User Tools

Site Tools


lbaops:atcanotes

15 February 2016 THESE NOTES HAVE BEEN UPDATED. Advice on how to change the tied array gain to correct bad recording statistics is now included.

ATCA VLBI Observing

The VNC sessions for general ATCA observing (usually xbones:1 & caccc1:1) should be already setup. A separate VNC session (usually xbones:2) may be setup as per the general observing notes, to run the VLBI-specific tasks on cavsi1 i.e. the DAS and cDisko.

Observing

Schedules

Generally the schedule will have automatically been copied to Narrabri. If not, copy the sched (.sch) file from the ATNF FTP area for the experiment onto $ATCA_SCHED (/atca/sched/) on xbones. [Schedules must have the channel bandwidth set to 64 MHz, and be using one of channel 17, 33 or 49 for the zoom channel. The channel in each IF can be different. Channel 33 is the centre channel, coincident with the IF central frequency. Channels 17 and 49 are +/- 512 MHz away from the central frequency. This should normally be set automatically.]

If you have to make the schedules yourself, or if there is an error in the automatically generated schedule, the following information may be useful. The offset channels 17 and 49 are only required when the CABB central frequency cannot be set to the required tied-array central frequency, and hence channel 33 cannot be used. There are only really two bands in which this occurs: L-band and C-band.

When the required L-band central frequency is lower than 1730 MHz, you must use offset channel 49, which has its central frequency 512 MHz lower than the CABB central frequency (because L-band is LSB). For example, if you wanted to observe 1400 MHz in the tied array, you would set the CABB central frequency to 1912 MHz, and use zoom channel 49, which will be centred on 1400 MHz.

When the required C-band central frequency is lower than 4928 MHz, you must use offset channel 17, which has its central frequency 512 MHz lower than the CABB central frequency (because C-band is USB). For example, if you wanted to observe 4836 MHz in the tied array, you would set the CABB central frequency to 5348 MHz, and use zoom channel 17, which will be centred on 4836 MHz.

When using the CABB scheduler, ensure that the channel width (in the drop-down box next to the continuum central frequency input) is set to 64.

CAOBS

Antenna control is from the normal caobs program. Usually a calibration schedule will be created automatically with the name <exper>-cal. This will have a 1 hour “relative LST” scan on 1934-638 or 0823-500 to phase up the array.

Note: The UTC schedule used by VLBI cannot be used with the “track” command in caobs i.e. the scans cannot be extended. Hence the need for an LST schedule for calibration.

CABB

VLBI is done using a special 64 MHz “zoom” mode. This gives 64 MHz of data with 2048 spectral points for delay, phase and amplitude calibration. By default the inner 50% of this 64 MHz is used for calibration, and the central point is automatically flagged to remove a strong birdie.

The current version of cacor does not support VLBI (30-Jan-2014). For the moment you must follow these extra instructions

13-nov-2012: Further notes added about the new hybrid VLBI mode (2048 x 1MHz for internal ATCA baselines in IF1, and 64 MHz for tied array in IF2).

For “standard VLBI”:

  • stop and exit caobs
  • quit cacor
  • In the terminal which cacor was running type 'cccversion VLBItwommlos' then 'cacor'
  • restart caobs (don't forget the stop/start problem)
  • For “standard” VLBI, use the cfb64_vlbi CABB config. In cacor, enter the edit screen, 'get cfb64_vlbi', 'prog'. (For the hybrid mode, refer to blue text below.)
  • Once you exit the edit screen (by pressing the 'Enter' key after a successful programming) you need to give cacor the command

calband 3,4

For “hybrid” VLBI:

  • use the cfb64_vlbi_1m-64m CABB config.
  • In cacor, enter the edit screen, 'get cfb64_vlbi_1m-64m', 'prog'.

Once you exit the edit screen (by pressing the 'Enter' key after a successful programming) you need to give cacor the command

calband 0,3

  • The calband command tells the correlator to calculate the delays/phases/amplitudes from the 64 MHz zoom channels in the standard VLBI mode (calband 3,4) and from the 1MHz band in IF1 and the 64MHz mode in IF2 in the hybrid mode (calband 0,3). This parameter is reset every time you leave the edit screen, so you will need to repeat this step after every board programming. Type calband by itself to see the current setting.
  • Set the reference antenna (same as caobs) as close to W104 as possible. In cacor:

refant X

or in caobs:

set reference ca0X

* In addition, in the hybrid mode, flag out other birdies in the IF1 band fflag f1 birdies

  • Set the integration time. In caobs type:

cain cycle 5

  • Set the tvchan range back to its default setting. In cacor type:

tvchan default

  • Start the scan in caobs

View the data links by clicking on “F1 DATA LINKS” and “F2 DATA LINKS” in cacor.

BOCF and MSYNC should be close to zero. The pairs of 1ms values should be within 20msec. If not, or any line is red the link needs to be reset. In cacor run

rtminit CA02 1B - for example.

Choose antenna, product and frequency as appropriate.

