next up previous
Next: About this document ... Up: news Previous: taxx and tax; November

Polarization modes enabled in correlator software; November 2005; Jayaram N. Chengalur

Polar mode observations are being carried out on experimental basis since GTAC Cycle 9. For more details, please contact Dr. Jayaram Chengalur (


A number of new correlator data acquisition modes are now available for test. These modes include:

(1) Full stokes (in either USB or LSB), (2) 256 channels in RR and LL (in either USB or LSB) for all baseband bandwidths 16 MHZ or smaller (3) identical data being piped through both halves of the correlator (either for possible SNR improvements, as routinely used elsewhere, or for testing) (4) In addition of course, the old "Indian Polar" mode continues to be supported.

Introduction of these modes has required a change in the lta format. Existing binaries of programs like ltahdr, xtract, gvfits etc. will be unable to read the new data files. Modified versions of all these programs which will read the new as well as the old format are available.

There has also been a substantial change in the way in which users initialize and reconfigure the correlator and das software. Rather than having to consistently edit several text files on several different PCs and supply consistent command line options, users now run a single program that guides them through a menu. (The menu is a clunky old text based one, volunteers to produce something more fancy are welcome). This program runs on the ONLINE machine ("aditya") and produces a parameter file ("/temp2/data/corrsel.hdr"). Wrappers running on aditya read this file and will initialize/reconfigure the correlator as appropriate for the selected observing mode. Information required for configuring acq30, dlytrk etc is also taken from this same file and transferred to appropriate locations by dassrv. This entire operation of transfer of parameters is transparent to the user.

Step by step instructions for running the new modes, looking at the data in real time, or offline, and converting to FITS follows:


1. log in to corracqa, corracqb, corrctl as observer and run "setdvl" on each PC. (This sets the paths and environment variables appropriate for the new "development" software. NB: if there are any shared memories floating around from the older das versions, you will need to delete them).

2. login in to mithuna and mithunb as observer and chose "d" when promoted for which software system to use. (Again, this sets the paths and environment variables appropriate for the new "development" software. NB: if there are any shared memories floating around from the older das versions, you will need to delete them).

3. login into aditya as observer, and cd to /home/observer/dassrv-tst

4. If you want to initialize the correlator, run corr init on *aditya* (This will run corr_config and newdly_config on the corrctl PC).

5. Choose the observing setup by running corrsel on *aditya* (This will setup the file /temp2/data/corrsel.hdr, which contains information on how you want the correlator configured. corrsel allow a menu based selection of the observing mode, integration time, clksel etc.

At the moment, the following modes are predefined in corr_sel:

0 IndianPolar 1 UsbPolar 2 LsbPolar 3 UsbHighRes 4 LsbHighRes 5 UsbCopy 6 LsbCopy 7 AllU130 8 AllU175 9 AllL130 10 AllL175

In addition, you can configure a vast number of modes by editing the DPC_MUX and MAC_MODE fields in the corrsel.hdr file.

The valid MAC_MODE values are RRLL RRRL and RR

The valid DPC_MUX values are any of the strings:

IndianPolar, UsbPolar, LsbPolar, UsbCopy, LsbCopy, /* new modes */ AllU130, AllU175, AllL130, AllL175,

arar_arar, alal_alal, brbr_brbr, blbl_blbl, /* historical */ aral_brbl, aral_alar, brbl_blbr, arbr_albl,

Again the names should be self explanatory. In addition, the mode can be a string of the form dpc_abcd, where 0<= a,b,c,d <=3. A mode of the form dpc_abcd means that the dpc mux sends channel a data to where channel 0 data normally goes, channel b data to where channel 1 data normally goes and so on.)

6. Reconfigure the correlator to the selected mode. To do this type

corr reconf on *aditya*.

7. start acq, acq30, sockcmd fstop, dlytrk,dassrv,collect on the appropriate machines. (acq can be running when you reconfigure the correlator, but as before, dlytrk and fstop should be stopped). Note that none of the programs need command line options, but instead, as explained in the introduction, will automatically understand the selected observing mode.

NB: The acqs on corracqa and corracqb should be started within 1 STA cycle (i.e. 128 ms!) of each other. The easiest way to do this is: (a) export the corracqa acq window to all desktops.

(b) go to the corracqb desktop and align the corracq and corracqb acq windows one below the other

(c) type "acq" in both windows (but don't type ENTER!) (d) hit the ENTER button in both windows in quick succession.

I agree that this is a pretty silly way to do things. A more automatic method is in the prorverbial pipeline.

8. Ask the operator to give an init and add a project from the master terminal. The bandmask while initing and adding a project should be correspond to the chosen mode.

NB: The bandmask is a bit mask with bits in the order USB130 USB 175 LSB130 LSB175. For a pure USB observation the bandmask is 3 (e.g. UsbPolar, UsbCopy, AllU130..) For a pure LSB observation the bandmask is 12 (e.g. LsbPolar, LsbCopy, AllL130..) For a dual sideband observation the bandmask is 15 (e.g. IndianPolar). It may be possible that putting bandmask to 15 will work for all modes, but I have not verified.


10. The only way to look at the data in real time is Manish's new dasmon. You have to run a modified version of this program which is in /home1/manish/new_dasmon/dasmon on mithuna and mithunb. NB: In the new modes the data is divided between the two halves of the correlator. For example in the UsbPolar mode U130xU130 and U130xU175 come in mithunb (i.e. the "ltb" file) while U175xU175 and U175xU130 come in mithuna (i.e.the "lta" file). Hence (a) dasmon in mithuna or mithunb will only show you one half of the data (b) "normalization" cannot be applied in some modes. To normalize U130xU175 one needs the U130 and U175 selfs, but both will be available only when one merges the data streams.

11. ltahdr and other lute programs will not work. Instead use tltahdr (tltahdr is backward compatible, but I plan a proper upgrade only after a bit more testing).

12. tax and xtract will not work, instead use ttax and txtract. Please remember that each file has only half the data. The programs do internally consistent things, but you need to be aware of what you are doing. Avoid normalization in txtract, of course.

13. you can merge the lta and ltb files using tltamerge. This merged file will have the complete data, and is easier for playing around with ttax and txtract.

14. the merged file can be converted to FITS using tlistscan and tgvfits. The individual files can also be converted using these programs, but I don't seem to be able to get AIPS to understand the polarizations in a file with only RR RL or only LL LR. I think that only some partitioning produced by the VLA correlator is understood. The combined RR RL LR LL FITS file is interpreted without problem.

taxx - For true polarization lta files. taxx output is easiest to understand when it is run on a merged (USB and LSB merged) lta file.

All the other offline programmes work on either IndianPolar or true polarization lta files.

next up previous
Next: About this document ... Up: news Previous: taxx and tax; November
WWW Administrator 2015-10-07