Compare Turret scans with heiles calibration
Turret scans move the turret back and forth in
a sine wave while the dome moves +/- 5 arc minutes. Total power data is
sampled at 100 hz in pol A and pol B. A baseline is removed and then a
2-d gaussian is fit to the amplitude, widths, and offsets.
The heiles calibration uses 4 strips offset by 45
degrees from each other. The correlator records the auto and cross correlations
every second. A 2-D gaussian is fit to stokes I. The fit parameters are
amp, offsets, side lobes, and coma.
The turret scan samples much faster but has the
drawback that the gain and tsys change as you move the feed off of the
optical axis. There is also no side lobe or coma fit (coma could be added
but side lobes would be difficult).
On 21mar01 CTA21 was tracked with sband narrow transit
to set using turret scans to verify the pointing model (during the day).
On 04apr01 this was repeated using heile's calibration method for
The black and red are turret scans polA and polB while the green is the
stokes I calibration. The 12 samples for the calibration method had
32 turret scan samples. The pointing errors differ. Part of thist
could be from the different fitting methods. The calibration scans fit
for sidelobes and coma while the turret scans just fit for the main beam.
The coma (especially at higher za where the pitch,roll,and focus errors
are larger) could be biasing the turret fits.
Pointing error in azimuth
Pointing error in zenith angle
Gain/Tsys as a percentage
beam width in azimuth
beam width in zenith angle
platform corner height - average height while tracking.
Since the runs were done on different days, it is
also possible that the platform angle was different on the two days. This
could happen on a hot day if we lost tension in one of the tiedown cables.
I checked the tiedown cable tensions and tiedown 4 tension was small but
it did not go to zero. The last plot of figure 2 shows the corner
heights of the platform for the two days while tracking the source. There
is a large difference in corners 4 and 8. I aligned the data by looking
at the laser ranging point where the za was 5 degrees for each day. The
alignment error could be because the calibration and turret scans do different
za motions (but i'm a little suspicious).
The Gain/Tsys for polB and stokes I are identical.
Pol A G/t is less. The tsys of polA, polB differ by about 5% so the stokes
I G/T is still about 5 % higher than the turret scans.
The za beam widths (middle plot figure 2) for the
calib runs and the turret scans agree. You can see the beam width increase
as the beam starts to spill over above 15 degrees za.
The az beam widths (top plot figure 2) do not
agree. The calib beam width is 10% smaller than the turret scan beam width.
The turret scan beam width in az is wrong since it should be smaller than
the za beam width (the az direction illumination is larger than the za
direction illumination). The turret scan is moving the horn away
from the paraxial ray as it scans the turret back and forth. The gain falls
off because of the telescope beam as well as the field of view at the horn.
When the beam width is computed, the plate scale at the horn is used to
convert turret degree offsets into arcseconds on the sky. The value -45
arcseconds per turret degree was used. If this value was off by 10%, the
two methods would agree.
(Note: The az,za errors were wrong until 06jan02 when i realized that
the definition of the error was different for the turret scans and the
calibration scans: (measured-computed) and (computed-measured). Putting
in the factor of -1 really helped the agreement!!!)