COMPASSDocumentation

on the

COMPASS test beam DAQ


Note:

The Data Acquisition System used for the time being is based on the WA89-DAQ. The information you can find in this document still contains many WA89 specific parts ....
The COMPASS-DAQ, that will eventually replace this system is described elsewhere.


Notes on the COMPASS test beam data acquisition

Index

  • Run control and tape handling
  • Changing tapes
  • Calibration files, evtcon, evtdump, eobdump, evtdsk
  • Tests of hardware, changing modules
  • Power fails and other troubles
  • Online Messages
  • Power supplies
  • About monitor tasks, paw and global sections
  • Load frontends and DAQ after power up

  • Run Control

    Note: Whenever you change conditions on your detector, open for access or change the tigger or readout you should stop the current run, write a message in the logbook and then, after the changes eventually continue. The steps to write to tape in sequence:
    1. Put in the tape
      Wait until the green LED comes on.
      If not, make sure the tape is inserted correctly.
      CAREFUL: Never remove the tape, if the green LED is off! Never try to remove the tape without operating the handle!
      Should the yellow LEDs not come on within 2 minutes, do a power off/on cycle of the tape drive, wait 1 minute.
      DO NOT START ANY SOFTWARE BEFORE THE YELLOW LEDs ARE ON!
    2. logon to MY (login-name: onl)
      Make sure (pwd) you are in /us0/onl/run ( cd /us0/onl/run if necessary)
    3. type ctl
    4. commands are START, STOP, PAUSE, CONTINUE, MOUNT, DISMOUNT and QUIT
    5. the first start also asks for the tape to be mounted
    6. tape unit should always be 0 (DLT).
    7. After you gave the command watch the response on the message page
    8. It takes ONE MINUTE to start or stop a run on cassette! Please be patient. The first spill after a start or stop will produce error messages: This is normal!
    9. Mounting a tape does NOT involve any physical action, so it will succeed even without a cassette in place. Mounting only declares the tape unit(s) for subsequent writes. If a run starts and no cassette is in place the system will wait infinetly until you put one into the drive(s)!
    10. The "label" you type in at response of the mount command is for bookkeeping purposes only and should be the label written on the cassettes sticker.
    11. lastruns (from the shell) shows the run/tape status
    12. Should you get messages "error writing to tape" on the message screen:
      STOP the run
      DISMOUNT the tape
      wait until the green light comes on.
      Lift the handle and take out the tape.
      Put in new cassette and START the next run.
      If errors occur again:
      STOP run and DISMOUNT again
      switch the power off/on in the back of the tape drive chassis.
      Load a new tape and start again.
    13. Runs will be written to tape until the tape is full. Then there will be the message: "tape dismounted". Wait until the tape is dismounted and the green light is on. Lift the handle and take out the full tape. Then load a fresh one (it should have its proper sticker), use ctl to mount the new cassettes and finally start the next run.
    14. **** If you want to take away a cassette:
      1. STOP the run
      2. wait until you see the message "run stopped"
      3. DISMOUNT the tape (command dismount)
      4. wait for the dismount message and the green light
      5. unload the cassette

    Changing tapes

    1. If the end of cassette is reached, wait for the green light to come on (don't push any buttons). Then lift the handle on the front of the drive and take out the tape.
    2. Move the WRITE-PROTECTION ( switch at right side of tape) so that you see a . Write the last run number (see DAQ Monitor) on cassette.
    3. Lift the handle .Put the tape smoothly into the drive. Close the handle. WAIT UNTIL the yellow LEDs comes on.
    4. Start a new run. In case there are tape write errors for the tape just started:
      STOP the run (ctl)
      DISMOUNT the tape (ctl)
      Wait until they are ejected
      Replace the bad cassette.

    The start of run calibration files:

    they have to be entered in dxhypm:/us0/onl/run/filest.com The files themselves may be anywhere. The files will be copied at 80 characters per line.

    Tape writing:

    1. Run summary information is in a readable file.
    2. Every end of run creates an end of file on cassette.
    3. There is an automatic change of run every 200 spills.
    4. There are NO physical tape labels. There is a label field generated into the Start of Run record consisting of the station number and the number of the first run on tape

    The program evtcon

    evtcon configures paramters for evtbld, e.g. the number of spills per run. If you don't want the run get terminated automatically set this number e.g. to 10000. Most of the parameters in evtcon are for experts only ! Don't change them. If in doubt reload the default values (option 3). Once changed the parameters keep their value until y1 is booted. At boot time the defaults are loaded.

