Higher Noise Floor and Incorrect SNR in Groundstation vs SDR++

[damianross]

On the same machine (Raspberry Pi 5 with rtl-sdr v4), in Groundstation the noise floor and signal are much higher than with other programs (e.g., SDR++).

The signal is also higher, so the waterfall graph scale might be incorrect.

However, an online SNR calculator shows a much worse SNR in Groundstation compared to SDR++.

Settings (same in both):

• Bias-tee: ON (for LNA)
  • Similar behavior is observed even when it is OFF
• Gain: 0
• Sample rate: 1.024 Msps

Comparison Screenshots (Groundstation vs SDR++)

[sgoudelis]

There is indeed an issue with the power calculation in the FFT. I'll test something out and roll out a fix soon.

[sgoudelis]

Try version 0.4.1 and let me know how it goes.

[damianross]

Thanks for the quick response!

I tested with satdump aswell. Noise floor seems to be more in line with SDR++ and SatDump but the signal is lower in comparison. SatDump and SDR++ seem to be consistent with eachother. For reference I'm testing with the signal from ATIS broadcast from a nearby airport so it should be pretty consistent.
Maybe the problem isn't just the scaling of the FFT?

[sgoudelis]

I am still checking this out, I ll let you know...

[sgoudelis]

Try out version 0.4.2 and could you please tell me what db values you are seeing now?

[damianross]

-45/-81 for groundstation and -40/-78 for SDR++. They seem to be pretty similiar now though sdrpp and satdump get a slightly stronger signal. This is of course assuming their scales are correct. I can send you an audio recording from both though they sound pretty similiar. Ideally someone could try and replicate this. Thanks again for the rapid response!

[sgoudelis]

No thats fine, I dont need a recording. Glad to hear it is working for you.