Noise factor in homemade software-defined GPS receiver

Question

I'm building a homemade software-defined GPS receiver. I'm currently at a point where I can acquire and track strong signals but struggle to even detect weak signals.

I've ruled out problems with my software -- I can easily detect all satellites from sample RF signals that I've downloaded from various sources (e.g. gnss-sdr's getting started guide).

The problem appears to be in the way I'm capturing the RF baseband. My setup includes an inexpensive GPS antenna with an integrated LNA, and an AirSpy R2 SDR (with a built-in bias-tee).

I'm considering buying a more expensive SDR (USRP B205) which has a ~5dB lower noise factor at the GPS C/A frequency (1575.42 MHz). However, I'm unsure about whether this is going to help because the overall NF is supposed to be dominated by the NF of the first LNA, which is part of the antenna.

1. Am I right that improving the NF of the SDR is not going to meaningfully affect the sensitivity of my receiver? In particular, going from an AirSpy R2 SDR to a USRP B205 should decrease the NF of the SDR by ~5 dB, but, because this is after the 28 dB LNA of the active antenna, Friis formula tells us that the overall effect should be minimal.
2. What can I do to improve the sensitivity of the receiver? I have enough gain that the SDR is using a significant fraction of its dynamic range, so throwing more LNAs is not likely to help.

Thanks!

Answer 1

Can your SDR measure signal strength at specific frequency?

If so a simple test is to measure the noise power at a frequency where no signals are present (maybe place Antenna/LNA in shielded box). Then disconnect LNA and put 50Ohm terminator on the SDR input. Remeasure the noise at the same frequency.

If it decreases without the LNA your NF is dominated by the LNA. No change (<3dB) suggests NF is SDR dominated.

Other thoughts: Do you know the NF of the LNA and do you believe it. Seems to be a low cost item, could have embellished specs. The actual antenna in the unit may not have much passive gain. Active gain in the LNA wont help too much if the antenna itself doesn't have good sensitivity.