    How to dump events to your terminal:

    evtdump dumps events
    eobdump dumps eob records
    - type the command and answer questions

    How to transfer data to disk without writing a tape first

    rm fort.2
    ln -s /us91/ofl/test.tap fort.2     # this is an example!
    evtdsk
    1                #first question: trigger mask (octal)
    1000             #second question: number of events
    0                #third question: station mask, normally 0
    1,eqbnr          #fourth question: Your equipment number
                     #0 if you want all equipments
    
    EOB records (1103) are always transferred as well. evtdsk can be run from ANY account (e.g. your own).
    If you want to run the following in the background: Answer the two questions, then press Ctrl-Z and bg

    What to do if there are no events ?

    1. Most likely this is a hardware problem (no beam ?) or a trigger problem. (pulser switched off ?)
    2. Watch the status display "Data acquisition active" If this display refreshes once a spill (look at first line) you know that y1 is up and running and so is evtbld.
    3. check the online messages. There might be reasons why evtbld rejects all events. Frontend -9 is the SVX frontend, frontend -5 is the main frontend.
    4. The triggers may be halted because of a blocked readout coincidence etc.

    Hardware tests, changing modules etc:

    1. First, stop the run (if one is active!).
    2. Log on to y1 (onl) directory /us0/onl/run
    3. suspendevtbld.com will SUSPEND evtbld until you type
      continueevtbld.com to restart it.
      - You may switch off Fastbus or Camac crates in between these commands.
      When you switch Fastbus crates on again: Check the RAILS! The BH LED must go OFF!!!!
      Type continueevtbld.com only when power is on and the BH LED is off!

    What to do in case of a general power failure:

    1. In case of a general power cut: IMMEDIATELY SWITCH OFF ALL MAIN POWER SWITCHES on all wall power outlets and mobile outlets used by COMPASS equipment. Do NOT touch the power switches on the crates! This switching off will prevent the power from coming on automatically after power is restored (causing spikes and failures).
    2. Then: DO NOT WASTE TIME. CALL IN ALL EXPERTS! For every detector one expert must come to switch on the equipment. NEVER switch on High Voltage without the expert being present. BEFORE the computers are booted the Fastbus crate and CAMAC crate 1 must be on.

    The online message system:

    The online message system consists of a set of log files written to a commonly accessible filesystem /onlmess on all computers. There is one file per task. You can also get the 20 last messages extracted from all files via a WWW interface.
    1. The frontends:
      TSTCH2 Main CHI
      WA89C9 SVX CHI
      If messages occur only occasionally they are not serious, UNLESS they are accompanied by messages from EVTBLD reporting bad data structure, missing data etc. In this case or if messages appear EVERY spill call the equipment's expert. Pause the run (if one is active).

    2. TSTBL2
      Most messages coming from the EVTBLD task are VERY SERIOUS. EVTBLD reads the data from the frontends, combines the equipments and builds events. It checks for consistency of the event. It it finds anything wrong it will stop building events for this spill and reports the problem. If already the FIRST event in a spill is inconsistent, or if the overall structure is bad: THERE WILL BE NO EVENTS transported, as a result there are NO TRIGGERS displayed on the Status display. The problems above result in the event rejected. If already the first event in a spill is bad, NO EVENTS are accepted.

      Note about changing power supplies:

      There are two types of supplies, one for NIM crates (6 Voltages, +-6,+-12,+-24 Volts) and one for CAMAC crates (4 Voltages +-6,+-24 Volts) The CAMAC supplies come in TWO physical sizes. If a camac crate got a large size supply, always replace it by another large size supply. CAMAC crates are regularly checked by Software so there will be an alarm in case something is wrong. If a camac crate is faulty (dead or alarming) the fan unit might be bad and the power supply ok. If you want to change the fan unit and cables are in the way, LABEL them before you disconnect them!

      NIM crates are NOT checked by software. If they fail you might notice problems with the trigger etc. So wonder through the various barracks, look at crates and listen to a possible sound of alarm. If a NIM crate is faulty (dead or alarming) the fan unit might be bad and the power supply ok. If you want to change the fan unit and cables are in the way, LABEL them before you disconnect them!

      How to use the various monitor tasks on the (offline) computers:

      In general:

      • Log on to hh (any of h2,....,hg), account onl
        from the online Workstations: type "hhterm"
      • Everything (including the logon) MUST be lowercase unless stated otherwise.
      • Type "all top -I" to find out which machine is most idle and logon there to run the task.
      • It IS possible to run a given task on more than 1 machine, but usually this will be UNREASONABLE. So check on whether your task already runs on one of the riscs.
      • Global sections: They MUST be viewed by pawX11 on the machine the task runs.
      • HBOOK files: they can be viewed from any machine
      • To stop the task:
        ps -x shows you the current tasks : read its id
        kill -9 id stops it

      About PAW:

