[evlatests] Ka-Band test

[Rick Perley]

    I haven't figured out this JIRA business yet -- this will have to do 
in a pinch ...

    Michael observed our favorite northern point-source calibrator, 
J0217+7349 (formerly known as 0217+738), along with a position 2 degrees 
north, plus another position offset by a larger amount.  Referenced 
pointing (using C-band, I think) was utilized, apparently (judging by 
the gaps in the high frequency data) three times.  The frequencies  used 
were 33.496 and 33.624 GHz, both with 128 MHz-wide subband widths.  Each 
position was observed for one minute, cyclically.  The total duration 
was 53 minutes. 

    Gustaaf was kind enough to provide me with a UVFITS version of the 
data, with flags already applied.  These seem to have worked well, 
except that the seven antennas without receivers (11, 13, 14, 16, 17, 
18, 26) were not flagged -- apparently we're not checking power levels 
as a flagging diagnostic? 

    Not all receiver-equipped antennas are working well:  ea22 is 
extremely weak on all IFs -- barely detectable fringes on the 2.75 Jy 
source.  Antenna 21 is also weak on all IFs -- the noise is about a 
factor 2 too high.  Antenna 20, LCP only, has noise about 3 to 4 times 
too high.  (Note:  these rms determinations are made following amplitude 
calibration). 

    Antenna 7 is special:  7C (LCP on IF#1) gave no fringes at all.  7B 
and 7D (IF #2) stopped fringing about 2/3 the way through the run, and 
did not return.  (Neither of these issues were flagged by the system). 

    After flagging these, regular calibration proceded.

    Amplitude stability is outstanding, strongly suggesting that the 
referenced pointing is working well.  (Note -- this is not proof!)

    Phase stability is good -- the 'jumps' between calibrator scans are 
well within those expected by weather -- they are identical in both IFs 
and polarizations for any given antenna, and grow with distance from the 
phase reference antenna. 

    Delays seem with acceptable boundaries -- the maximum w.r.t. antenna 
24 (by reference) is 15 nsec -- about two turns across the 128 MHz 
subband width.  However -- the two adjacent subbands give very different 
delay solutions, and the phase does *not connect* between the two 
subbands.  This is a totally different behavior than the wideband WIDAR0 
tests, where in all cases, phases connected smoothly across contiguous 
subbands. 

    Bandpass solutions generally look fine.  Some exceptions: 

    * 5D has a significant standing wave (approx. period of 25 Mhz). 
    *  9C has a resonance feature (in both amplitude and phase!) at 
33.48 GHz (most odd -- I've never seen anything like this before!).  The 
same feature is seen weakly in 9A, but is completely absent in 9B and 9D.
    *  An amplitude slope of a factor > 2 (4 dB in power) is seen across 
21B -- this slope is far larger than in any other antenna. 

    I have not yet had time to do imaging, etc. 


    *  A standing wave ripple of about 16 Mhz period is in 23B.