Completion of the Data Acquisition Computer Upgrade Project
Contents
1. Nor were we making changes in two different places that then had to be reconciled e Multiple releases A release technology was created that allowed us to make formal releases By having multiple releases on the system we could engineer in radical changes without affecting current operation In summary there has been a very low rate of problems caused by computer change At the same time the com puter group has felt free to service custom needs as well as make progress in whatever direction was required 6 The Future While first semester 1989 has seen the completion of the original data acquisition and computer upgrade project it should not be thought of as a static situation There are not only new features being requested but there are new instru ments on the horizon What we have created is a foundation and technology that not only supports our past and current endeavors but our future plans as well Of course these projects will continue to be in addition to the daily challenge iring data ot acquiring data Jon Brewster 3 00E9 YJ 4 aua aO om 3CR208 1 2 5069 2 components A QSO mg 19 3 z 1 02 B Seyfert mr 20 8 z 0 156 2 00E9 1 50E9 j d A B 3 9 arcsec 1 00E9 i li oll 4 Le Reducteur Focal du CFHT Modes Longue Fente et PUMA Le r ducteur focal du CFHT utilis jusqu alors en mode multi spectroscopie PUMA a t dot d un mode longue fente la fin 88 et utilis avec2. le CCD PHX1 de 7 de bruit de lecture en Juin 89 A cette occasion nous rappelons les principales caract ristiques de cet instrument sp cialis pour la spectroscopie d objets faibles tout en offrant une capacit d imagerie Une description compl te de l instrument et de son fonctionnement est donn e dans le User s manual for long slit and multi spectroscopy with the focal reducer and PUMA disponible sur demande La combinaison optique comprends les diff rents tages suivants Plan focal F 8 longue fente ou masque PUMA peu vent y tre ins r s lentille de champ collimateur Roue filtres 9 75 mm quatre positions Roue Grisms 9 65 mm quatre positions 2 filtres 2 Grisms e Objectif F 2 9 CCD ll est donc possible de s lectionner une configuration imagerie ou spectroscopie Le tableau donne l chelle spa tiale et les dispersions mesur es en A pixel pour les 3 CCDs utilisables cette table corrige celle du manuel d utilisation distribu avant le 1 Juin 89 Table 1 Grism B400 A 5200 Grism V150 a 6100 Grism 0600 a 5900 arcsec pixel Il est possible d ins rer dans le plan focal une fente ou un masque multi objets PUMA 1 La proc dure ins rer retirer la fente ou le masque s effectue manuellement sur l instru ment Le support de fente peut recevoir des fentes de 0 5 2 5 avec une longueur de 45 Un mouvement fin en rotation permet un aligne ment pr
3. FH Canadian Applications Committee CFHT c o Director Herzberg Institute of Astrophysics National Research Council Canada Ottawa Ontario CANADA K1A OR6 DEADLINES Postmark date For time in first semester August 15 For time in second semester February 15 Institut National des Sciences de l Univers M le Directeur 77 avenue Denfert Rochereau 75014 Paris FRANCE DEADLINES Date of receipt For time in first semester September 1 For time in second semester March 1 UNIVERSITY OF HAWAII Director Institute for Astronomy 2680 Woodlawn Drive Honolulu Hawaii 96822 U S A DEADLINES Date of receipt For time in first semester September 1 For time in second semester March 1 Requests for observing time on the Canada France Hawaii Telescope are made to the member agencies There are two competitions per year one for the first semester January June and the other for the second semester July December The mailing addresses and dead lines for proposal submission are given below for each of the three agencies Les demandes de temps d obser vation avec le T lescope Canada France Hawaii doivent tre sou mises aux agences associ es L attribution de temps sur une base comp titive est effectu e deux fois par ann e une fois pour le premier semestre janvier juin et une fois pour le deuxi me semestre juillet d cembre Les adresses postales et les d lais de soumiss
4. averaged surface brightness and corresponding high inte grated metal production of this flat spectrum population that we discussed above do suggest that this is a phase that a substantial fraction of galaxies must go through and that by studying these objects we are looking at a cosmologically significant episode of activity 4 Towards a Unified View of the Universe atZ 3 Our picture of objects at high redshift is still very patchy with just a few objects found by diverse but nevertheless highly pre selective methods Whether the objects are found through their strong radio emission or their vigorous star formation as discussed here or through high column densi ties of neutral gas as in the systems producing quasar ab sorption lines or through being associated with luminous quasars we can be sure that strong selection effects are present It is therefore premature to attempt to fit these sparse data into any comprehensive picture of galaxies at high redshift Nevertheless our progress in just the last few years has been dramatic We will need more objects found by these various tech niques and the development of new techniques so that the base of our knowledge is widened Once further objects at high redshift are found high resolution imaging in both the optical and infrared wavebands will be essential before we can begin to understand these galaxies at high redshift and hence solve the puzzle of galaxy formation from a direct ob
