The pulsar filterbank data acquisition system was designed for pulsar observations using the 13-beam HI (Multibeam) receiver at Parkes. It can, however, be used with any other receiver or pulsar filterbank.
Data are acquired with PMDAQ (Pulsar Multibeam Data Acquisition system) running on the DEC/Compaq Alpha puma01. This is controlled by PMCTRL, running on any Sun workstation, usually pavo. PMCTRL has a Tcl/Tk user interface, PMPAR. PMCTRL interfaces to the telescope control system and AT clock via the PC program DESK and to the feed-control system OPERFCC which normally runs on one of the three screens of the Linux PC goonumbla. Monitor data are obtained from PMDAQ and processed and displayed by PMMON, running on a Sun workstation.
Please note that this system is under contining development. We will make every effort to ensure that these instructions are kept up-to-date. Different parts of the system are the responsibility of different people and errors, problems or suggestions should be reported to the appropriate people, as follows:
The feed translator is controlled via the OPERFCC Tcl/Tk interface normally displayed on one of the three screens on goonumbla. This interface can run on any of the control-room Sun workstations provided you are logged in as `observer' or `pulsar'. In normal operation it is displayed on the middle of the three goonumbla screens. As goonumbla is a Linux PC, and as OPERFCC at present runs only under Solaris, you will need to find or open an xterm on this screen, and log into the Solaris machine sagitta first. From a Solaris prompt, type;
>operfccThis should bring up the GUI and open the connection to the FCC in a few seconds. The "Status" lights should all be green.
To manually select a receiver, highlight it on the list, select track or lock, and press 'Move selected receiver on axis'. If there appears to be a problem with the FCC system, consult a local expert (Brett Preisig).
For operation with PMCTRL, connection to the GUI is automatic (no matter which machine it is on), and you can select receiver, feed angle, etc under program (including schedule) control.
Currently running on PC in the computer/correlator room (pkdesk). Performs telescope control functions using REMOTEOBS via port 5025 and gets telescope monitor data using REMOTEMON via port 5030. It is also is host for AT clock system, via port 5040.
If there are major problems with telescope control, pkdesk can be rebooted by pressing reset switch on front panel (behind door on front of PC). It shares a keyboard and colour monitor with three other PCs - to select DESK, select the "D" display. It takes about 30 sec to reboot. Call the on-call person (after-hours) or a local expert if problems persist.
SHOWTEL normally runs on the MicroVax YOWIE. Occasionally it will will start beeping and display the message "DESK may be down". This means either that DESK is in some abnormal state (most likely), or that YOWIE has lost its connection with DESK (less likely), or both. If the problem persists, try rebooting pkdesk (as above). If the SHOWTEL errors still persist, you should try the following: log into YOWIE as `observer' (its console near pkdesk should be already - or you can telnet to YOWIE from any other machine), and type netcom. It should respond with "nonexistent process" and then the id of the created process.
A unix version of SHOWTEL is now available. This can be run on any machine logged in as `observer' or `pulsar'. Just type showtel at the prompt. If this is used, the PM system has no dependence on yowie.
The digitizer (bottom of Rack E) requires several control lines. These are normally left connected. The External 1 MHz light should be green. When observing the Enable light comes on.
Lines at the rear of the digitizer are:
Currently the 5-sec tick and an 11 Hz signal used to drive the pulsed cal for testing purposes are driven by the Front-End computer, located in Rack 1. The signals come from a panel of BNC connectors in Rack B, the 5-sec tick from EVDS8 and the cal pulse from EDV0. The Front-End computer is controlled from the Servo terminal; turn the rotary switch to Front End Computer, do ^C (Control C) twice to get the . prompt, then type
.r wpp 3to get both signals.
First move the receiver on axis (see OPERFCC above).
The RF output of the 1st LO synthesizer (normally the top sig gen in the right-most Rack) should be connected to IN1 of the small panel just below the synthesisers. Ensure that synth1 is selected on the switch on this panel. Set the synthesizer to RF 791 MHz (to give an LO of 1582 MHz) and Level +5 dBm.
