[evlatests] L-Band Amplitude Behavior

[Rick Perley]

    The phase jump test yesterday has been analyzed for amplitude 
behavior.  There are a number of minor issues to report:

    1) The test consisted of two scripts, back-to-back.  The first 
followed the holography script, so we know the EVLA and VLA antennas 
were on the same wrap, and so the antennas on both arrays were pointed 
more or less in the same direction at the beginning of each half of the 
phase jump test.         It was noted that for the first scan of each 
half of the test, the EVLA antennas started fringing 10 seconds later 
than the VLA antennas.  (A minor issue, to be sure!) 

    2) Close inspection of the visibility amplitudes show that, for 
nearly all EVLA antennas, there is a (generally) small (maximum of 10%, 
more generally a couple %), but discernible ~20-second settling time at 
the beginning of every scan.  None of the VLA antennas demonstrate 
this.  The origin of this effect is very clear:  The 'settling' is 
reflected perfectly in the system temperature profile, and is thus 
introduced by the post-correlation correction of the correlation 
coefficients for the change in system temperature which occurs when a 
frequency change is made. 
       The VLA is absent this effect because the Tsys values are 
determined at the front-end, and are done with a faster time constant.  
EVLA antennas must use 'back-end' values, which are smoothed (because of 
a potentially narrower bandwidth, thus reducing the SNR of the 
determination) to ~10 seconds. 
       It is suggested here that we shorten the smoothing time of the 
back-end Tsys calculation.  Visual inspection of the Tsys values for 
this experiment -- done with a narrow 6.25 MHz BW-- indicate the curves 
are very smooth, and hence have very good SNR.   Except for the very 
narrowest bandwidths, we should be able to stand a shorter time 
constant.  (Can the smoothing time be made a function of BW?) 
    This is a transition issue only.    For the final EVLA, the 
synchronous cals will be detected in the correlator, with (generally) 
lots of BW, permitting a suitably short time constant for correction of 
the correlation coefficient. 

    3) One antenna -- number 24 -- has peculiar amplitude  jumps, whose 
nature changes depending on the frequencies employed in the experiment:
       - In the first half of the test, where the frequency alternation 
was between 1465 and 1445 for IF1, and between 1385 and 1405 of IF2,  
low amplitudes were commonly seen on single 3-second records on IF1, and 
rarely on IF2.  The drop was large on IF1 -- by about 40%, and small on 
IF2 -- by about 5%.   The opposite polarizations behaved identically.  
All drops were 'on 10 second ticks',  (NB  my averaging time was 3.3 
seconds), but not alll records on '10-second ticks' were affected. 
       - In the second half of the test, where the frequency alternation 
was between 1465 and 1450 MHz for IF1, and between 1385 and 1390 MHz for 
IF2, the nature of the drops changed dramatically!  For IF1, all the 
drops were 'doubled-drops', and all occurred on consecutive records, 
always 'on the 7s' and 'on the 10s'.  The initial drop (on the 7s) was 
the larger (of amplitude 30%), the 10s drop much smaller (~10%).   As 
before, the drops did not occur on all '7' records.  For IF2, the 
frequency and size of the drops was about the same as in the first half, 
but all drops occured 'on the 7s' rather than 'on the 10s'.