I have been experimenting with an Octagon LNB (Low Noise Block) which I’d obtained from the late John Roberts G8FDJ. Originally it had two outputs. John had modified this according to Bernie G4HJW’s guidance and had removed one output to allow room for an external reference oscillator input of 27 MHz. I used John’s TCXO (Temperature-Compensated Crystal OScillator) incorporating a 3MHz crystal and a 27 MHz harmonic filter. I hadn’t spotted the MIMIC buffer/booster amplifier.
I found a length of satellite coax with an F-connector on it & sealed this onto the LNB with self-amalgamating tape. I used a separate “Wi-fi” thin coax lead with SMA connectors and sealed this on the other LNB socket for the external reference input. I made a little clamp from some toilet seat bits and a bit of sailing rope and clamped the LNB close to the existing horn on the off-set dish arm. It is an 80cm diameter dish. I slanted the LNB, hoping to receive horizontal or vertical polarisation. I fed the satellite coax from John’s bias tee. The LNB output fed an RTL dongle (I’d given up with the FCD) and the data fed into GQRX running on Ubuntu Linux 18.04.
At first, I failed to get any sensible output from the LNB; I just got noise. The LNB did not resolve anything; certainly not the GB3FNY beacon, which is very strong here. The LNB output noise lifted when I added the reference oscillator input via 40dB of attenuation. This seemed a lot of attenuation. The LNB didn’t seem to work on it’s internal crystal alone.
Barry G8AGN brought round a neat Arduino-controlled marker that he had made. I got no reading off this either. Barry did a neat job of replacing the LNB’s reference feed-in capacitor with a 1nF chip capacitor and did a general clean-up of the LNB’s board & case. We tried again. Still nothing from his marker. I took out 10dB of attenuation. A signal! It turned out that John had included a MIMIC in the reference oscillator box. Also the PLL seems quite fussy about the level of 27 MHz reference input.
Back in my lab/shack I sealed the die-cast case using outdoor Bostik sealant and put the LNB back on the dish arm. I tried 27 MHz from my signal generator. The LNB produced noise with above -40dBm input up the coax. With the reference input a bit stronger I finally found the Finningley beacon appearing about 60 kHz higher than where I was expecting it.
I adjusted GQRX to compensate for the dongle frequency error. The RTL dongle was remarkably stable. Overall the whole receiving assembly seemed as stable as my DB6NT-K3 receiver with GPSDO reference, so this looked very promising.
So I found raised noise from the LNB does not necessarily indicate PLL locking. I need a lock indicator! I settled on 26 dB attenuation in the 27MHz feed. (assuming 4dBm from the OXCO+MIMIC, this gives -22dBm into the coax feeding the LNB’s local oscillator).
Peaking the dish on GB3FNY, I found I needed to dip the dish about 5 deg. to for maximum LNB signal. Azimuth, (on my sloppy rotator) seemed about the same.
Saturday 16-Feb-2019 I aimed the dish at Az 154 deg., El 24 deg. Nothing. I took the power off the LNB and put it back on: signals! The PLL must have locked. Fortunately, there were signals on the narrow band; the band edge beacons were switched off.
I compared the Goonhilly SDR with my LNB/GQRX output:
The top window is GQRX with a waterfall falling down. The lower window is the Goonhilly WebSDR with a waterfall falling up. The left-most trace is the lower beacon. The right-most trace is the upper beacon.
Update: I had failed to get my FUNcube dongle working with GQRX, so I bought an R820T dongle from cosycave.co.uk priced £13. This stabalises after some minutes & the frequency accuracy is much better than the dongle I have been using. Barry G8AGN showed me how to drop out the coax socket that it comes with (using my hot-air gun) and I soldered in an SMA socket and it all worked fine. A couple of days later, its little blue LED stopped working, but I can live without the LED.