Check the IF bandpass. The two polarizations for Beam 1 are split off behind Rack G and are at the top of Rack 2. Check these with the spectrum analyser (Rack 4); they should have a relatively flat bandpass to about 385 MHz at a level of about -65 dBm. The notch near 417 MHz is centred (at RF) on the Parkes Airport beacon at 1175 MHz.
If a pulsed cal is required, connect EVD0 (Rack B) to Pulsar Cal (Rack C) and Pulsed Cal Drive In (Rack 4) to Pulsar Cal (Rack 3). The Cal 5V (Rack 4) should be connected to the Multibeam Cal to Aerial Cabin (Rack 2), and the switches on the Cal control panel (Rack 4) set to Pulsar.
Now you must make sure that all the cables (IFs into filterbanks, filterbanks into digitiser) are connected properly. To do this it might be helpful to have some plots handy: MB frequency relations, MB pulsar back end, MB signal conditioning, MB band splitters, MB filterbank assignments, and MB digitiser configuration. See Filterbank Systems for further details.
Once you're sure all the connections are right, that the LOs are turned on properly, and that the feed is in position, it is time to check the input levels to the filterbanks. Firstly, make sure that the attenuator settings are set on the MB cable equaliser attenuation manual control at the bottom of Rack G. The nominal values for each polarisation and beam to be used for pulsar observations are shown under the appropriate heading on the card attached to the unit.
The levels can be checked with the filterbank monitors at the top of Rack E: use the `Start' dial to cycle through every 32-filter card (96 MHz-wide IF), for a total of up to 78 different voltages, using the code printed below the top panel. On average the voltages should read about 1 V, but you can only adjust attenuation one polarisation at a time - so you must calculate a rough average of the voltages in the three cards that make up each and every polarisation (e.g., 8-A), and add or subtract dBs with the manual control in Rack G accordingly. [For the time being, however, the attenuators on the downconverter chain produce birdies that are a function of attenuator setting. We choose settings that minimize interference at the expense of sensitivity for the digitizers. Do grep -A3 Attenuator pmobs.log for the latest settings, which should correspond to those given on the card.]
From a Sun workstation (normally pavo) telnet to puma01 and log in as `pulsar.' Then
puma01% script [puma01|1] pmdaq- PMDAQ then waits for commands from PMCTRL. If it is necessary to kill PMCTRL, then, for the time being, PMDAQ should also be killed with a ^C and restarted before restarting PMCTRL. The optional `script' command results in the generation of a log file `typescript' which can be used to diagnose faults. If a PMDAQ crash occurs, save it with the commands;
[puma01|99] exit puma01% mv typescript typescript_2002-11-23to save the contents of the `typescript' file, then restart as above.
PMCTRL and PMPAR run on a Sun workstation, normally pavo. The run file pmrun and other required files are under /psr1/cvshome/pulsar/soft_atnf/search/pmsurv/ (env-var PMSURV) in pm_obs/ (env-var PM_OBS). Log files (pmobs.db and pmobs.log) are in pmsurv/database/.
Log in as `pulsar'. Ensure that PMDAQ is running on puma01. Also you must make sure there is a tape in the Exabyte or DLT tape drive on puma01 (Rack B, to the right of the old puma in Rack C) if you will be writing data to magnetic tape. If you're using the DLT, press the `Select' button on the bottom left of the drive several (eight) times until the `35 GB' and `Density Override' lights are lit (i.e., we do not want to compress the data). Then
%cd $PM_OBS %pmrun- this runs PMCTRL and starts the Tcl/Tk script for the Graphical User Interface PMPAR, bringing up the control windows. This allows interactive control and parameter input into the system, either manually or through pre-prepared files, and also provides a monitor of its operation.
The top two panels in the main control window provide a status display and any messages which will also be entered into a log file. Just below this is a text `Log Entry:' line through which manual entries may be added to the log. At startup, a mandatory entry is required declaring the identity of the observer(s). Hit CR (Return) to complete this (and any other) Log Entry which will then appear in the window above.
Immediately below this lie two areas for entry of data, either Telescope Control Parameters or Data Aquisition Parameters which determine the operation of the telescope and PMDAQ hardware respectively. Data can be entered into the windows manually or they may be obtained from lists of preset values in listboxes which may be activated by clicking on the small green icons to the left of the individual windows. Select an individual item by clicking on it. They may also be read from or stored in named Configuration Files (having the extension .cfg), as described below. These parameters are only transferred to the equipment at the start of an observation.