5. ay is in the main observing room A smaller HP 9000 S300 UNIX machine called shemp This machine is primarily a smart bitmap display server needed to provide a user environment in the old ob serving room For instance FTS observers must re main in the old observing room to be near the control electronics Think of shemp as a very smart windowed terminal to moe A second small HP 9000 S300 UNIX machine called larry This machine is like shemp except that it is placed in the third floor observing area to provide a user environment for coud Think of this machine also as a very smart windowed terminal to moe All of our machines are connected together via and only via a summit network This network connects to our Waimea development machines to provide fast debug turn around and efficient monitoring of user state The network also provides for automatic data storage in Waimea mo ments after the data is acquired 4 Remote Observing The above hardware description points out that all ob serving at CFHT is actually remote observing from the com puter point of view Since the main data acquisition machine moe has no screen a second computer sometimes small sometimes large is used to provide the display hardware through the network This means that logging into a Waimea bitmap screen as focam is technically identical to the summit situation Remote observing is now reduced to non computer issues 5 The Live Wir
6. cis de la fente avec les lignes du CCD Un logiciel a t d velopp pour la spec troscopie multi objet avec la machine PUMA construite par l Observatoire de Toulouse Il 500 5000 5500 6000 5 00 9 Mgl z 1 02 4 50E9 L LL ee li 4500 5000 5500 6000 lambda angstroms Figure 7 6500 7000 6500 7000 permet entre autres la s lection grossi re des objets avec le curseur sur une image du champ pr alablement obtenue suivi du calcul pr cis de leur positions un contr le des positions pour viter un recouvrement des spectres l addition de trous sur le fond de ciel l envoi d ordres de percage des trous la machine PUMA Les trous de 6 300 um sont perc s avec une pr ci sion de 10 um dans un clinquant dont le sup port peut tre ensuite ins r dans le plan focal F 8 Une proc dure compl te de selection des objets et cr ation d un masque prend environ 15 minutes Une proc dure d offset precis du t lescope Page 9
7. data management High Resolution Camera Detector and data management CIRCUS network disc and tape services 2 Consistency The key to all of the above mentioned environments is consistency All observing accounts look and feel the same The exposure management is the same The file formats are the same Quick look and tape management are the same The only things which appear different are those fea tures which are unique to the current instrument And even these features are presented in a standard fashion in the observing environment screen forms Consistency is important for three reasons visitors often use many instruments over time support astronomers must support several instruments each and the computer support team can leverage much of its work A best case example from the computer team point of view was the Fabry P rot session It was decided on fairly short notice to support a scanning Fabry P rot run We needed a custom log in ses sion instrument control user interface forms FITS cards and automatic scanning during iterative exposures The to tal amount of work was three normal days for two people 3 Hardware Behind the Scenes The final hardware solution is slightly different from the originally envisioned set up described two issues ago Two influences are responsible IRAF has been successfully ported to our HP 9000 UNIX machines multiple simultane ous instrument support has increased The first
