Alfa spider scans taken 29nov04
12dec04 (updated 06jan04)
links to plots:
results of the calibration runs (,ps) (.pdf)
beam maps for the 7 beams (.ps) (.pdf)
Spider scans (heiles
calibration scans) were taken with alfa on 29nov04 using the calibrator
B1756+134 (3C365, 2.3 Jy at 1420Mhz with 1.2 % polarization (from NVSS)).
The setup was 100 Mhz bandwidth centered at 1420 Mhz. Each strips took
60 seconds and covered 6 beam widths. An NVSS source search showed that
the closest source above 10 milliJy is 464 arc seconds so there was no
problem with confusion from other sources in the field.
13 patterns were done starting with pixel 0 and
continuing on thru pixel 6 and then pixel 0 thru pixel 6 (skipping pixel
1 the second time). It took about 43 minutes to do 7 pixels (6 minutes
per pattern). The telescope covered the za range of 16 to 7 degrees za
(50 minute before transit to 30 minutes after transit). For each
pattern, a pixel was centered on the source (using the built in CIMA offsets
for each beam) and then the spider scan was done. The analysis used the
cal values provided by ganesh back on 21apr04 (the newer cal values have
not yet been installed in idl).
The plots show the
results of the calibration runs (,ps) (.pdf).
Each pixel is coded with a different color.
Fig 1: gain, Tsys, SEFD, and averageBeamWidth. The gain and Tsys
are affected by the accuracy of the cal values used. The sefd only depends
on the source flux. The values change because the za/az is changing as
well as the motion from center pixel to outer pixel.
Fig 2: coma parameter, first sidelobe height, main beam efficiency,
and main beam efficiency + first sidelobe. The coma parameter measures
the asymmetry in the main beam. The beam efficiencies depend on the accuracy
of the cals.
Fig 3: The pointing error in azimuth and za. The plots have za error
vs az, za and azimuth error vs za ,az. The mean and rms errors were computed
over the 7 pixels.
Fig 4: The alfa rotation angle that minimizes the pointing error.
The array was set to 0 degrees in hardware. Offline I computed the offsets
for rotation angles -5. to +5 degrees in .1 degree steps. The measured
pointing errors for the first 7 measurements were then modified to reflect
this new setting. The rms errors in azimuth and za for the 7 pixels were
added in quadrature and then plotted versus rotation angle. The minimum
pointing error was found at a rotation angle of 1.25 degrees. This
means that when the users set the rotation to 0 degrees, it is actually
sitting at a rotation angle of 1.25 degrees.
The processing steps for the calibration are:
Beamaps were then made using the results of the
main beam and sidelobe fits. For the sidelobes, only the first 6 terms
of the fourier fit were used. This smoothes the sidelobe measurements a
plots show the beam maps for the 7 beams (.ps) (.pdf).
The data is contoured with 3db steps (below the peak) out to -21 db. Each
page has the beam maps for a particular beam. The central az, za position
for the beam map is printed on each plot (-az is before transit and +az
is after transit).It
The bottTop is pixel 0, bottom is pixel 1. Pixel 0 is symmetric with a
first sidelobe level of about -14 db. Pixel 1 has a coma lobe with a peak
of about -10 db.
Fit the main beam and sidelobes for each strip using 1d gaussians (allow
coma for the main beam).
Remove the sidelobe fit from each strip then do a 2d gaussian fit to the
model the 8 sidelobe measurements with a fft of 8 terms.
The central pixel 0 is coma free with sidelobes of -14
db. The outer pixels have coma lobes of -10 to -8 db. The coma lobe is
always opposite in direction to the feed offset on the array e.g.: pixel
2 is offset in the + azimuth direction on the array, the coma lobe is on
the -azimuth offset side of the beam . Some of the variation in the
sidelobe level for a particular beam can be explained by the az,za difference
Fig 1 pixel0: 2 beam maps taken at a za of 15.2 and 4.9 degrees.
The sidelobe level is -14 db. The map taken at za of 4.9 is a bit more
asymetric. This may be from the telescope blockage.
Fig 2 pixel1: Only 1 beam map was taken at a za=13.9 degrees.
The sidelobe level is -10db
Fig 3 pixel2: beammaps at a za of 12.4 and 5.5 degrees with a sidelobe
level of - db.
Fig 4 pixel3: Beammaps at 11.0 and 6.3 degrees za. Sidelobe levels
of -10 and -8 db.
Fig 5 pixel4: Beammaps at 9.6 and 11.4 degrees za. Sidelobe
level of -8 db.
Fig 6 pixel5: Beammaps at 8.5 and 8.7 degrees za. Sidelobe levels
of -8 and -10 db.
Fig 7 pixel6: Beammaps at 7.4 and 10.0 degrees za. Sidelobe levels
of -10 db.
This first set of observations was to get the analysis
software running. We now need to:
processing: x101/x102/newcal/alfa/29nov04.pro, 29nov04_beammaps.pro
Track one of the pointing model verification sources rise to set with pixel
0. This was never done after the pointing model was made since the calibration
scans were not yet working. This will give us the az, za offsets
for the alfa array to 1 or 2 arc seconds. These offsets should be similar
to the other receivers if alfa was placed correctly on the rotary floor.
Track pixel 0 rise to set on a polarized source to do the full stokes calibration.
Process the data with the old and new cal values. Compare the deltaG measurement
to see the consistency in the relative measurements of the polA, polB cal
Track a non zero pixel rise to set on a polarized source to do the full
Track a source rise to set stepping between the beams. Taking more than
1 cycle rapidly would allow you to interpolate between the pixel 0 measurements
for pointing error and maybe separate out the pointing model errors from
the beam offset model.
Track an outer pixel at 0 degrees rotation and look at the beam shape.
Then track the same source over the same az/za range with a different rotation
angle to see how sensitive the beam shape is to the rotation angle. Check
that beam 1 beam maps look like beam 2 beam maps when alfa is rotated by
60 degrees (putting beam 1 in the position of beam 2).