Compressor monitor system (CMS)
06jun11: testing the compressor monitor system on
14apr11: testing the compressor monitor system in
06jun11: on telescope testing of CMS. (top)
The 327 vlbi run on 05jun11 reported a 1 Mhz
comb in the data. On 06jun11 we did some tests to see if
the problem was coming from the compressor monitor system (which
had been turned on during the 05jun11).
The setup was:
Processing the data:
- 327 receiver, center freq 327 Mhz, 302=352 filter installed
- mock spectrometer: 43 Mhz bandwidth, 8192 channels (5Khz
resolution), cfr 327 Mhz, 1 second dumps.
- Since the compressor monitor is on the dome, the az,za position
should not make much difference.
- Data was taken for about 750 seconds.
- Around second 400 to 500 seconds, willie iguina openned the
huffman box of the compressor monitor to get access to turn off the
- At 542 seconds the compressor monitor system was turned off.
- a bandpass to flatten the spectrum was computed by:
- average the 750 seconds of data
- do a robust fit (throwing out outliers) using a first order
polynomial and a 43 order harmonic function.
- divide each of the 1 second integrations by the bandpass
generated by the fitting (then remove 1.)
- For each 1 second record remove the median value (to get rid of
continuum source drifting through the beam).
- I used a fit for the bandpass correction instead of a median
bandpass so that contstant rfi would not get removed.
The image shows the dynamic spectra
for the 750 seconds of data (.gif):
The plot shows average
spectrum with the compressor monitor on and off (.ps) (.pdf):
- the image was made using polB.
- the horizontal dashed line at 542 seconds is where the compressor
monitor system was turned off.
- The image has been smoothed and decimated by 6 so that the image
will fit on the computer screen. This changes the resolution bandwidth
from 5 khz to 30khz.
- There are no obvious lines that stop at this point.
- Top frame: polA, bottom frame polB.
- The black line is the average (500 seconds) with the compressor
- The red line is the average (250 seconds) with the compressor
- A birdie from the compressor monitor would have a black spike
with no corresponding red spike. I don't see any of these.
- No birdies were seen from the compressor monitor dynamic spectra
using 1 second and 30 Khz resolution.
- An average spectra of 250 seconds and 5 Khz resolution also
showed no birdies from the compressor monitor.
14/15apr11:Compressor monitoring system in
the lab (top)
The Compressor monitoring system was tested for rfi
by luis Quintero 14apr11 in the screened room. It was retested 15apr11
after some added shielding.The setup was:
The plots show rfi from the compressor monitor system:
- Compressor monitor system inside huffman box.
- agilent E4445A spectrum analyzer, 601 channels, and preamp on.
- 1 MHz to 10 Ghz. I've included the data from DC to 6gHz.
- ETS model 7405 902 probe: magnetic field. upper resonance >
- ETS model 7405 904 probe: E-field probe upper
resonance > 1Ghz
- A separate set of plots was made for each of the 2 probles used.
- Move the probe around the device looking for maximum value
(spectrum analzyer set to max hold).
- Do the tests with the huffman box open and repeat it with the
huffman box closed.
- The Compressor monitor has a 62.4964 MHz comb
- With the box open this extends up to 3GhZ.
- With the box closed it is not seen.
- With the box closed there is strong rfi 0 to 125 MHz probably
coming from the switching power supply.
- After some additional shielding of the device, the H field rfi
belwo 125 MHz went way down while the E field measurement went up.