It is almost always useful to do a rtminit all during the first scan after the correlator is reprogrammed.

If a delay jump is reported at the correlator and/or vis shows a problem, check the links as above and reset with “rtminit” as described earlier. Even if no obvious errors are seen in cacor but vis jumps in phase/amp are seen, then try “rtminit all”.

Please BE PATIENT as corrections take a few cycles to show in vis.

To reset all the counters, in cacor

rtmr - useful so one can see the error rate for the session.

Other useful resetting commands when in deep trouble, are:

diginit - to reset the digitisers at the antennas (cf. rtminit)

reset acal - to reset and restart calibration

reset abdel - to reset cross-polar delays

reset delays - to restart the delay calibration BUT please use with EXTREME caution and only if you understand what they do and their effects.

Calibration

  • Only once all the boards and links are working should you proceed with array calibration. First, set the attenuation to equalise the power coming from each antenna. In cacor:

userms - this uses the whole band to compute rms

tatts 20

atts on - wait for rms of ~20 to be reached

atts off

usegtp - substitutes GTP for RMS in SAMPLERS window on cacor & uses only tvchan range.

  • View GTP in cacor SAMPLERS window and choose median value
  • In cacor,

tatts X - sets atts target to X, being the median value you just chose

atts on - attenuators servo to get GTPs to X.

(But don't use this in the hybrid mode as the GTPs are quite different between IF1 and IF2)

  • Once settled, in cacor:

atts off

abph X X - The AB phase offset to get circular pol from the tied array. See the table below for correct settings for different frequencies.

for hybrid mode, abph in IF1 should be 0, and refer to table below for appropriate value for IF2 for the current observing frequency.

  • Check that the delays and phases look stable. You may want to average some of the spectral points together to get a better delay solution. For example, in cacor:

delavg 8

Then to calibrate delay and phase, in cacor (or use “cor <cacor command>” in caobs):

dcal - Delays should go to zero

pcal - check that the AB phases go to +/-90 degs.

To get Tsys measurements working for ATCA, you must follow these altered instructions now.

We now change back to using the “wideband” data which can be used to keep track of Tsys.

calband 0,0

This step is the same for the hybrid mode as well. If tracking one of the primary cals i.e. 1934-632 or 0823-500, in caobs

corr acal

If you are not tracking a primary cal, you can get the flux of your calibrator from the ATCA calibrator database at the frequencies you're observing (click on “search calibrator” in the CABB scheduler). In this case you can (in cacor):

acal F1 F2

Where F1 and F2 are the fluxes of the calibrator at the two IF frequencies.

Check that delays and phases are still ok, especially for the cross polarisations ab & cd (which should be an -+90 deg). You may safely do another pcal to reset phase without affecting anything else.

It should be noted that the calibration in calband 0 0 might be different than for calband 3 4, since the “continuum” band will be at the central CABB frequency, while the “VLBI zoom” band may be one of the offset channels at a different frequency. This will most likely have the biggest effect at L-band, where the RFI and phases may vary significantly over the 512 MHz difference. If this is proving to be a problem, you might want to consider going back to calband 3 4.

  • Setup the tied array. In cacor:

tie reset - to reset tied-array timing.

tie - to report the setup it returned to from previous time. You can reset the tied array setup by:

tie ant 12345 12345 - choose which antennas to tie (usually not ca06)

tie gain 0.1 0.1 - set the gain

tie pol c c - set circular polarisation

tie - to report what the setup is

Check the VLBI DAS levels and statistics on the DAS GUIs. Unusually low/high levels are indicative of setup problems. Try different gains in catie to get the stats right (usually 33 and 17).

Re-check in vis that all levels are as expected.

During the observations, the gain of the tied array may need to be changed to keep the statistics correct. To do this, change tie gain for each IF. Do not attempt to change the levels into CABB using the attenuators, especially while using the calband 0 0 mode, since the levels will usually be nowhere near the tatts setting you had during the setup stage.

Setup parameters

Band Phase zoom channel DAS profile
X -90 -90 33 vsop.pro
C (4.8 GHz) -90 -90 17 vsop.pro
C (6.7 GHz) -90 -90 49 vsop.pro
L (1.4 GHz) +90 +90 49 vsop_n.pro
S (2.3 GHz) +90 +90 33 vsop_n.pro
K (22 GHz) +90 +90 33 vsop_n.pro

Note 1: The vsop.pro profile is inverting and vsop_n.pro non-inverting. Recorded band would always be non-inverted. Note 2: If a vlbi_64_1_tda config is used in cacor, only the 1st value should be set by abph. The 2nd value should be set to normal 0 value.

Troubleshooting

The most common trouble is a non-responsive channel in the DAS window - the power to this channel will usually be too high or too low and will not respond to changing the catie gain. The solution to this is to get someone at the observatory to press the “FPGA reset” button on the appropriate tied array unit in the screened room. Each press resets the timing of the tied array link, and this may fix the problem. The timing will come back randomly however, so it may take more than one press to fix it. Each of the four channels on the tied array boxes has a set of LEDs above the FPGA reset buttons, and the channel that isn't working is usually obvious as its LEDs will look different to the others, and the FPGA reset buttons should be pressed until the LEDs match the expectation.

