[evlatests] correlator offsets (from Neal Miller)
Michael Rupen
mrupen at nrao.edu
Wed Aug 22 13:04:35 EDT 2007
---------- Forwarded message ----------
Date: Thu, 9 Aug 2007 10:40:49 -0400 (EDT)
From: Neal Miller <nmiller at skysrv.pha.jhu.edu>
To: ksowinsk at nrao.ed, rperley at nrao.edu, mrupen at nrao.edu
Cc: Ed Fomalont <efomalon at nrao.edu>, Ken Kellermann <kkellerm at nrao.edu>
Subject: correlator offsets
Hi Ken, Rick, and Michael.
Sorry to single you out, but I'm not registered with
"evlatest" so I figured I'd share this with you. It
is regarding the correlator offsets, in this case for
program AM889 (L band, 4 mode of usual 3.125 MHz channels).
Both Ed and I had the impression that the correlator offset
errors were worsening, so he suggested that I send out a message to this
effect. I wanted to provide some quantification
of this, so I've investigated it in a rather simple manner.
For the AM889 data after the June 15 observations, the errors
have been relatively small. My reduction procedure has been
to ignore them until the end. I do the usual editing and calibration, image the
field (using ~120 facets, ~3/4 of
which canvas the primary beam with the rest handling brighter
flanking sources), use the images to self calibrate (phase
only), then perform another round of imaging. At this point,
I subtract the clean components from the (u,v) data. On
occasions where there has been some RFI I will do a little
more editing, but we've been pretty fortunate so far.
To make a correction for the correlator offsets, I take the
uvsub data and run UVMTH (opcode 'sub', input and in2 files
are the uvsub data). This should take the baseline averages
(the offsets) and subtract them from the data. I return the
CC's to this new file, and save that as my cleaned up (u,v)
data for that day's observations. With a lot of data coming
in and the goal being to eventually combine the days, I have not saved images.
However, I have been checking the
UVMTH step by doing quick dirty images of the central facet
(256" x 256") from both the UVSUB data and the UVMTH data.
This means my notes do have some numbers that serve as an
indicator of the magnitude of the correlator offsets for
each date. Visually, the improvement is obvious, too.
There are three files attached. The first is an ascii table
of the information, with the date of the observation, the
date converted to a number of days since the program started,
the rms for the UVSUB image (i.e., prior to any correction
for correlator offsets; in uJy), the rms for the UVMTH image,
and the GETJY fluxes for the calibrator source (IF1 and error,
IF2 and error, in Jy). I included the calibrator information for completeness -
I know that apparent variability is going
to include the correlator offsets, errors in flux calibration,
etc. The other two files are plots of the data. One shows
the absolute improvement, simply defined as the rms of the
central facet in the UVSUB image minus the rms in the UVMTH
image. The other is the same thing but with the improvement
displayed as a percentage ((uvsub-uvmth)/uvsub). I've color
coded the points by pointing, too, since there are variations
in where real sources lie and these will get mixed up with
the correlator offsets in my reduction procedure. None of
the fields have a particularly strong source; Pointing 2 does
have a strong source in an adjacent facet (~3' from the phase
center).
Things do look a bit worse recently, but it doesn't appear
to be a simple trend. The July 28 and 29 data were good,
while the July 26 and 27 and August 2 data had larger
errors. We had the very strong offsets for the first two
observe dates (June 3 and 15; I did not follow the above
procedure for these, instead attempting to correct the offsets
prior to calibration and imaging), and I had thought that
the fix was to turn off the self-test function of the correlator. Anyway, I
hope that these data can help elucidate
the origin of the problem, perhaps by looking for trends compared with VLA
observing modes of programs running before
the AM889 ones.
If there's more information that I can provide, please let
me know.
Thanks,
Neal
-------------- next part --------------
#Date num Ptg rms_cl rms_uvm calibrator
jun03 1 1 40.0 - 0.93771 0.00485 0.95272 0.00594
jun15 13 2 36.4 - 1.00539 0.00219 1.00484 0.00184
jun24 21 3 28.7 - 0.95113 0.00288 0.95345 0.00231
jun25 22 4 30.0 28.1 0.96530 0.00244 0.96619 0.00216
jul01 28 6 25.3 24.3 0.95926 0.00577 0.96763 0.00619
jul06 33 5 28.7 27.7 0.95813 0.00547 0.96087 0.00573
jul12 39 1 29.4 28.5 0.96262 0.00645 0.93843 0.00491
jul13 40 2 28.0 27.2 0.94247 0.00520 0.94546 0.00491
jul14 41 3 26.6 25.0 0.94877 0.00556 0.94967 0.00449
jul15 42 4 27.9 26.0 0.94998 0.00542 0.94761 0.00493
jul16 43 5 26.9 25.2 0.96825 0.00591 0.97136 0.00538
jul17 44 6 26.6 24.3 0.96677 0.00599 0.96490 0.00515
jul19 46 1 26.6 24.9 0.95148 0.00571 0.95782 0.00524
jul20 47 2 43.7 40.3 0.92772 0.00505 0.95387 0.00518
jul21 48 3 26.3 25.3 0.94941 0.00563 0.95223 0.00493
jul22 49 4 25.4 24.6 0.95656 0.00536 0.95960 0.00503
jul23 50 5 27.0 25.9 0.93707 0.00599 0.93105 0.00523
jul24 51 6 26.1 25.1 0.94856 0.00496 0.95175 0.00468
jul26 53 5 35.2 32.0 0.93473 0.00502 0.94006 0.00446
jul27 54 2 29.1 25.7 0.89893 0.00581 0.90539 0.00486
jul28 55 3 34.0 33.0 0.91546 0.00413 0.92687 0.00480
jul29 56 4 30.3 29.0 0.92045 0.00514 0.92968 0.00453
jul30 57 1 33.6 30.3 0.91514 0.00430 0.92756 0.00367
aug02 60 6 35.0 32.1 0.89258 0.00436 0.89636 0.00451
-------------- next part --------------
A non-text attachment was scrubbed...
Name: CorrErr_abs.ps
Type: application/postscript
Size: 31277 bytes
Desc:
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20070822/42af344b/attachment.ps>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: CorrErr_frac.ps
Type: application/postscript
Size: 32419 bytes
Desc:
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20070822/42af344b/attachment-0001.ps>
More information about the evlatests
mailing list