Wide band look at AC birdies in 327 rcvr.
spectra while the old AC unit was turned off and then back on (.gif)
spectra of ACon - ACoff across 25 Mhz (.ps) (.pdf).
On 20oct05 the old AC unit in the dome was powered
on off while data was taken with the interrim correlator. The datataking
setup was 25 Mhz over 2048 channels and hanning smoothing. This gave a
channel width of 25 Khz. Previous measurements showed that the birdie was
not resolved in a 95 hz channel so the 25Khz channel width is not a very
good match. Data was dumped once a second for 300 seconds. The telescope
was sitting still.
The data analysis consisted of:
Compute the median bandpass for the 300 seconds. Using corblauto() fit
a 9th order harmonic function to the median bandpass. Divide each spectra
by the fit. Birdies that are constant will not be cancelled
by this bandpass correction.
Flatten the image by averaging over 50 freq channels that are clean
and then divide the bandpass of each second by its value.
Compute an average spectra while the AC was on and and average when the
Ac was off. Use the bandpass corrected, flattened spectrum to do this.
The dynamic spectra image (pol A) shows where
Old AC unit was turned off and then back on (.gif) :
Average spectra were also made for the AC on and AC
off records. The difference of these averages shows the
AC comb across 20 Mhz (.ps) (.pdf).
The power to the old AC unit was turned off at record 147 and back on at
record 240. The horizontal dotted lines show where this occurred. You can
see that the birdie went away during the off period at: 310.344,
313.345, 315.346, 318.347, 327.351, and 328.351
Mhz. These are all separated by a multiples of 1 Mhz.
When the old AC was turned off, the new AC was still on. It also has a
set of 1 Mhz birdies that are shifted from these.
Previous measurements showed that the comb elements were not resolved in
a 95 hz channel width. This image uses 25 Khz resolution (and only
1 second integrations) so some of the weaker comb elements will not be
The data shows that the AC unit extends across the entire band. The strength
varies between polA and polB. The 2nd Ac unit is creating a similar comb
(although offset in frequency).
The data was averaged for 50 seconds AC on and 85 seconds AC off.
The Black plot is polA and the red plot is polB. The vertical scale is
Tsys. PolB has been offset for display. The dotted green lines are spaced
by 1 Mhz after aligning with the strongest birdie.
The strongest birdie is in polB at 331.35 Mhz. It is about .15 Tsys in
a channel width of 25 Khz. The 95 hz rbw data taken on a previous day showed
the signal to be about 8 * the 95 Hz channel power.
More info can be found at (AC
units in the dome create rfi).