Alfa spider scans taken at 1665 Mhz
links to plots:
power versus sample along the strips.
100 Mhz filters centered at 1665 Mhz were installed
after the dewar on alfa. Heiles calibration scans (spider scans)
were then taken centering each alfa beam on the source 3C43 (the flux used
was 2.46 Jy at 1665 Mhz). 24 spider scan patterns were taken but only 13
of these were analyzed (the other 11 patterns spanned multiple files and
were not analyzed). An attempt had been made to measure
the alfa cal values at 1665 Mhz but the cal coupling into the horn
at 1665 is very poor. The program ended up using the cal values at 1520
Mhz so the gain and Tsys were not accurately measured.
Each pattern contains 4 strips of 6 beam widths
each (about 3.5 Amin per beam width). While tracking the source, the four
strips drove :
The drive time was 60 second per strip with 1 second sampling using
strip 1: -az to + az (no za offset)
strip 2: -za to +za (no az offset)
strip 3: (-az,+za) to (+az,-za)
strip 4: (-az,-za) to (+az,+za)
The first set of plots shows the total
power versus sample along the strips. Each page is a separate pixel
of the array. Black is polA and red is polB. Dashed green lines separate
the four strips. Each division on the x axis is 3.5 arcminutes (about 1
The sidelobes for some of the pixels are large (-6 db).
This may in part be do to the pointing error that did not take us through
the source peak. If you believe the pointing errors from the fits, then
a 56 asec error gives a 1 db loss. You can also clearly see the coma lobes
in most pixels.
Fig 1 thru 7 Top plot: The 4 strips (separated by dashed
green lines) are plotted versus sample using a linear power scale.
The dashed black and red lines are the baselines that were used to compute
the Db plots on the bottom of each page. There are two sets of black, red
lines since each pixel was measured twice (except for pixel 5 which was
only measured once). Some of the pixels (3,4,6) have offsets between
the two measurements (mainly for the first strip). This may be a real pointing
error, but I think it is because of the uncertainty in the synchronization
of the antenna drive and the data sampling.
Fig 1 thru 7 bottom plot: The bottom plot y axis has power
measured in db below the peak value for each strip (note that the strips
do not necessarily go through the source peak). To remove the system
temperature, a linear fit was done using the 4 lowest points on each side
of the middle (the dashed lines on the top plot are the baselines used).
The second set of plots show
the system performance. They were generated after passing the measured
data thru the heiles calibration software. This fits individual gaussians
for each strip (the main beam and two sidelobes). It then does a 2d fit
for the beam. The values for each pixel are plotted in a separate color.
Fig 1 top plot: This is the SEFD measured for each beam using 2.46
Jy as the source flux.
Fig 1 2nd plot: the average beam width in arc seconds for each measurement.
Fig 1 3rd plot: The coma parameter for each pattern. The coma parameter
is defined as : exp((x-x0)^2*(1-coma)/sigma^2). I think it is constrained
be by less than .2 .
Fig 1 bottom plot: The average sidelobe height. This value is 3
to 6 db less than the single strip plots. The pointing error (next plot)
would account for about a 1 db difference. It is also caused because this
is the average sidelobe height. The large bumps that we seen in some of
the total power strips are only on 1 side (coma) and not necessarily in
the azimuth and za directions.
Fig 2: This is the azimuth (top) and zenith angle (bottom) pointing
error in arcseconds derived from the fits. It assumes that the az, za positions
that are used for the strip positions are correct. Most pixels have an
azimuth and za error of about 40 Asecs each. Pixel 4 differs with a zenith
angle pointing error of -15 and 5 asecs. One of these patterns had lots
of interference but the 2nd pattern was relatively clean.
processing: usr/a1800/ohcalib/alfaspiderOhtp_jul04.pro, alfa.pro
The sefd at 1665 is 4.5 to 6.5 Jy (this depends on the accuracy of the
flux for 3C43).
The individual strip plots show sidelobes as large as -6 Db with a large
coma lobe. The average sidelobe fits are -15 to -10db below the peak.
The fitted pointing errors are about 40 asecs in both az an za.
These offsets may be from real pointing errors, or from inaccuracies in
aligning the az, za offsets of the strips with the data samples. The jumps
(in az) between the two measurements of the same horn are probably alignment