At the bottom of the display is an area for parameter input which, together with the buttons down the right-hand side, provides the main control. The top left-hand window of this area allows one to determine the Origin of Control, which may be 1) Local, 2) Schedule (from named Schedule Files, having the extension .shd), 3) as a Slave of the Caltech pulsar timing system or 4) as a Slave of the HI multi-beam system.
In order to read in a Configuration File, select from the pull-down menu or type in the required file name (with extension .cfg). Hit GetCfg to load the config file. Modify entries if necessary and possibly change the file name, then hit SaveCfg to store for future use.
There are also facilities in this area to allow reading from named Schedule Files, to select lines for observation from a Schedule File and to manage the output Tape Operations.
There is a further, detached window which, when under schedule control, displays the selected schedule lines in the Observation List and permits their sequence to be adjusted (see under 'Schedule Control').
When not observing, you can do Tape operations (Mount, Skip, EOD, Rewind, Eject) by hitting the `Tape' button and selecting the appropriate operation from the pull-down menu. `Mount' assigns the output device and must be done before observing can commence. Current valid output devices are DLT1, DLT2 (if installed), EXA1, FILE and NULL. The tape device names on puma01 are EXA1: /dev/nrmt1a, DLT1: /dev/nrmt0a and DLT2: /dev/nrmt9a (not currently connected) . If FILE is chosen the data are written to a file in /ATOMS/data on puma01, the name of which is derived from the date and time; maximum file length must never exceed 2 GB. For NULL the data are written nowhere.
Before mounting a tape device, check that the tape label is correct. PMCTRL checks the tape.log file in pmsurv/database and indicates if the tape has been previously used. EOD skips the number of files listed as present on the tape when it is mounted. For the DLT - if you are sure you know what you are doing - you can skip forward fewer files than have been written and then start writing. The tape label is ignored for FILE or NULL output.
When not observing, you can change the output device. Just change the device name and label (if a tape) and select Tape Mount. If you have previously used a tape device and you haven't rewound or ejected the tape, you can continue writing to the tape from its present position, even if you have quit the program or been writing to another device. Note that in this case, you must Mount it again and continue writing from the present tape position.
Regardless of the history, when you Mount a tape device, PMCTRL will check the current tape position and advise you on whether or not to skip files to EOD. It does not automatically skip to EOD.
When finished with a tape, use Tape Eject on the PMPAR GUI. This keeps the tape history file in order. If you are forced to eject a tape by pressing the button on the drive, you will get an error message next time you mount the tape, but it normally reports the correct number of files and GB written. Check these, and if they look OK you can proceed with a Skip to EOD as usual.
Changing Tapes To change tapes, first hit Tape Eject on the PMPAR GUI. (The DLT is ready for unloading when the `Operate Handle' light turns green.) Remove and replace the tape. If using a DLT, reset the mode as described above to `35 GB' and `Density Override'. Then check that the tape label is correct, mount the tape and skip files or EOD if necessary. If you are about to overwrite data already written on the tape, you will be given a single warning about this when you start observing. You may then remedy the situation or hit `Observe' a second time.
At the end of each observation, PMCTRL lists the amount of tape used. Present experience indicates that 34 GB can be written to a DLT tape in uncompressed mode. 5 GB can be written to an Exabyte tape. When the tape reaches 95% of capacity, warning messages appear, which become more insistent as 100% is approached. The tape utility tape_used described below also lists tape usage, but doesn't take account of rewinding and overwriting a tape.
Under `Local' control, you may perform single telescope or data-acquisition observations based upon the parameters displayed on the PMPAR GUI.
To position the telescope, without starting any data acquisition, enter the Start coordinates or select a pulsar. If you hit (Return) after entering a Start coordinate, the entry is duplicated in the Stop coordinate box. This is convenient, especially for non-scan observations. Select the desired receiver and ensure that the Feed Control parameters are as required. Then hit `Posn' to move the telescope to the coordinates listed and the selected receiver into place. PMCTRL then waits for further instructions. `Posn' is particularly useful to save time while waiting for tape operations to complete.
