[evlatests] Ku-Band Behavior

[Rick Perley]

    I have completed the full calibration of the Ku-band data taken 
during the 30 hour 'flux densities' run. 

    The 'fringe failures' have been previously reported.  I list here my 
conclusions concerning amplitude and phase behavior. 

    There were 105 individual observations during this run.  The two IFs 
were tuned to 15.029 and 13.212 GHz (center) of the AC and BD IFs, 
respectively.  Referenced pointing was applied, using X-band observations. 

    Phase and amplitude calibration was done using the four 'standard 
objects':  3C48, 3C138, 3C147, and 3C286 -- there were 32 observations 
of these four objects. 
    There were eight antennas equipped and operating:  7, 10, 12, 14, 
20, 22, 24 and 26. 

    A)  Phase stability. 

    Phase stability within each scan was very good.  But the phase 
stability between subsequent observations was worse than poor -- there 
is no repeatable phase at all within either IF (or within any subband).  
As the weather was very good throughout, this instability is certainly 
not atmospheric. 
    The R-L phase is fairly good, but some antennas (notably 7, 12, 20 
and 24 show variability of 10 to 20 degrees (w.r.t. antenna 26, which I 
took at standard). 

    B) Amplitude stability. 

    All eight antennas are showing a large diurnal change in PDif, 
indicating that the amplifier gains and/or switched power source is 
sensitive to the 24-hour cycle -- likely temperature.  The amplitude 
change is shockingly large -- about 15%, with the minimum in the 
switched power (and hence in the amplifier gains) in the afternoon of 
both days.
    This large variation is not nearly as large in the visibility 
amplitudes -- which have been corrected by these switched power values 
-- so the evidence suggests it is the amplifiers which are most 
sensitive, not the switched power itself.  Some antennas -- notably 7 
and 12, show clear diurnal residuals, of ~2% -- either the noise diodes 
are variable, or the switched power measurements do not fully reflect 
the gain changes. 
    Good elevation gain functions were derived with ELINT for all 
subbands.  Some of the antennas give notably better fits than others:  
The best are:  10, 14, 24, and 26, where residuals are typically less 
than 1% following correction for elevation gain changes.  (Note:  An 
opacity model has been applied). 
    Antenna 7 in RCP in IF 'B' showed three different power levels 
(despite using 'set and remember'), amongst which the system cycled.  
These gain changes were nicely removed by the switched power system. 
    Overall, after applying both the switched power and elevation gains, 
the system stability for most of the antennas is stable to a pk-pk of 
about 2% in amplitude. 

    C) Polarization. 

    Wideband polarization was measured, with quite excellent results.  
Using antenna 26 as a reference, the 'differential D' terms are 
typically 2% across both IFs.  antenna ea07 is notably worse than any 
other -- with these differential polarizations 4 to 6 percent at the 
lower frequency tuning. 
    All D terms were seen to be high at the higher frequency tuning (AC 
IFs) -- likely because of the reference antenna used (26). 
    Eric's new software provides lovely plots of the antenna and source 
polarizations as a function of frequency -- the latter show the expected 
smooth changes in source polarization (amplitude and phase).