      The newest version of paw (the regular paw, not paw++) is available as pawX11. It includes now our extra features like "CHDIR" and support for true shared hbook memory. See below. Workstation type 1 (or 2,...10) for X-windows, 7878 for Falcos. Copy /us0/wbr/higz_windows.dat to your home directory for a good choice of window position and size. Edit this file according to your needs. The parameters are xposition, yposition, xsize, ysize One line for each window. Please note that your private pawlogon kumac must be called ".pawlogon.kumac" in your home directory. In case your kumacs don't work with pawX11: Remove all window sizings. to get a hardcopy:
      select "opla" BEFORE you start plotting (or "oppo" for portrait)
      select "clops" AFTER you plotted
      select "ppso" for the O-printer and "ppsb" for the B-printer and "ppsv" for the V-printer.

      True shared pawc for HBOOK/pawX11

      • With the new library $NEWCERNLIBA we can now map PAWC directly into shared memory (as on VAX/VMS). This means: no copying, no extra copy of pawc, no bookkeeping needed anymore, and you can see your histos grow in real time (event by event) WITHOUT overhead in time or memory!
      • To use the new shared memory feature:
        • remove all CALLs to ICOPGL !!!!
        • remove all CALLs to ICREGL and IMAPGL (if used for hbook!) (The calls above are NOT compatible with the new feature)
        • remove your "COMMON/PAWC/.... " line EVERYWHERE
        • remove CALL HLIMIT and replace it by CALL HLIMAP(100000,'NAME'), 100000 would be your size limit as previously in HLIMIT, NAME is your global section name. (The maximum size for HLIMAP is 1 Million words) (Only UPPERCASE for NAME) CALL HLIMAP must be done BEFORE any other HBOOK call!
        • start pawX11 glo name maps you to your global section greset id clears histo id (immediately), id=0 clears all
        • There is no need to change to //pawc !
      • Note: The global sections will be deleted automatically when not used. The Deletion can take up to 15 minutes. If you want to delete the section manually: "gsshow" lists allocated global sections "gsdelete name" deletes the global section "name" (verify with another gsshow. If sections does not get deleted: It is attached by a program or pawX11)
      • Note: If you increase your shared PAWC (by increasing the size in HLIMAP) and use the same name: You must FIRST delete the section BEFORE you start the program
      • Note: Global sections can be attached only at the machine that created them!
      • IMPORTANT: BEFORE choosing the name of your global section: edit the file /us0/onl/global.sections (h*) choose a name that doesn't exist yet and enter this name into the file
      
      
      
      
      

      Testbeam, load frontends and DAQ after power up

      1. Switch on Fastbus crate (make sure the "BH" LED is OFF)
      2. Boot my (if it has been booted with FB crate off) To test this: "mt -f /dev/CHIP rewind" if it says" no such device" you must boot
      3. load of SVX CHI
             On my:                              On any terminal server:
             connect_CHISVX              connect CHITEST4
        
             ctrl C                      ctrl C
             host 9600                   host 9600
             tra                         tra
                               	       	 
             ... login as chip pw ********       ... login as chip pw ********
             ctrl A                              ctrl A
             lo,p='get svx'              lo,p='get svx'
             ... wait until loaded               ... wait until loaded
             go 30000                    go 30000
             Ctrl-\ C                            ... BREAK
                                                 ... DISCONNECT
             
      4. load of primary CHI
             On my:                              On any terminal server:
             connect_CHIMAIN             connect CHITEST3
        
             ctrl C                      ctrl C
             host 9600                   host 9600
             tra                         tra
                                         
             ... login as chip pw ********       ... login as chip pw ********
             ctrl A                              ctrl A
             lo,p='get fblib'              lo,p='get fblib'
             ... wait until loaded               ... wait until loaded
             go 30000                    go 30000
             Ctrl-\ C                            ... BREAK
                                                 ... DISCONNECT
             
             daqstart.com
             
        ... logoff from my and logon again only to start a run on tape
      5. to switch off the FB crate (rarely needed)
             on my/onl/run:   suspendevtbld.com
             to continue:     continueevtbld.com
             
      6. REMOTE procedures (if you cannot go to the west hall or if you are at DESY)
        on my: "connect_CHIMAIN" or "connect_CHISVX"
        also connects to the CHIs, terminate with CTRL \ , then "c"
        to press the RESET button by software (careful, once the primary CHI is RESET, you cannot reset the secondary anymore), (only AFTER suspendevtbld.com !!!):
             on my: "/us0/wbr/testbeam/chiboot2"  (resets the svx CHI)
                    "/us0/wbr/testbeam/chiboot"   (resets the primary CHI)
             


      COMPASS home | test beam | general | schedule | activities | contact | documentation

      Last update: 06/08/1997 by Lars.Schmitt@cern.ch.