[evlatests] Encouraging Results on P-band Linearity

[Rick Perley]

    Tests taken with our new P-band system, late last September, 
demonstrated that the system was being set up at far too high a power 
level for linearity to be retained between observations of a standard 
calibrator and Cygnus A.  Cygnus multiples the total power by a factor 
of 4 to 5 at this band. 

    On Oct 2, a small script was generated, which observed only the pair 
3C405 (Cygnus A) and 3C380, one of the 'Scaife&Heald) standard 
calibrators, whose flux density is reasonably well known in this 
frequency zone.  Ken then modified the script to set both the T304 and 
the requantizers.  The test comprised three parts:

    1) Set the power levels up on Cygnus A, and observe 3C380.
    2) The reverse of the above. 
    3) Set requantizers separately for each. 

    The last test did not work as expected.  The first two worked well. 

    Results:

    It was immediately apparent that after Ken's setting of the power 
levels (including the requantizers), all the antennas had exactly the 
expected power levels as seen in the 'PSum' monitor.  This is in stark 
contrast to the September experiment, where the range of powers was 
close to an order of magnitude!  
    It was also immediately clear that the 'PDif' compression problem is 
entirely independent of whether the power levels were in the low (set on 
Cygnus A) or high (set on 3C380) state.  And indeed, even for the 
September experiments, where power levels varied by well over an order 
of magnitude, the 'PDif' compressions were unaffected. 

    All the above was reported earlier by me, prior to my European 
trip.  I hadn't had the time then to see if the derived flux ratios 
(3C380/3C405) were as expected.  I have now done the proper calibration, 
and the results are very encouraging. 

    I used only the 'low power' data (those from setting the power 
levels on Cygnus A).  3C380 was used as the calibrator, with the fluxes 
from the Scaife and Heald paper.  I made images of 3C380 (it is slightly 
resolved, with background objects, so a decent full-field image was made 
at each subband), and determined the total flux using 'IMEAN'.  I did 
the same for Cygnus A, and found its total flux via image integration.  
I then determined the flux density of 3C380 by using the Baars et al. 
expression for Cygnus A.  The results are in the table below

Frequency        S(3C380)           S(3C380)
    MHz               S&H               Baars et al. via Cygnus
--------------------------------------------------------
   224                 56.9                     57.3
   239                 54.1                     53.6
   260                 50.7                     52.2 **
   274                 48.7                     48.4
   288                 46.9                     46.3
   304                 45.0                     44.7
   320                 43.3                     42.1
   336                 41.7                     41.3
   352                 40.2                     39.3
   368                 38.9                     40.1 **
   384                 37.6                     37.3
   400                 36.5                     37.9 **
   416                 35.4                     34.6
   432                 34.4                     33.4
   447                 33.5                     32.9
   464                 32.5                     31.6
-----------------------------------------------------

The values with ** marks subbands with clear errors in the ratio -- 
mostly due to RFI issues. 

The agreement between S&H and the 'Baars et al.' values is very good,  
especially in the central subbands, with deviations less than a couple 
percent. 

    The next step is to repeat this experiment, using a few other 
objects.  A script containing 3C286, 3C295 and 3C380, along with 3C405, 
has been prepared. 

    Note:  The ratio between 3C405 and 3C380 is substantial -- slightly 
more than 140.