[evlatests] OTF/ACU test at X-band

[Rick Perley]

     Another test done yesterday afternoon strongly indicates that 
pointing residuals are the root cause of the few hundred millisecond 
timing discrepancies noted in the February long-slew OTF/ACU tests.

     This latest test had the antennas cross the main beam, back and 
forth, in RA and in Declination, at X-band.  Referenced pointing was 
done and applied.  Three of the four cuts were arranged to start and 
stop at the 2nd null -- this to ensure the antennas were at the right 
speed when crossing the main beam:  At the 4 x Sidereal rate, it takes 
only 5 seconds to cross the main beam.  Phase/delay steps were made 
every 1.2 seconds (this to test whether this fast a rate is stable).  
The dump rate was 10 Hz.  Due to an error on my part, the last cut 
(declination, south to north) started and stopped at the half power points.
     The source used was 3C84, which was at HA = -1, with a parallactic 
angle of about 120 degrees -- hence the 'RA' cut is mostly along 
elevation, the 'dec' cut mostly along azimuth.

     Results:

     1) A 1.2 second phase/delay rate is not fully stable.  Perhaps 10% 
of the initial phases were wrong after the step.  Bruce has apparently 
arranged to removed the 1 second limitation within the station boards -- 
I'd like to test this next week.

     2) The timing residuals for the two RA cuts (back and forth) are 
*very small*, with a dispersion of 0.07 seconds about the mean. This 
corresponds to 4.0 arcseconds at the rate selected -- this is completely 
consistent with residual pointing errors, following referenced pointing.

     3) Antennas with larger residuals always had opposite signs, and 
nearly equal magnitudes, for the two RA cuts.  The most discrepant are 
listed below:

      ea05   +.07, -.08 (seconds)
      ea07   +.09, -.06
      ea16   +.08, -.12
      ea17     -.12, -.05   (some sort of offset crept in ...)
      ea21     -.19, +.16   by far the largest error
      ea23     -.05,  +.07
      ea25     -.14,  +.13
     I checked the pointing solutions -- all of these antennas had good 
solutions.  Note that, due to the parallactic angle, the axes along with 
the pointing is determined (az and el) are not the same as those along 
which the timing is determined (RA and Dec).

     4)  The two declination cuts were taken with different rates (4X 
and 1X 'sidereal').  Furthermore the latter of these (S to N) started 
the movement only 1.2 seconds in advance of the first scan. Hence, this 
last beam cut is dominated with antenna oscillatory behavior.  However, 
the N to S cut was done correctly (i.e., from 2nd null to 2 null), and 
the timing offsets are again small, but about twice the value of the RA 
cut residuals.  Note that this cut direction was primarily in the 
azimuth direction.

     5) The cuts from 2nd null to 2nd null clearly show we have optics 
misalignment.  The difference in height of the first sidelobes on either 
side of the main beam is a sensitive indicator of subreflector 
misalignment.  The typical difference between these is currently 3 to 5 
dB (!).  Some are as large as 8 dB.

     6) ea21 (new ACU) showed long-lasting oscillations in the 
amplitudes for the 'RA' cuts, easily visible 8 seconds into the cut.  
The period is about 1 second.

      Conclusion:

     These 'short-throw' tests strongly indicate that the residual 
timings for the beam crossing a source are due to residual pointing 
errors.  There is no evidence for an offset in the time, or rate of 
motion, amongst the antennas.
     Still unexplained are the much larger offsets from the 'long-throw' 
tests done in February.  These were done in only one direction -- so no 
way to separate pointing residuals from timing offsets.  The code has 
been changed since these tests were done, so it seems another long-throw 
test will be needed, with opposing directions.

     Next week.