Re: SNR meter, bandwidth and gain adjustment for transverters


I'm sorry guys - my autisim has kicked in and I have to get this off my chest before I explode.

As originally conceived the S-Meter was an indication of channel power. Meter ballistics and capacitor charge and discharge created a quaisi-peak reading over time. It takes a little time for the meter reading to rise and it generally if not originally takes longer to fall.

This served multiple purposes. I'm not sure which came first, protecting the operator's ears or providing a relative power indication for the operator. I suspect the latter was easier to implement before the days of AGC systems. Certainly IF alignment is facilitated by reading audio output power with any AGC disabled. Probably somebody figured that would be handy for operators to give some relative power readings during active communications. In some radios special meter calibration was used to give quasi accurate readings 6dB/S-unit. That did not sell so we got the 2 dB per S-Unit of the ProII coming out of the mess. (I have one so I pick on it's shortcomings.)

The important point above is that the S-Meter responds to instantaneous power changes slowly. This gives it the IARU's quasi-peak sort of performance much like a VU meter provides, as perhaps distinct from a Durrough meter. The value you see on the meter bears a fairly good indication, if the meter is accurate, of you real ability to communicate.

Once we move to the spectral domain from the time domain this "quasi-peak" loses meaning. It has to be applied between successive spectral computations. Since there are FFT peculiarities such as windowing involved it is difficult enough to be called impossible or at least quite impractical to make real accurate readings for say a 2.4 kHz bandwidth out of a spectrum with 10 Hz bins.

Fortunately in most cases one filters incoming data before switching to baseband processing. In the case of SSB it's a zero IF receiver so filtering after conversion can be used but power can be derived in much the same way. Take the IF filter output IQ values, square each one, and add them together. That is a real power value in dB relative to a dooby value. With a calibrated source the dooby value can become a real dBm value. (For AM demodulation the square root of that power value becomes your AM output without the distortion products the usual diode demodulators provide.)

Once you have that power value derived from I^2 + Q^2 you can filter it per the IARU quasi peak requirements and get correct S-Meter readings and behavior.

Do note that the I^2 + Q^2 reading above is NOT necessarily well related to selecting the highest level bin in an FFT output. With broadcast FM it very emphatically does not. I have seen peak spectral element values on a broadcast FM signal vary by more than 20 dB from silence to high modulation level. That is not in any way shape or manner proper S-Meter behavior.

Now maybe I can get this out of my head. I hope I did not get abusive above. (This is why I enjoy math. There is only one correct answer or set of answers. It's the way my brain works. And it's made my life "complicated" to say the least.)


On 20210705 12:02:37, Bob Dengler wrote:
At 7/5/2021 09:37 AM, you wrote:

Is this Continuum mode? I would need an exact specification were I to add
Not sure if I understand the question.  I'm referring to the channel power or S-meter, which if I understand correctly current indicates peak channel power.  It would be nice to have the option to have it read average channel power, then also have the ability to store those readings to a CSV file (2 columns: time & dBm) every x milliseconds.  This is exactly what HDSDR offers.  Once calibrated with a known noise source or hot/cold load, this allows using an SDR as a radiometer.  Also give us one or two significant digits after the dBm decimal point.

I'll put the above in the SDRC Requests forum if needed.  Thanks!

Bob NO6B 

Join to automatically receive all group messages.