To start a single observation, make sure Origin of Control is `Local', and that all parameters are correct, then hit `Observe'. This will position the telescope and receiver and then start an observation. An audio signal is sent to mark the end of the observation. The volume of this can be adjusted using
An observation or drive to a source can be stopped at any time by pressing the `Abort' button. If data have been taken, an EOF will be written and the file number will increment on the next `Observe'. The `Stop' button is only effective in schedule mode and will stop data collection at the end of the current observation.
During observing, the current pointing name (ID for survey observations), start UT and other parameters for the current observation are displayed in the Status window.
Log Entries can be added by the observer in two ways. A one line `Comment' (64 characters max) can be entered before starting an observation. This is recorded both in the tape header and in pmobs.log. Additionally, during an observation, comments can be entered into the `Log Entry' window. Log entry lines are terminated by CR (Return). It is important to log ALL significant events, i.e., anything which affects the data. In most cases, this will be your ONLY record of what happened. On completion, the log line is tagged with the UT and copied to the monitor window and to the pmobs.log file. Messages appearing in the monitor window and, at the conclusion of the observation, the .cfg file used, are similarly copied to the log file.
The schedule lines to be observed can be selected from the Select Schedule list using the pull-down menu and adds them to the bottom of the Observing List in the separate window. The item at the top of this list is the next to be observed and is removed from the list as the observation starts. Simple editting of this list is possible - clicking the Left mouse button on an entry cuts that entry into a buffer. This buffer may be pasted before any other entry in the list by clicking on the entry with the Middle mouse button.
Pointings already Selected or Observed are so marked in the Schedule list and can also be checked in pmsurv/database/pmobs.db or pmobs.log. Also, if you (re)load a schedule that has had some sources observed, the date and time of observation will be appended as a comment to the end of the respective line. When ready to go, hit `Observe'. The program will go through the Observing List, taking lines off the top, and stop when the list is empty. The Observing List can be worked on at any time during an observation. It is also possible to read in new Schedule Files and transfer lines to the Observing List while observing. Green buttons exists to transfer the whole contents of the Schedule File to the Observing List and to clear the Observing List.
When under schedule control, data from the .cfg files specified in the lines in the schedule file (< name>.shd) over-ride menu items. In particular, the configuration file given in the schedule determines any parameters not explicitly listed in the schedule. Typical schedule lines are as follows:
J0437-4715 pmc1.cfg t120 s600 f0 ! What a pulsar! B0833-45 pmc1.cfg t1800 s300 f90 ! Old faithful G123-07 123.45 -7.654 pmc1.cfg t1800 s300 f45 b3 ! Wow P0908-06 09 08 07 -06 05 04 pmc1.cfg t1900 s220 p135 G4691506 287.899 1.212 PMSURV.cfg ! Observe twice a dayEntries are free format in the sense that the number of spaces is irrelevant. The first entry must be the pulsar/pointing name and the first character of the name determines the coordinate mode:
J - a J2000 pulsar name B - a B1950 pulsar name P - J2000 celestial coordinates. The next six strings must be the hr min sec of RA and the deg min sec of Dec, respectively G - Galactic coordinates. The next two strings must be the longitude and latitude, respectivelyThe next string must be the Configuration File name. Entries after that are optional and identified by the leading characters as follow:
b - beam number (for multibeam receiver) f - Feed Angle, no parallactic correction. p - Position Angle of feed r - receiver s - sample interval (in microsec) t - observation time (in sec) ! - comment (up to 64 char)The comment must be the last string.
%watchdogand in a puma01 window, type
%watchdogThese scripts will ring a warning bell if PMCTRL or PMDAQ (respectively) crashes. Thanks to John Reynolds for writing these scripts.
The format of the file names is as follows:
e.g. 0437-4715_70cm_256_250_1024.ps has
band = 70cm
nchan = 256
tsmp = 0.250 ms
nblks = 1024
First log in to sagitta as `pulsar'. Then
%monrunThis brings up the `Mon GUI' (Tcl/Tk), `PGPLOT Window 20,' `Mon Diagnostics,' and `pmdaq_mon Server' windows.
Use the `pmdaq_mon Server' window to log in (telnet) to puma01 as `pulsar.' Then, if observing with the multibeam system (ie nchan=96),
%pmdaq_mon -d 1otherwise
%pmdaq_mon -d 1 -c nchanwhere nchan is the number of channels (e.g., 512).
