[evlatests] Backend and Frontend Tsys

[Rick Perley]

    This is not an EVLA issue, per se, but it may reveal some related 
problems.

    I had a semi-successful test run last Thursday night -- other things 
learned
or seen will be reported later.  Because sensitivity is the front-burner 
issue,
I decided to look into the reported values for antenna Tsys.   For the VLA,
the default values for Tsys which are read by AIPS are the 'front-end' 
values.
But since these are not calculated for EVLA antennas, we must look at
the 'back-end' values instead.

    Hence I created two AIPS databases from the Thursday data, one with
front-end, the other with back-end, values for Tsys.  It then occured to
me (with encouragement from Ken) to compare these values for VLA antennas. 
We had expected the values to be similar. 

    They are not!

    I looked at both values for Tsys, for
all VLA antennas in IF `A', at L, C, X, and K bands.  I formed the
ratio:  Tsys(backend)/Tsys(frontend) for each antenna, at each band,
for IF 'A'. 

    All 104 ratios are positive, varying from 1.10 to 1.70.  If we believe
these values, then the system temperature, as seen at the backend, is
up to 70% worse than at the front end. 

    The `degradation' factor is highly correlated, in the sense that if
the ratio is high at one band, it is high in all bands for that antenna-IF.
For example, antenna/IF 20A has ratios 1.46, 1.36, 1.33, and 1.35 at
L, C, X, and K bands, while antenna/IF 17A is 1.07, 1.11, 1.09, and 1.13
for those bands.  This type or relation holds for all antennas.  (But I 
didn't look at the
other IFs, other than IF C at L-band, where 25 of 26 ratios are positive). 

    It seems unlikely that the signal is degraded by such a degree (Ken
recalls Durga finding ratios of order 1.01 or so), and more likely that our
calculation mechanism is faulty.  One possibility is an offset in the
synchronous detector voltage -- but is it likely they are all off in the
same direction?  And in any event, these offsets are supposed to be
monitored and corrected for regularly by operations. 

Is is possible the electronic gain factor (15 for front-end Tsys, but 
apparently
a different value for the back-end system) is offset by 10 to 30%?   The
offset would have to be all in the same direction,and of different values. 
Or perhaps the Vtp (total power monitor voltage) is offset from the
assumed values (3.0 for front-end, something else for backend)? 

    Rick

    Rick