[evlatests] Results from ACU tests using OTF modes
Rick Perley
rperley at nrao.edu
Wed Mar 4 14:16:40 EST 2015
On Feb 19 I ran an extensive set of ACU tests, using the nifty OTF
modes. These tests were intended to determine of the antennas scan at
the same rate.
This is a lengthy report (apologies to the impatient). The
appropriate figures to explain and support the results will be shown at
next week's test meeting.
The test was set up this way:
1) Scans through a calibrator source were made driving at 4, 8 and
12 times 'sidereal' -- meaning, 1 deg/min, 2 deg/min, and 3 deg/min,
respectively.
2) Scans were arranged to be 2, 6, and 11 degrees in length.
3) Scan start/stop points were arranged such that the scan ended
one degree beyond the calibrator. Hence, the scans started 1, 5, and
10 degrees before the calibrator.
4) Scans were done in both RA and Dec directions. All RA scans
went from East to West. All Dec scans went from North to South. There
are thus nine RA scans (three lengths times three speeds) and nine Dec
scans.
5) The target object was J0121+0422 -- chosen for near zero
declination, and such that the observations were taken near zero hour
angle. Hence, the RA scans were almost purely in azimuth, the Dec scans
were almost purely in elevation.
6) The observations were taken at X-band, using two IFs: 8.3 and
11.3 GHz. Note that at this frequency, the FWHP is about 5.3
arcminutes, and 3.9 arcminutes. With the selected scanning speeds, the
higher frequency beam is traversed in 3.9, 1.95, and 1.3 seconds.
7) The scan mode utilized 1 second delay/phase steps, and 0.1
second (10 Hz) dumps.
8) X-band referenced pointing was done at the beginning, and
applied to all subsequent observations.
9) A regular calibration observations was done at the beginning and
end.
Data calibration/analysis proceeded thusly:
1) Delay and phase calibration was done using the two calibrators
scans.
2) Bandpass was next done, using the two calibrator scans.
3) Amplitude/phase calibration (using unit flux density) was done.
4) The data corresponding to the actual transits were then
extracted, with the calibration applied. (With the beam being transited
in 2 -- 5 seconds, the vast majority of this test comprised empty sky).
There were then 18 databases, each corresponding to a single transit.
5) FRING was then run again, using 0.1 second averaging (i.e., each
and every data point). The results were applied, with two-point
interpolation. (This step is needed to prevent delay smearing of the
source -- with such a fast transit and short integration, the SNR
corresponding to a single channel is not good enough).
6) CALIB was then run, with 0.1 second averaging. The amplitude
solutions are what we are after, then give the amplitude of the beam
pattern as it crosses the source.
7) The transit time was solved for using a special program, SNFIT,
written by Eric Greisen for this purpose. This program fits a
polynomial (a quadratic in this case) to the amplitude solutions, and
solves for the transit time and beam transit width.
I'll point out here that the 1 second step is too fast (I wasn't
aware of this limitation when the run was performed). Curious things
happened in the first few records (up to 4 of them) for each new
delay/phase step. Investigation of this uncovered an old 'feature' in
AIPS which prevented accurate calibration of data taken this fast --
this has now been corrected. The 1 second limitation is an artificial
one -- we'll likely test tomorrow just how fast we can step.
The accuracy of the beam transit time is very high -- typically 0.1
seconds, determined by comparing the results for the two frequencies and
two polarizations available in each of the 18 transits.
I extracted from the solutions all 26 x 18 x 4 x 2 = 3744 transit
times into a spreadsheet for analysis. The mean time of antenna passage
was determined from the mean for each of the 18 separate transits, and
the deviations computed. Some interesting results emerged.
1) Most antennas behave remarkably well - 18 of the 26 antennas in
the array transited with 0.1 seconds of the mean in RA, and 15 antennas
did this in Dec. The dispersion within the nine transits for these
antennas is typically 0.05 seconds. This means that these antennas all
behave very similarly. There was no dependence on either the antenna
speed, or length of the travel.
2) The remaining antennas (8 in RA and 13 in Dec) show much larger
dispersion. All of these are consistently 'early' or 'late', by up to
0.5 seconds (which translates to 0.5 to 1.5 arcminutes). There is again
no obvious dependence of the time crossing with the speed of the antenna
or the length of the scan. It is consistent with the antenna starting
its travel either too soon, or too late. But the 'lateness' or
'earliness' is often not consistent. Antennas usually 'early' will
occasionally be 'late'. There are no patterns or dependencies I can see.
3) There is no correlation between 'lateness' or 'earliness' with
the offset pointing solutions. (The theory being these may not have
been applied). In any event, a real pointing offset would have shown a
simple relationship with beam crossing times, which is not there.
4) Some antennas are consistently early, or consistently late.
Here's a short list of the more egregious:
A) In Azimuth:
ea01: Early by 0.3 seconds
ea02: Late by 0.2
ea14: Late by 0.42. (Very consistent -- dispersion was 0.12
seconds)
ea21: Late by 0.33
ea27: Early by 0.19. It was early in 7 of 9 scans, but
equally early in the other two.
ea28: Early by 0.38.
B) In Elevation:
ea02: Early by 0.26 seconds
ea07 Early by 0.25 sec
ea12: Early by 0.34 seconds
ea26: Late by 0.33 seconds.
For reference: Here is a table of the averaged (over the nine
transits) time differences for each antenna from the mean time of
transit, and the dispersion for each mean.
Ant Az Time (sec) Disp (sec) El Time (s) Disp (s)
---------------------------------------------------------------------------------------
01 -.31 .16 -.01 .13
02 .21 .18 -.26 .15
03 .05 .07 -.08 .04
05 .03 .05 .05
.06
06 .00 .05 .14
.19
07 -.13 .21 -.25 .20
08 -.02 .07 .20 .13
09 .08 .07 -.12 .07
10 .09 .06 .00 .06
11 .05 .03 .21 .11
12 -.05 .15 -.34 .18
13 .05 .05 .05 .06
14 .42 .12 .02 .07 New ACU
15 .05 .05 .07 .08
16 -.03 .08 -.04 .04
17 .00 .08 .03 .03
18 -.06 .06 -.22 .12
20 .01 .15 -.16 .11
21 .33 .05 .05 .07 New ACU
22 -.05 .09 .05 .10
23 -.09 .10 -.02 .05
24 -.01 .05 .06 .05
25 .07 .06 -.04 .09
26 .04 .08 .33 .15
27 -.19 .24 .04 .05
28 -.38 .18 .22 .23
---------------------------------------------------------------------------------
More information about the evlatests
mailing list