[evlatests] Test of UX Alternate Path

[Rick Perley]

    Data were taken yesterday to test the alternate path of the UX 
converter.

    Summary:

    It looks good.  The sensitivity for data routed through the 
alternate path is the same as through the straight through path. 

    Details:

    The array was commanded to observe alternate in two different states:

    A) At a 'standard' frequency of 22.416 GHz (for both IF pairs), in 
which the EVLA data were routed through the UX converter's straight 
through path.

    B) At a frequency pair of 25.816 GHz and 22.416 GHz, for the AC and 
BD IFs, in which the BD IFs were routed through the alternate path.  As 
this path can be commanded only for antennas 18 and 24, the data for 13, 
14, and 16 were no good. 
Note that the BD frequencies were the same in both states, to allow 
direct comparison. 
    Ken modified the jobserve scripts to enable the alternate path.  He 
also modified them such that while testing the alternate path (IFs B and 
D), the A and C IFs were set to 25.816 GHz, in order to get around some 
`features' of the EVLA's LO tuning.  (You'll have to ask Ken about this, 
if you wonder what this means.  Be prepared to stand at a blackboard 
...)  The interesting result of this is that all the VLA-VLA baselines 
were at 22.816 GHz, while the EVLA-EVLA baselines were at 25.816 GHz, 
and of course all VLA-EVLA baselines were only  noise. 

    Observations were made in spectral line mode, 12.5 MHz BW, 16 
channels for each of the four parallel hand correlations.  Averaging was 
3.33 seconds.   Observations were alternated, 5 minutes each, on 3C273 
and on cold sky 8 degrees to the north. 
    Bandpass calibration was done by standard means.    A 'channel 0' 
pseudo-continuum database was then constructed, using only the central 
channels (4 through 13) which did not have Gibb's ringing or unusual 
phase 'hooks'.  (More on this later). 
    Amplitude calibration was done using the calibrator 3C273, for which 
25 Jy flux density was assumed.  (Close enough for this study). 
   
    The sensitivities were estimated using the well-tested 'AIPS 
weights', after calibration was completed. 

    Results:

    1) As described above, the comparison of the 'alternate' to the 
'straight through' path can be done only in the B and D IFs, for 
antennas 18 and 24.  For these:
       - The AIPS weights are the same for both paths, for all four 
IFs.  From this, we conclude the alternate path does not degrade the 
sensitivity. 
       - As an aside -- I'll repeat the conclusions reached in my prior 
email:  Antenna 18's sensitivity is a little below par.  Antenna 24, in 
IFs B and D, looks quite fine. 

     2)  The observations at 25.816 GHz, in IFs A and C, taken by the 
EVLA antennas only when in the 'alternate path' mode, gave two 
remarkable results:
          a)  The sensitivity of these channels was *far* better -- by a 
factor of more than 2 in AIPS weights, or about 1.5 in real sensitivity 
-- than the VLA's.  Almost certainly, this is a result of being well off 
the H2O line, which in our current weather conditions must be adding at 
least 20K to the system temperature.  I tried to check this from the 
Tsys values -- but these are all zero, except for antenna 24, where the 
values are stuck at 2.4K.    Why are we getting no Tsys?   (I filled 
with 'backend' values). 
         b) The bandpasses for this frequency are absent the 'Gibbs' 
ringing so clearly evident at the other frequencies.  The ringing is due 
to a sharp filter cutoff -- not the VLA's backend filter.  Why did 
tuning the frequency up by 3 GHz remove the cutoff? 

   

    3)