You may also find that after reprogramming that several baselines in a particular IF won't give fringes. These may or may not be common to one antenna. In this case, you should first stop and restart the scan; this usually fixes it. If not, the solution to this problem is to reprogram the blocks that are used to make the zooms, which by default are block 7 for IF1 and block 27 for IF2. Simply go into the Edit window in cacor, type prog 7 27 and wait for them to successfully reprogram, then exit the Edit window and reissue the calband 3,4 command in cacor. This usually fixes this problem, but may need to be repeated a couple of times.

Sometimes block 7 or 27 will be taken offline by the correlator. When this happens, you will see a red message in the correlator window to this effect, and assistance will start complaining about the block being offline. If 7 is offline then no data will be generated for DAS1, and if 27 is offline then no data will be generated for DAS2. If this affects the experiment you're running you will need to fix it. To do so:

  • Stop the schedule in caobs with

stop

  • First reboot the block with the “spray” command. In a terminal in the caccc VNC, type

spray

and press enter. Accept all the defaults until it prompts

Blocks:(? - help) [ 1 ] (note that the numbers in the square brackets may be different - it will be showing the last set of blocks that were rebooted with the spray command).

Here, specify which block you want to reboot and press enter. The same prompt should show up again, this time with the block number you gave in the square brackets. Press enter again to get the

Command? (? - help) [.system reboot]

prompt, and press enter again here. A little while later you should get a message saying

Block 27 (l-bcc27) - Return Status: OK

(of course the actual block number will be the one you rebooted).

  • Now quit the spray command with Ctrl-C and start pinging the block you just rebooted from the same terminal. If you rebooted block 7, then

ping l-bcc07, and replace “07” with “27” if you rebooted block 27.

Wait until the block has responded successfully to 10 pings at least before stopping the ping command. If the block doesn't start responding to pings within a couple of minutes, or the block stops responding to pings soon after it starts, contact the DA.

  • We must now reinstall the block into the correlator. Open up the “EDIT” menu on the cacor GUI. From the prompt that appears in the “EDIT” window, type

block and press enter, then

7 or 27 (depending on which block you rebooted) and press enter again.

You will see a menu describing the block you just rebooted, and at the top it will say

Block 27 (INSTALLED) (for example).

Remove the block by typing

rem then enter,

and the menu will now say

Block 27 (REMOVED)

Reinstall it by typing

inst then enter.

If something goes wrong at this point it will say something like

Block 27 (INSTALLED.. BUT NOT THERE)

In this case, remove the block again and reinstall it. Continue until it works, or you've been trying it for a few minutes. If you can't get it installed, contact the DA. If you do get it installed, press enter to get back to the main menu.

  • We must now reprogram the booted block. You should still be in the “EDIT” menu from the previous step, but if you have exited the menu by mistake, bring it back up. Now, if you rebooted block 7, type

prog 7 and press enter.

You should see the correlator attempt to program the block, and if you don't receive any error messages you should eventually return back to the prompt again. If there is an error, type

prog 7 again.

If this doesn't work within a few attempts, call the DA. If it does work, exit the edit menu by pressing enter at the prompt.

  • You should now reset the calband by giving the

calband 3 4

command in the cacor GUI's command box. Once you've done that, and you see the blue

CONFIG: Fringe rotator firmware version for CABB IF1 = 1

message in the log window, you can restart the schedule in caobs. Everything should now be working normally. If you don't think it is, call the DA.

Returning to non-VLBI observing

DON'T REPROGRAM CORRELATOR WITHOUT ADVICE

Currently a different version of the CABB control software is use for VLBI and normal observing. On 2/2/16 the normal observing version was newzmsy and the VLBI version VLBItwommlos. You can check the current and available versions by typing

cccversion

in a terminal pn caccc1.

To change versions

  • Stop the current scan
  • Quit caobs
  • Quit cacor
  • On a caccc1 terminal, run

corkill

cccversion newzmsy (or whatever version was running before VLBI)

  • Restart cacor in a caccc1 terminal (cacor)
  • Choose the right CABB config in cacor and click Edit, and program as per normal
  • Restart caobs

You may also need to run the following (definetely needed if not restarting cacor and caobs):

In CACOR must reset:

abph 0 0 - remove 90 deg phase offset

and also

userms

tatts 20

to return to the standard attenuation servo setting.

Reconfigure the correlator to the standard ca_2048_2048_2f config.
You'll need to reset “tvchan” to more appropriate values for the intended setup.

calband 0 0

tvchan default

fflag f1 birdies

fflag f2 birdies

N.B. On reconfig please check the F1 and F2 data links and rtminit any link that has failed. In fact it may be prudent to do

rtminit all

in cacor in all cases.

lbaops/atcanotes.txt · Last modified: 2016/07/29 12:44 by cormac