Link budget validation for a UHF satellite ground station

Question

I built a UHF satellite ground station. It has a yagi with G=16.1dBi and an LNA with G=20.1dB. It’s a 430 - 450 LNA with NF=0.68. The RF cable, rg174/u has I think close to a 1dB/m, which would mean around 5dB loss.

The output of the yagi is connected to an SDR receiver by a thin 5-meter UG174/U RF coax cable as follows:

Antenna —> 5-m cheap RG174/U coax —> LNA —> SDR —> Rx Software

The Challenge is poor packet reception from CubeSats, which could mean either my signal processing or poor link budget is to blame. I tested this hypothesis by first transmitting a playback of recorded satellite signals from 400-meters away. The receiver was able to the get packets without any problems. This test was of course done without considering Doppler shift. However, I do Doppler correction when I track satellites. I use TLEs to calculate the shift in real time. I have verified that my calculated Doppler shift is similar to GPredict.

Before I dissect the DSP functionalities, I have decided to rule out other issues such as link budget first. In particular, I’m worried that the cheap and not-so-well-shielded 5-meter cable from the antenna to the SDR might be gathering too much noise/interference (Friis formula). What do you think of this hypothesis? I welcome all ideas.

UPDATE 1

I moved the LNA closer to the antenna. I replaced the cheap 5-m RG174/U coax with a 16-cm coax of the same type. The LNA is connected to the SDR through a relatively better RG213 with a 0.65 dB loss. The connection now looks as follows:

Antenna --> 16-cm cheap RG174/U coax --> LNA --> 5-m RG213 --> SDR --> RX Software

The new setup shows significant improvement. The SDR software can demodulate and decode packets (with a few misses here and there). I have a few more ideas/questions to further improve the link budget.

1. I still use a cheap 16-cm coax between the antenna and the LNA. Will there be any significant change if I replace this with a better cable?
2. The LNA has an internal 10-MHz bandpass filter mentioned in the first answer, which I can't replace without desoldering the LNA casing. Suppose I get a narrower bandpass filter, say 3 MHz bandwidth, and put it at the LNA input. Will this add any improvement in case I don't remove the internal 20 MHz bandpass filter (thinking about Friis' formula)?

Answer 1

There could be many things going wrong so this is a slightly rambling answer, a primer on low noise ground stations. I suggest reading about amateur EME stations, there is more published there than on satellite ground stations. This is what I would check, in order of priority:

1. The LNA is not at its best if there's some coax before it. With 5 dB of coax loss in front of the LNA, given the good noise figure and that you're looking up, the system performance is degraded by more than 5 dB. That's a lot of SNR to give away.
   Try using a cascade calculator website or spreadsheet to see the difference between LNA-first and LNA-after-coax. If I assume the antenna has a noise figure of 1 dB for some ground pickup, and the cable is a 5 dB attenuator with a 5 dB noise figure, then LNA first the chain NF is 1.6 dB, while the other way it is 6 dB, so the Signal to noise ratio is degraded by 4.4 dB.
   I don't think that the coax will be picking up noise or interference - the antenna will do that much better than even RG-174. But some better coax might help.
   Solution: move the LNA close to the antenna feedpoint.

2. The LNA should ideally have a filter to keep out FM radio, pagers, general mobile radio, pagers, TV etc. The amateur bands are reasonably quiet but in any major population centre you will find a lot of strong signals that will overload an LNA that isn't filtered. The antenna will only help a bit with rejecting lower frequency signals, but you should consider it a reasonably good collector of all signals. The SDR is probably worse than the LNA in terms of being overloaded and mixing everything together.
   Test: Since you have and SDR, try scanning the whole band with one of the wide band scan programs, with the AGC off, and see what you see. Ideally you will see only signals in the 420-460 MHz range, and not much anywhere else. Some people report seeing unexpected signals all over the place, repeated copies of radio stations at various wrong frequencies. This is a sure sign that the SDR is overloaded.
   Solution: If you see strong signals in the rest of the band, consider building or buying a bandpass filter. It goes between the antenna and the LNA. It could be a specialised amateur band filter, though a mini-circuits filter would make a big difference too.

3. Antenna pointing. Is the antenna definitely pointing at the satellite? Do you drive it around manually to search, or just trust the rotator?
   The beamwidth of your yagi will be about 30 degrees, so if it's not perfectly calibrated you might miss the satellite. 15 or 20 degrees wouldn't be obvious from looking at it, but it would significantly reduce the signal strength.
   Test: You could check the calibration by having it track the sun and looking at the shadow, at a few times in the day.

4. Damaged cable, connectors, LNA, etc. Yesterday I found one of my cheap no-name SMA-BNC adapters had developed a short circuit inside it, even though it looked fine... it took a while to solve this.
   Test: check the signal chain with a (nano) VNA - but use a 40 dB attenuator so the signal source doesn't overload the LNA. Check the antenna SWR is low at your frequency. Check that the SWR rises dramatically when the antenna is pointed at a large metal object. Find a weak signal source at your frequency and see if you can detect it when held in front of the antenna. The noise from some unshielded electronics might work as a source, something like a raspberry pi, a phone charger or electronic toy. If it's visible on the SDR on the bench, it should be stronger when held in front of the yagi+LNA.

5. In general you could try listening to other, stronger satellites, or even terrestrial signals like distant ham repeaters, to check that the system sensitivity is what you expect.

6. Software problems, as you say. Try listening for the satellite with gqrx or SDR# instead of your special DSP, see what appears on the screen. Record the samples and play them back through your demodulator later.

I have some experience with UO-14, long ago. It apparently had 10 watts of power at 435 MHz and on a high pass was readable with a quarter-wave vehicle-mounted antenna and a regular ham radio, or full quieting with a 5-element yagi and a handheld radio. With a 16 dBi yagi and an LNA you should get good results if the satellite transmitter is 1 watt. If it's less than that, share your link budget calculation and test results in a new question.