Intro

    On 12apr21 we found that the sband signal rfi was so strong that it was causing glitches in the xband signal. The 12meter multiplexes the sband and xband rf signals onto the fiber optics and sends them down to the rcvr room. The udc (up down converter) then selects either the sband or xband signal to convert to 1-2 GHz. The measured power levels showed that the fiber optics was not saturating (fiber output was about -30 dbm, with about 24db loss in fiber the input was less than -5 dbm.. The 1db compression for the fiber is 10dbm.). Later measurements showed that the udc's 1db compression was -35dbm.. So the compression we were seeing in apr21 with the sband signal was probably in the  udc .


    On 7dec21 we placed pads in front of the sband inputs to the rf box in the 12meter pedestal. We tried 10db and then 15 db pads. The current 12meter config also has [15,20] db pads (polA,polB) before the UDC inputs.

---

Data taking setups used:

    The 12meter tracked blank sky while the data was taken.

sband:

• We took 2 172 MHz bandwidths centered at 2250 and 2350 MHz
  
• band 1 2164 to 2336 MHz
• band 2 2264 to 2436 MHz
• the rf filter in the 12meter sband rf goes 2185 to 2364 MHz
• The mock spectrometers were set to 172MHz bw, 1 second dumps, 8192, channels, an a/d rms of 60 counts.
  
• 60 seconds of data with 1 second dumps was taken
• 1 60 sec scan with 10db pads
• 1 60 sec scan with 15 db pads.

xband:

• 7 x 172MHz bands were taken at [8219,8363,8505,8647,8789,8931, and 9073 MHz.
• 1 300 second scan with 1 second dump were taken
  

---

Looking at the data:

    The first plots shows the sband spectra (.ps) (.pdf)

• Top plot: spectra with 10 db pads
• bottom: specdtra with 15 db pads
• the vertical scale is a log scale in mock counts.
  
• colors:
• black: band 1 polA
• green: band1 polB
• red:   : band2 polA
• blue:  : band2 polB
• the sband rf filter of the 12meter 3db fall off is at 2364MHz
• the signal beyond 2370 MHz is noise from the IF filters (so the sky signal is not very far above the noise floor).
• The green,black bump around 2300 MHz is an i/q image of the large rfi at 2190 MHz.
• The large value at 2350MHz has hit the 31bit limit of the mock accumulator.. we would need to increase the attenuation, or decrease the dump time of 1 second to stay away from this limit.
• the spikes in the red/blue 2250-2290 part of the spectra are not seen in the black,green overlap. So these spikes are being generated in the 12meter system.

    The second plots show 300 second of xband total pwr while sband connected (.ps) (.pdf)

• page 1  spectra and rms/mean by channel
• top: 7x172 MHz average spectra.
• the band passes have not been removed
• black is polA
• red is polB
• I've clipped 4% of each side of the bandpass in the overlap region.
• bottom: rms/mean by channel vs freq
• computed using the 300 1 second spectra.
• The expected rms is .0069 (1/sqrt(bw*time))
• the negative going spikes are the dc channel of each spectra.

Summary

•  The sband spectra show where the large rfi signals are in the sband portion of the 12meter receiver.
• It looks like we should put a different filter in to cut out these strong signals
• The xband signal did not show any glitches with the sband signal active.
• If we are not going to use the sband till we get the filter, it might be better to terminate the sband inputs to the rf box in the pedestal just in case something is glitching the rf signal.

processing: x101/211207/sb12m.pro, xb12m.pro