On the `Mon GUI' window hit `Enable' (PMDAQ host should default to puma01; make it so otherwise). PMDAQ link should turn green shortly thereafter. If it doesn't, make sure pmdaq_mon is running on puma01:
%ps aux | grep pmdaq_monIf it's not, repeat procedure described above to start pmdaq_mon.
Select required functions (press green bars - red light comes on), select beam, nr of blocks, select the desired hard copy cycle, then press `Start' to commence automatic cycling, or `One Go' for single pass through selections. Press `Stop'/`Start' to change selections, `Quit' when finished.
Pgplot files: Pgplot files produced by the automatic hard copy cycle or by the manual `SNAPSHOT' are stored in the subdirectory $PMMON_RUN/plots/pool/. There is a `ghostview' facility incorporated into `Mon GUI' to access the plots in the pool. To use it:
Periodically you should move the plots you want to save to an appropriate directory (e.g., pmmon_run/plots/aug31/), and remove the others.
Channels: The button `Channels' allows the following processes on the data:
When using the `Channels' option, you can select `BeamNr' as `Auto'. In this way the pulsar (or candidate) beam is selected automatically by the PMMON program. This works ONLY with the PMCALx.cfg and PMCALx.shd files (in this case PMMON looks at the first five characters of the file name to check if it matches `PMCAL'), OR when using schedule files having the `@#C' or the `@#P' flag and the pulsar parameters information at the end of each schedule line (hexview, which is used to generate schedules, provides this feature automatically.)
Interference issues are discussed in other web pages. Certain things are known to cause interference and (unfortunately) many of these are within the Observatory and even within the focus cabin.
Before starting observations check these pages.
Occasionally, puma01 crashes for no obvious reason, bringing PMCTRL down with it. If this occurs in mid-observation (as it usually does), there are two possible ways to recover.
The safe way is to restart PMDAQ and PMCTRL, manually eject the tape (by pressing the eject button), remove and reinsert the tape. Then remount the tape and skip to EOD, restarting as normal. The down-side of this method is that you lose the data from the incomplete observation prior to the crash.
A somewhat more risky method which saves the prior data is to restart PMDAQ and PMCTRL, leaving the tape in place. Then mount the tape and do a WEOF. It is possible to proceed with normal observations from here, but this leaves the tape log in a bit of a mess and file numbers will be wrong. Preferable is to then eject the tape, remove and reinsert it, mount it again and skip the required number of files. N.B. this will probably be one more than stated for EOD, and so must be done by entering the number of files to skip on the GUI (bottom right) and doing Tape Skip. File numbers should be correct after this. When the tape is next remounted, it may report a tape log error, but this is not normally a serious problem.
You may one day come across a bus error, or otherwise find that puma01 does not respond. Check the console, a vt320 terminal on the circular bench opposite the machine itself. If there's no life on the console and you can't log in remotely, or if you're otherwise convinced the machine needs rebooting, open the hinged door on the front of the machine and cycle the power button. This takes a minute or two - watch the console and wait for the login prompt to appear.
If puma01 has been rebooted, and maybe on other occasions, there can be a DMA timeout in the digitiser interface to puma01. This is indicated by a sequence counter error followed by a puma01 INTERNAL ABORT and a general crashing sound. A fix which usually works is to power cycle the digitiser. The power switch is a black switch at the back of the rack, mounted on the left wall of the rack adjacent to the digitiser. Another possible cause for this error is a broken connection between the digitiser and puma01 - check the data cable and plugs.
There are now several DLT tape drives on site. If puma01's for some reason becomes unusable, you can borrow one of the other ones (if not in use!), make sure the SCSI ID switch in the back is set to 2, and daisy-chain it to the existing DLT drive on puma01. Ensure it is properly terminated. If puma01 wasn't rebooted since the last time it had the second DLT on it, that's all you need to do: the device is /dev/nrmt9a. Otherwise, you should reset puma01 (see above.)
To stop the screens switching into power-save mode after 10 min, do RMB on the background, select Workspace, Properties, hit continue on the message about colour, RMB on Category, select Miscellaneous and turn the Screen Saver off.
Return to top.