be designed Probably the existing design can be adapted to the larger photomultiplier tubes to be used for the lower part of the vertical plane Instead of fixing the black foil with adhesive tape a frame construction should be used which is screwed onto the detector As in the focal plane the spatial distance between the channels is bigger the construction of this part of the hodoscope should be more simple than the other parts 8 2 Conclusion 105 Figure 68 FrED board prototype The PMT signal enters on the left connector An amplified analogue signal as well as the digital signal is output on the right 2 3 4 Only a basic functional check was made for the PMTs which will be used for the lower part of the vertical plane as well as for the focal plane Before building the remaining parts of the detector their behaviour at high rates and their efficiency should be investigated The electronics which was used for the experimental test is not final Rather a new electronic design is under w
150. um one gets the threshold curve f x Sapc x 4 hApc x i s n d x t x 52 The spectrum originating from real hits sapc x s d x is found by fitting f x to the complete ADC spectrum The exponential term is left out to separate the signal s d x from the noise n d x Replacing s x by s d x in Equations 45 and 46 to include the distortion gives the discriminator efficiency min 5 d X ER X Xmin u SRA NA X Xmin Edisc 53 Tel Detection Efficiency of the Prototype channel coincident channels Og Edisc Oei e Os 2 1 3 0 9706 0 0002 0 9653 0 0001 1 0055 0 0003 3 2 4 0 9102 0 0003 0 9763 0 0001 0 9323 0 0003 4 3 5 0 9507 0 0002 0 9942 0 0000 0 9563 0 0003 5 4 6 0 9918 0 0001 0 9935 0 0000 0 9984 0 0001 6 5 7 0 9227 0 0003 0 9873 0 0001 0 9346 0 0003 7 6 8 0 9280 0 0002 0 9931 0 0000 0 9344 0 0002 8 7 9 0 9580 0 0002 0 9597 0 0001 0 9982 0 0002 Ss 87 Table 8 Discriminator efficiencies uncorrected and corrected detector efficiencies d x not negligible If however d x significantly distorts the signal seen by the ADC it not possible to get the real threshold curve f x sS x n SADC X nApc x N a E t x AR s n d d x 2 When using 1 x instead of t x to calculate amp gisc the value will differ from the real value The effect can be understood qualitatively Due to d x the threshold curve will seem broader than it actually is If the curve 1s near t
151. umbers can vary The given periods are valid for this test 90 Data Analysis x10 300 250 200 150 100 50 02 04 06 08 10 12 14 16 t s Figure 57 Spill structure of the electron beam N is the accumulated number of entries in one scaler during Afpin 0 1 s 2 The spills themselves show a structure Only spills with an almost constant rate may be selected The rising and falling edge should be left out Keeping this in mind suitable spills are defined with the following constraints 1 A spill begins when the rate measured in the scaler goes above 50kHz and ends when the rate falls below this value 2 Only spills with a length between 3s and 5s are used 3 The leading edge is cropped by 1 5s to remove the bump at the beginning of a spill The trailing edge is cropped by 0 5s 4 The rate averaged over each 0 5s interval must not deviate by more than 20 from the mean rate of the spill These values were chosen to minimize the error of the rate determination and to maintain a sufficient number of spills The rate for each channel and spill is then obtained by summing the scaler entries and dividing by the sum of the entries in the 1 MHz scaler which is the time reference 7 2 4 Scaler versus Primary Electron Current The measurement of the electron current which leaves the stretcher ring and enters the BGO OD experiment is completely independent from the data collection of the tagging prototype On one hand
152. ure 20 a The same pattern of s channels can be produced by two coincidental electrons Figure 20 b which would not be the case without the larger overlap Furthermore the triple overlap leads to patterns of hit s channels which cannot be asso ciated reliably to the correct c channels even when it is evident that more than one electron is detected see figure 20 c and d Four s channels are hit implying that more than one elec tron hit the detector For two electrons there are two possible origins one electron in c channel 4 and one in c channel 3 or one electron in c channel 4 and one in c channel 2 C channel 4 is correctly identified in both situations A similar situation occurs when five neighbouring s channels are hit With six or more s channels the situation is no longer ambiguous seven or more s channels are not possible using only two electrons 46 Detector Design S1 52 3 S4 Ss SE S1 2 3 S4 Ss SE Figure 20 Possibilities for multiple electron events The scintillator bars are all pictured with the same size to simplify the graphic The dashed lines represent electron tracks The c channels c to cs are defined by the dotted lines Struck scintillator bars are coloured grey a A single electron hits three s channels b The same pattern as in a can be produced by electrons of adjacent c channels c Two electrons from adjacent c channels can hit four s channels d The same pattern can be produced by two elec
153. y L00 T 1 99 L0 0 F80 05 L0 0 FOIIE co co E W o Qo TO L 09 IN ann a oO oO 0500 F 0057 500 Fo097 000 F 08 S Figure 70 Left side plane of the chassis 115 A Technical Drawings W EV epu dAjojoug 1ebbe DEE HERE HERE BEE NAAA OCT usoa asnaeyan 6 TEST TTT ome pres Ap TT 1909 UNIUIUN V JEIISLEN yezuy L0 0 FLY 99 L0 0 F80 0S i L0 0 FO9LE 00 0 200 07 0500 FSZ 050 0 5 W IN Z 09 AN U Figure 71 Right side plane of the chassis 116 Appendix Material Aluminium 6061 h16 Einschub hinten 2s Anzahl 1 N ce o H m e o 1 29 0 02 VO a gt O S o 0 02 0 S G 9 Pat Name orm W 31 58 0 02 71 00 x 0 02 Figure 72 Left side of the middle slide Tagger Prototyp 1 AN 117 A Technical Drawings EY BEE a aces BEE dAjojoug 1ebbe WIN sj suJOA qny3sulg g A pep pp pw roue q qoo cu HA 1909 UNIUIUN V JEIISLEN yezuy W c0 0 FLEO Hl AN IN Figure 73 Right side of the middle slide Appendix 118 W O po p pog AREA V sj U94UIY QNYISUID Uu pep T V sen used 1909 UNIUIWN V JEIJOLEN juBzuv g 9p J N H NN gt EA gt T N UJ d d Figure 74 Left side of the top slide 119 A Technical Drawings W
154. y coincidences of two scintillator bars are counted For neutrons the detection efficiency 1s much lower and the probability that one neutron induces a signal in two scintillator bars is even smaller namely 1 96 as estimated above 3 6 Selected PMTs and Scintillator Two different photomultiplier tubes are used for the tagging system the Hamamatsu R7400U and the ET Enterprises 9111SB Table 2 shows their most important properties The R7400U was chosen because of its fast response its compatibility for high rates FPO9b and its small dimensions It features a transit time delay of tg 5 4ns with a spread of Oty 0 23ns Thus it 1s used for the low photon energy part where rates are the highest and the scintillator bars are the smallest The 9111SB has a slightly bigger outline and larger transit time delay of tg 15ns with a spread of Oty 1 2ns It is therefore used for the higher photon energies and lower rates Both PMTs are sufficiently insensitive to magnetic fields the R7400U due its dynode structure and the 9111SB due to a shielding with Mumetal As socket assembly the Hamamatsu E5780 and the ET Enterprises E673 ASN2 are used Both are designed for use with a negative high voltage As scintillator the Saint Gobain BC 404 see table 3 was chosen It has a short rise time of 0 7ns comparable to the rise time of the R7400U and thus allows for a very fast counting Another possible choice was the BC 418 offering an even be

Design of the BGO-OD Tagging System and Test of a Detector

Contents

Download Pdf Manuals

Related Search

Related Contents