[evlatests] EVLA tests, Sept 20 2005

[Ken Sowinski]

                    EVLA Tests, September 20, 2005

Rick and Bob are using antenna 14 to try to get a good measurement
of efficiency and T_sys at (at least) C band. 

10 dB pads were installed between each downconverter and its D rack
so that we could turn up GAIN2 in the hope of suppressing the birdies
coming from the switching power regulators.  After that was done and 
antenna 16 was available Barry proved the software which adjusts GAIN2
by root(10) to accomodate the pads.  Mike has supplied filter coefficients
for most of the subbands.  We elected to try the subband centered at
1216 MHz (subband 6 in Mike's notation).  It appears both the loading
of the correct filter taps and setting the L302 to compensate all works.
This allows us to tune 512 MHz lower in X band than we could before so
that sky frequencies as low as 7940 MHz should work.  The remainder of 
the tests described were carried out using subband 6.

An attempt was made to see how well 16 works at all bands by setting the 
T_sys correction to unity so that amplitudes are just scaled correlation 
coefficients.  Small delay changes (as much as 5 ns) were needed for most 
IFs at all bands.  The first step was to verify sampler RMS which gives 
the best response.  This was measured using 16B at C band.

Output        Sampler    Amplitude
attenuator      RMS
20 dB          19.5       2.23
23             14.1       4.21
24             12.5       4.9
25             10.7       5.2
26              9.7       5.31
27              8.7       5.3
29              6.8       5.30
31              5.3       5.3

It is clear that the optimal power level drives the sampler to an RMS
of 5 to 10 digitizer units and response falls off starting at ten.  The
attenuator does not go beyond 31 dB and I did not try adding attenuation
at the input to the downconverter so do not know how far less than 5 
we can go before repsonse degrades.  There is at least a 6 dB range in
power that provides maximum correlation coefficient.  All the succeeding
observations were made with the sampler RMS between 7 and 10 for all IFs.

Having done this I looked at focus and discovered that C band focus was
wrong by about 2.5 cm.  similarly K band focus was wrong by about 2 cm
and  pointing was off by more than an arc minute in elevation at K band.
After correcting for all this we have the following, all responses are
expressed as power.
    ant 16      VLA Ants      Ratio (VLA/EVLA)
L   8 +/- 1       ~20          2.5
C   2 +/- 0.2    12 - 16       7
X   9.2 +/- 1    12 - 14       1.4
K   1.6 +/- 0.1    ~3          1.9

The uncertainty in the EVLA result only indicated the range over all
IFs.  The wind was at 5 to 10 m/s and referenced pointing was not used 
at K band so the VLA K band response might well be too low.  There was
still an elevation error for 16 of a few tenths of an arc-minute so
that response ought to be larger as well.  L band focus was not checked.
What stands out is that C band is so band and K band is surprisingly
good.  This makes it hard to implicate LO purity, but suggests the 
C band receiver is not as cold as we think or there is something ill-chosen
about the LO chain for C band.