Intro

    The filter bank box for the 12meter will reside close the to the dewar in the 12meter hub. It contains various digital elements that cause rfi:

•  raspberrypi,  fiber to ethernet converter,  ethernet to usb converter..etc

    Osvaldo placed the components of the filterbank in a huffman box and worked on the shielding:

• power line filters (to better than 1GHz)
• feed throughs from the box to the outside world.
• contact between the lid and box.
  

   After the shielding work was complete, we  placed the huffman box in the screen room and ran some rfi tests to see how well it performed.

• 02mar23 - we tested the system with a 1 gbit ethernet to fiber transceiver.
• 06mar23 - we re tested the system with a 100 Mbit ethernet to fiber transceiver.

---

Setup:

    The setup was:

• box placed in the screen room
• 1-10ghz log periodic antenna used
  
• 100Mhz to 10Ghz 25 db amp attached close to antenna
• cable from screen room to fieldfox (outside screen room)
• field fox spectrum analyzer used for measurements
• 300 Mhz span, 30khz rbw,10001 points, rf attn=0, avg detector
• average 10 sweeps. record avg,maxhold,minhold  (about 70 second to avg 10 sweeps)
• Step cfr from 2400 to 9900 Mhz (26 steps)
• Do a complete set of measurements with box open (antenna close to opening).
• Do a complete set of measurements with box closed and antenna close to output cables on side of box.
• this had the strongest leakage.
• Datataking: x101/rfiscrn/ffscrn  cmd_ffwbfb.dat  
  
• output files: /share/phildat/wbrcvr/scrn/filterbank/fb_(opena/closeda)_yymmdd_nnnnn.dat
• Each file as a 10 sweep average for a 300 Mhz span.

---

Looking at the data

    The data for each measurement (lid open, lid closed) was first read in.
The  26 spectra. each of  10001 points,  300MHz span were concatenated into 1 large spectra.
The 10 sweep average was used for the plots.
   
    Plots were made from

• The 02mar23 data (1 gbit transceiver)
• The 06mar23 data (100 Mbit transceiver)
  

The first plots show the average spectra across the band:

• Top frame:  lid open

• Vertical scale dbm input at field fox (after 26 db external amp).
• black:spectra
• red:  spectra smoothed with a 1001 length median filter (30MHz).
  
• this is used for the the noise floor of the measurement.
• bottom frame:lid closed.
• the step in the spectra at 7.5 GHz is probably the band step of the field fox.
• the noise floor is being set by the field fox noise floor.
• 02mar23 with 1 gbit transceiver (.ps) (.pdf)
• Box open has spikes above -65dbm (in 30Khz rbw)
• box closed has spikes up to -83dbm
  
• 06mar23 with 100 Mbit transceiver (.ps) (.pdf)
• box open max spike is  -84dbm (in 30khz rbw)
• box closed: no spikes are seen above the noise floor (about -97dbm)

The 2nd plots show the spectra divided by the noise floor:

• Page 1: spectra divided by the noise floor
• top: box open
• bottom: box closed
• the red + are points 6db above the noise floor when the box is open.
• Page 2: 230302 box closed blowups in freq
• Page 3: 230392 : blowup strongest peaks around 3 Ghz.
• 02mar23 with 1Gbit transceiver (.ps) (.pdf)
• page 1:
  
• box open. peaks up to 35 db above the noise floor
• box closed: peaks up to 17 db above the noise floor.
• Page 2: blowups 2.4 to 5ghz and 2.8 to 3.4 Ghz
• Page 3: blowup showing strongest spikes around 3 GHz.
  
• 06mar23 with 100Mbit transceiver (.ps) (.pdf)
• Page 1
• box open peaks 17db above the noise floor
• box closed: no peaks above the noise floor.

The 3rd plot shows the box rfi reduction for 230302 (1Gbit transceiver) (.ps) (.pdf)

• Top overplot box open and box closed(red)
• bottom: boxopen/boxclosed
• some peaks were reduced by up to 29db 
• the 06mar23 data (with the 100 Mhz transceiver) was not plotted since there were no spikes above the noise floor with the box closed.

---

SUMMARY:

• The huffman box gave up to 29 db of attenuation (before hitting the noise floor of  the fieldfox)
• With the 1 Gbit transceiver there were still many points up to 17 db above the -103dbm(30khz rbw) noise floor.
• With the box closed, most of the rfi was coming out the power cable (even with the > 1GHz power line filter).
  
• With the 100 Mhz transceiver we did not see any points above the noise floor when the box was closed.
• We will use only 100Mbit transceivers for connections close to the horn.
• You can compare the box closed rfi with the 1 Gbit transceiver:
• dewar monitor box originally from hamdi with large holes in the side of the box (.ps) (.pdf)
• There are about 10 spikes above -90db with the box closed
• the amp used in this test had a better noise figure at the higher frequencies then than the 02mar23 test.
  
• the AO filter bank box with the holes closed, and power lines filters (.ps) (.pdf)
• There is 1 spike above -90dbm with the box closed.
  

processing: x101/230303/fbank_rfi.pro, x101/230306/fbank_rfi.pro

---