Question about SNR measurement in SDRConsole

  • It seems to me that the SNR measurement of SDRConsole is not correct, according the following 4 images:

    Tuning in no signal and without zoom we have 6 dB SNR.

    Increasing only the zom and still no signal, the SNR rises to 18 dB !!! In both cases, the correct value would be 0 dB.

    With the beacon signal tuned and without zoom, we have 37 dB SNR. Zooming in on the beacom signal has 48 dB SNR !!! In both cases the correct value would be around 31 dB. (37-6 or 48-18)

    The SNR should be ZERO with no signal and should not change according to the zoom level!

    (I already sent an email to Simon but no answer ...)






    73 de Roland PY4ZBZ

  • Roland wrote: The SNR should be ZERO with no signal and should not change according to the zoom level!


    This was already discussed long time ago on the SDR-Radio support list after Simon introduced the SNR feature into the S-Meter. It seems that it is not a bug, but a feature :-)


    However, the change when zooming seems indeed a bug, hi hi...



    73, Oscar DJ0MY

  • He explains it also here why the SNR cannot be zero:

    https://www.sdr-radio.com/s-meter



    Note:

    When no signals are present with the SNR display it is quite normal to see a SNR reading. This is because Noise is the mean (average) whereas Signal is the peak value, so even in a quiet part of the band there will be a difference between the peak and mean. To see this yourself change the Smoothing algorithm to None (Ribbon Bar, View, Spectrum, Smoothing). The default smoothing is very good at eliminating noise.

    You see the same effect with a 'normal' receiver which has a S meter - even on a quiet band it will be moving and following peaks in the noise.

  • Thanks for the answers.

    I understood that the 6 dB SNR (and not zero) in the absence of a signal is because the weighting of the noise measurement (average) and the signal (peak) are different.

    But that does not explain why the SNR with more zoom increases ...


    Could someone tell me what is the power of the FSK beacons transmitted in Bochum and what is the gain of the antenna, or the E.I.R.P of the beacons?


    Thanks and 73s de Roland PY4ZBZ

  • PY4ZBZ: Hello Roland,


    According to an information that was distributed a while ago (and when I remember that right) the E.I.R.P. of the CW beacon is 950 Watt. Plausible, because the dish in Bochum is a 3m parabolic prime focus antenna and the power is around 300mW.



    Best 73


    Holger 'Geri', DK8KW

  • PY4ZBZ : Hello Roland,


    "S/N and (S+N)/N When measuring signal to noise ratio there are two basic elements to the measurement. One is the noise level and the other is the signal. As a result of the way measurements are made, often the signal measurement also includes noise as well, i.e. it is a signal plus noise measurement. This is not normally too much of a problem because the signal level is assumed to be much larger than the noise. In view of this some receiver manufacturers will specify a slightly different ratio: namely signal plus noise to noise (S+N/N). In practice the difference is not large, but the S+N/N ratio is more correct."


    (The above is a quote from https://www.electronics-notes.com ) This quote supports my observation that if you reduce the bandwidth the SNR rises, but at the same time the noise level without any signal present rises too. The ratio "Signal+noise/noise" stays the same. If you want to compare SNR readings you have to ask for the bandwidth too, or for the noise-level without any signal. I don't know if it has to be that way, but at least there seems to be nothing "wrong" with the SNR readings at different bandwidths.


    73 de Greg DF2IC :)& Peace

  • Hi Roland

    I have also been wondering about the way SDR Console defines SNR and how this affects the SNR number we read off the screen. I found an explanation on Simons webpage:


    ""

    Noise Floor

    A few weeks previously a reasonable logic was implemented for measuring the noise floor. Purists will not be happy - they rarely are, but it works for me.

    Take the output from the SDR radio, ignore 15% of the bandwidth at the high and low end of the output to avoid the ant-alias filtering, and we're left with a healthy 70% of the signal. Now sort the FFT bins by value, take the mean of the lowest 10% and that's the noise floor.

    ""


    Reading this doesn't fully explains the changing values vs. changing display bandwith, but at least it shows a different approach to "defining" noise level - hi.


    For what it's worth, I see the introduction of SNR reporting on satellite signals as a step forward. So thanks for putting that into SDR Console - even if noise bandwith is "undefined" and displayed numbers seems too high.


    73 Ole OZ2OE

  • Thanks Olger and Greg,


    Ooops.... Ole OZ2OE, my post happened while you did yours, hi hi ...


    S/N and (S+N)/N depend on the band in which the noise is measured. What is the band used by SDRConsole to measure noise?


    But the zoom applied to the spectrum display should only affect the visualization of the spectrum, and not the S/N measurement.

    The visual zoom applied to the spectrum (purely graphical display effect) does not affect the resolution of the FFT and therefore should not affect the value of the S/N?

  • Only now have I seen the noise and SNR information on Simon's website. I found the algorithm for noise measurement very interesting, in the fantastic SDRConsole software. Now I understand why we don't have SNR = 0 in the absence of a signal.


    In SDR # it is necessary to have at least a part of the tuning band in which only noise appears to have a correct indication of the SNR, as shown in the following figures. It seems that this software searches within the tuned band for the largest signal (the signal S) and the smallest signal, which is the noise N, to calculate SNR.


    73s de Roland PY4ZBZ