[evlatests] Adventures with Cygnus A -- Ka-band Edition

[Rick Perley]

    I have reduced and analyzed the Ka-band data from last weekend's 
Cygnus A test. 

    All three high frequency bands (K, Ka, Q) were observed in this 
10-hour run.  At Ka band, 5 pointing positions were utilized, spanning 
the 2 arcminute width of the source.  For each position, about 1800 
on-source integrations (each one second) was obtained, so about 30 
minutes integration time.  I chose 29 and 35 GHz for the two frequencies 
(BD and AC, respectively), using OSRO1 mode. 

    Referenced pointing, based on the very nearby calibrator J2007+4029, 
was done at C-band, once every hour. 

    Phase calibration used this same source.  The observing cycle took 
about 30 minutes, so calibration was done on this timescale.  This is 
quite coarse, but as Cyg A provides a lot of flux (and has an unresolved 
nucleus of 1.5 Jy), self-calibration was expected to work (and did). 

    Notes from the editing/calibration stage:

    1) About 5% integer zero records, nearly all associated with 
particularly antenna-IFs.  (see below).  Virtually no 'zero' data were 
found after ~15 seconds after the start of a scan.
    2) I used a 15-second 'QUACK' to get rid of all initial setup 
issues.  (I chose not to use the provided flags). 
    3) Antennas 11, 13 14 16 18 and 26 have no receiver at this band.
    4) Antenna 20 had a serious pointing issue. 
    5) Antenna 6, IF 'B' was blank (integer zero) throughout the entire 
database. 
    6) Antenna 22, IF 'D' gave only noise -- no fringes. 
    7) Only modest flagging was required to remove 'failure to tune' IF 
pairs, although in some cases, whole antennas (i.e. both IF pairs) 
'failed to tune' -- this seems unlikely!  Specific examples:  Antennas 
1, 10, and 4 each provided no fringes for an entire observing block (the 
5 pointing positions plus the calibrator -- these were observed in 
succession, so there were no band changes within the block)-- 12 minutes. 
    8) We had a single instance of a single-IF (5B) providing only noise 
-- this lasted for 1.5 hours, at the beginning of the run. 
    9) Other than these instances, there were only 3 or 4 scattered 
instances of a 'failure to tune'. 
    10) Bandpasses were quite stable (no changes > 1% in amplitude) with 
a couple of spectacular exceptions:  17C on two instances showed +/- 10% 
sinusoid of 60 MHz period, and 27B, with the same period, but smaller 
(7%) amplitude. 
    11) Delays were normal (less than 5 nsec) except for 15, as noted 
earlier.
    12) Amplitude (gain) stability was 'pretty good' -- roughly 5% pk-pk 
in amplitude, with most notable effects affecting all four IFs equally 
-- I would claim this is evidence that the variations seen are due to 
pointing errors.  A clear exception is 27B, with a 50% drop over about a 
6 hour duration.
    13) Phase stability is very hard to assess, due to atmospheric 
effects being rather large.  Plotting the IF1 - IF2 phases (AC - BD) 
(which removes most of the atmospherics) shows the usual gang of 
suspects providing phase jumps:  ea04 (once only), ea8, ea10, and ea17 
(continuously for all). 
   
    Comments on Imaging:

    The 5 target positions, and the nearby calibrator, were 
self-calibrated and imaged.   For these, I split out the data as a 
single 110 MHz-wide 'pseudo-continuum' channel, after applying the 
bandpass, delay, and polarization. 

    A) Calibrator (about 3.5 Jy).  A single A&P self-cal provided an 
apparently noise-limited image, with DR = 31,000:1 at 29 GHz, and 
25,000:1 at 35 GHz.  (The noise was 0.11 and 0.12 mJy, respectively). 

    B) Nothing close to this was found in the Cyg A images.  The primary 
beam is ~ 1.5 arcminutes, and with Cyg A (to first order) being a 
triple, with a bright nucleus and two very bright hotspots located 1 
arcminute away on each side, we can expect pointing issues to be 
paramount.  These expectations are clearly met in the images! 
The best image was with the beam center near the eastern hotpots -- the 
noise is 4.1 mJy (40 times that of the calibrator!!!), the DR is a 
modest 2000.  The nucleus  appears highly distorted (due to pointing 
wobbles, or more likely, bad referenced pointing), the opposite hot spot 
is in the first sidelobe, and is barely visible. 

    Images with the beam between the hot spots are much worse.  The 
worst is the 'on nucleus' observation, with a noise of 14 mJy, and a DR 
of only 150. 

    Unsurprisingly, the 35 GHz images are poorer than the 29 GHz images. 

    Interestingly the Q and U images are much better, with the noise 
level a few times lower than in I (but still far from the calibrator 
noise level). 

    I now progress to the Q-band data (without high expectations ...)