8. e Fire Drill Dilemma A Live Wire analogy was developed by the computer group By this we mean that our systems are in use con stantly Any maintenance or changes let alone real prog ress must be done without interrupting service This anal ogy is probably appropriate for the whole company As we were redesigning our computer architecture we had a new opportunity to tune our solution with this in mind The Fire Drill is the other horn to our dilemma We can not just fill the computers fulkof epoxy and never touch them Every few days the telescope undergoes a transformation into an en tirely new tool Also each time an observer comes to CFHT a new feature may be required The problem then is to provide a malleable but rugged software environment Rug ged in this sense means internally rugged The main solutions to this problem are The use of many separate processes This allows us to know and prove that the majority of the software is untouched for any given change Even recompiling untouched code is risky The separation of user interface and specific function ality Most of our user interface is handled by sepa rate processes that in turn invoke the real process This allows us to change either without affecting the other Stateless programs Most of our programs do not hang around and keep state in their data structures Instead they run to completion and keep any state in Comit canadien de demandes C
9. influence will allow the summit SUN machine to migrate to Waimea where it is badly needed while at the same time reducing out summit computer support problem to one vendor The sec ond influence requires an example The June 1989 FTS run includes the sharing of each night with a few scattered hours of CCD shearing interferometer work Also the final day of this run needed to support Reticon Coude set up This is not only three different instruments but three different detec tor control technologies While the software architecture and instrument control hardware supported this type of shar ing we found we did not have sufficient displays in all the right places to support the interactive user sessions Our final hardware solution is composed of A large fast fully configured HP 9000 S800 UNIX ma chine called moe This machine has all of the instru ment control hardware connected to it It also runs all of the data acquisition software even when there are Page 7 multiple sessions However it does not have a display We find that we get higher performance if we use net worked displays since we then have more CPU s work ing for us A smaller less fully configured HP 9000 S800 UNIX machine called curly This machine serves two pur poses It acts as a backup computer in case moe fails It also has a high resolution bitmap screen which is used through the network to provide one of the user en vironments The curly displ
10. ion sont indiqu s ci apr s pour chacune des trois agences Page 8 c o M le Directeur Institut Herzberg d astrophysique Conseil national de recherches Canada Ottawa Ontario CANADA K1A OR6 DATES LIMITES cachet de la poste Pour le premier semestre 15 ao t Pour le deuxi me semestre 15 f vrier AGENCE FRANCAISE M le Directeur institut National des Sciences de l Univers 77 avenue Denfert Rochereau 75014 Paris FRANCE DATES LIMITES date de r ception Pour le premier semestre 1er septembre Pour le deuxi me semestre 1er mars UNIVERSITE D HAWAI Director Institute for Astronomy 2680 Woodiawn Drive Honolulu Hawaii 96822 U S A DATES LIMITES date de r ception Pour le premier semestre 1er septembre Pour le deuxi me semestre 1er mars a separate file which can be modified by the computer group to change behavior without recompiling e Reusable code A lot of attention has been given to both high level and low level tools A user interface language was developed to handle forms and invoke runtime control packages Device control libraries were developed to take care of the tedious details These features are what allowed us to create the Fabry P rot session in three days One source code tree Using the T1 network between Waimea and the summit allowed us to keep a single copy of the up to date source code This guaranteed that we never used an old version by mistake
11. servational stand point There is little doubt that CFHT will play a major role in this effort Simon J Lilly and Lennox L Cowie University of Hawaii Relevant Biblography Lilly S J 1988 Discovery of a radio galaxy at a redshift of 3 395 Astrophys J 333 161 Cowie L L Lilly SJ amp Gardner J 1988 A Cosmologically significant population of star forming galaxies Astrophys J Lett 332 L29 Lilly S J 1989 Faint identifications of 1 Jansky radio source empty fields A new look at radio galaxies at high redshift Astrophys J 340 77 Cowie L L amp Lilly S J 1989 A field galaxy at a probable redshift of 3 38 1989 Astrophys J Lett 336 L41 LATEST NEWS ON INSTRUMENTATION Completion of the Data Acquisition Computer Upgrade Project 1 Overview The new computers and the new software architecture that goes with them are fully released and operational at the summit All of the currently available CFHT instruments and many of the visiting instruments are supported The list of directly supported instruments includes FOCAM full support e UVPrime full support Herzberg full support TIGER full support e Multi Object Spectrograph full support with object selection punch machine and long slit mode Coud Reticon and CCD Fourier Transform Spectrometer full support Fabrey Perot full support Shearing Interferometer Detector and
Download Pdf Manuals
Related Search