Alfa sky, absorber cal measurement at 1665 Mhz
100 Mhz filters centered at 1665 Mhz were installed
after the dewar on alfa. The cal values were measured by using absorber
and sky as the hot and cold load. 5 second cal on followed by 5 second
cal off was repeated 3 times at each position. Since the absorber did not
cover the entire array, 4 separate positioning of the absorber were done
to cover all of the horns.
After looking at the data, it was obvious that the
cal was not coupling very well into the system. The cal is being
injected before the omt. Instead of computing the cals from the Y factor
(since the coupling was so poor) i measured the cals as a ratio of the
absorber temperature and sky temperature using the following parameters:
|scattered ground radiation
The absorber temperature was measured with a thermometer, the scattered
radiation is a value that has been used in other receivers.
The 4 absorber positioning covered the following horns
absorber position #
show the results (ps) (pdf)
the measurements. The numbering of the plots has board 1 thru board 7 (which
is pixel 0 thru 6).
Fig 1. The band passes for absorber position 1. polA (blue CalOn,
red calOff). polB (purple calOn, green calOff). You can see the rfi
leaking through the absorber.
Fig 2. CalOn - calOff in correlator units. polA ( red abs, blue sky)
polB (green Abs, purple sky). The scales for the absorber and the sky were
not the same. Unity was normally close to the off cal system temperature.
You can see the large variation in the cal power getting into the system.
Fig 3-6. These are the cal Values in Kelvin computed using the 4 different
absorber positions. The computation was (caldeflection/TsysCalOff) *TsysCalOffInKelvins.
The Cal measured from the sky temperature was over plotted in each. The
shape of the cal curve is similar for both the absorber and the sky. Setting
the TsysSky to be 38 K overlaid the two lines (since the match is so bad
this is probably pointless...).
The cal varies by a large amount across the band. If the cal
is being injected at a null in the wave guide then the variation across
the band could come from the changing wavelength. The band covers +/- 3%
in lambda (+/- 10deg in phase). If the distance from the injection point
to the probes is about a meter then this is 1/.18= 5 lambda's. So we might
accumulate 50 degrees of phase across the entire band. Probably not enough
to cause the large variation we see in 20 Mhz.
The probes are not at the same spot so the pattern
should be shifted for the two polarizations.