Intro

    On 27jun23 a4005 was doing 5 min on,offs looking for hc3n. The upper part of the bandpass was oscillating with  aperiod of about .5 Hz.
On 29jun23 i took some more data looking at this problem.

---

Setup:

• The 12meter receiver used the wb6 xband filter
  
• 1 mock spectrometer were used:
  
• 172.032 MHz clock with a bw divisor of 7 gave a bandwidth of 24.567MHz
• 4096 channels gave channel width of 6Khz (about .2 km/sec) at 9103 MHz.
• band center freq of 9103.3319MHz
• dump time was 1 second.
• pol A and B were  recorded.
• pointing pattern
• 5 min on,off with a 10sec cal on,off at the end of each on,off
  

---

A4005 data from 27jun23

The first plot shows the 5 min on/off for file number 900-1200 on G351 (.ps) (.pdf)

• Black in polA, red is polB
• Page 1: on/off -1 and  rms/mean by freq channel
  
• Top frame: on/off -1 for the 5 minute on,off
• the problem covers 9105 to 9116
• The  negative going values points to it being stronger in the off than in the on.
  
• bottom: rms/mean by channel for the 300 1 second on records.
• The expected rms/mean of 1/sqrt(.006E6*1.) - .013  is close to the measured value outside of the problem area.
• The variation is stronger in polA than polB.
• Page 2: total power 9210 to 9114
• Top: total power  vs time
• Bottom: spectrum of the total power.
• The data is sampled at 1 second so the highest unaliased freq is it can measure is .5 hz.
• It looks like the problem is periodic with a period of 2 seconds

The dynamic spectra shows the 300 1 sec spectra of file 900 polA  (.jpeg)

•  The spectral variation is strongest between the red lines 9110 to 9114.
• You can still see variation in 9106-9109
• The total power between the red line was then computed (see below)

processing: x101/230627/xbrfi.pro

---

230629 test data.

    On 230629 i took some  test xband data to look at the problem. I decreased the spectral dump time to .1  seconds.

The setup was:

• Mock setup:
• 4096 channels over 24 MHz, .1 second dumps
• box 0: cfr 9103 MHz
• box 1: cfr 8900 MHz
• this was centered on a know source of rfi at xband.
• telescope
• sitting at az=255, el=40 deg
• 300 seconds of data were taken
  

The first plots show the avg spectra and rms/mean for the 2 bands (.ps) (.pdf)

• Page 1: 9103 MHz band
• black is polA, red is polB
• top: 300 sec average spectra.
• Bottom: rms/mean by freq channel for the 300 seconds.
• we don't see and variation in the 9110 to 9114 MHz part of the spectra.
• Page 2: 8900  MHz band
• Top average specdtra
• bottom: rms/mean by freq channel
• this rfi if part of the 8730 to 9000 MHz rfi that had been previously documented   (more info)
• it is a chirped radar with 20MHz bands covering 8730 to 9000 MHz
• a 22.2 usec pulse,
• 1.2096 ms ipp
• 4 second rotation rate.

The dynamic spectra show the 300 secs of data for the two freq bands

band

9103 PolA (.jpg)

8900 PolA (.jpg)

8900 PolB (.jpg)

• 9103A dynamic spectra does not show any obvious problem at 9110 to 9114MHz
• The red lines show where the total power is computed (see below).
  
• the 8900 MHz dynamic spectra show the 15 MHz radar  with the adjacent guard bands.
• back in feb23 the20MHz bands included the guard bands.
  

the next plots have the total power spectra for the 2 bands (.ps) (.pdf)

• Black is polA, red is polB
• Top: total power spectra for the entire 8900 MHz band
  
• middle: total power spectra for the entire 9103 MHz band
• bottom: total power spectra for 9110 to 9114 MHz
• The 8900 MHz rfi has a .5 hz comb (fundamental period of 2 seconds)
  
• back in feb23 the comb was .25 hz (fundamental period of 4 seconds)
  
• we see the .5 Hz comb in the 9103 MHz band (though not as strong at a4005 saw it on 27jun23).
• a more careful observation showed that this periodicity is probably from compression across the entire band.

processing: x101/230629/xbrfi.pro

---

Summary

• on 27jun23 a4005 saw an unstable bandpass in their 9110 to 9114 MHz portion of the band
• It was seen in polA and polB
• it had a period of 2 seconds.
• on 29jun23 i took some .1 second spectral data at 9103 and 8900 MHz
• the 9110 to 9114 MHz portion of the band was not unstable.
• the 8900 MHz known rfi showed a periodicity of 2 seconds.
• the 9103 MHz band total power spectra showed a periodicity of 2 seconds (probably from compression).
• The 8730 to 9100 MHz rfi  has previously been seen in xband observations with the 12meter (more info)
• It previously had a periodicity of 4 seconds.
  
• It now has a period of 2 seconds.
• The 24MHz band centered at 9103 MHz is :
• initially sampled at 172.032 MHz 
  
• the rf input s 9017 to 9189.
• the analog filters in front of the samplers are not very steep.
  
• This sampled data in then filtered digitally down to 24MHz.
• My guess is that the 9000 MHz rfi is folding into the 9110-9114 MHz part of the band.
• What can be done:
• If it is seen again, you might shift the freq by a MHz or so and see which way the rfi goes.
• You could add a second band that sits on the 8900 mhz rfi and then correlate this with the problems in the 9103 MHz band
• the problem probably occurs when the radar points at AO.
  
• If you increased the sampling to .1 seconds, you might be able to exclude the worst part of the rfi.
• Of course your disc space will increase by a factor of 10.
  
• the previous 4 second rotation period showed  the system saturating for about 30 milliseconds.