I'd need to see math that proves this is
the way to handle coherent sidebands as opposed to summing a
large number of complex numbers in pairs of matching
sidebands. Are you doing something such that no filtered I and
Q value is ever present before the demodulator? That is where
it's automatic you get the right value.
Frequency, so using the FFT output. The
FFT output is in dBm (log10), so has to be converted back
from dB using pow(), averaged and then converted back to dB
using log10. This is trivial, doesn’t use much CPU at all.
Averaging across frequency or time?
Across frequency summation is perhaps closer, particularly
if you presume it's all noise and simply add power per bin.
But with coherent sidebands on a signal you can get
surprised simply adding power I suspect. Alas, I do not KNOW
that. I figure it's easier to do the I^2 + Q^2 operation on
the IF before the demodulator. With that signal and
something a bit like your noise averaging the signal
strength reading and noise reading could actually help you
know if there is enough signal to expect your demodulator to
provide something understandable, especially for digital
modulation schemes. (I'd take your minimum 10% data, sum the
powers (I^2+Q^2) in each bin, divide by number of bins, and
multiply back by number of bins in the filter bandwidth to
get the noise power estimate. That last could probably be
reduced to something close to a constant multiply
operation.) This makes WGN environments quite predictable.
Other environments suffer in comparison. So if the
calculation above gives too low an SNR, or value converted
to Eb/N0 is too low, then you know a
digital mode is not even worth trying. Of it is high enough
you know you have a sporting chance.
On 20210705 22:46:33, Simon Brown
FM power is always constant, this *could*
be an area for an averaging signal meter.
I’m slowly losing the will to live 😊 .
Simon Brown, G4ELI
[Edited Message Follows]
[Reason: Initial send occurred for seemingly no reason.]
I probably misspoke at some point if/when I said BCFM
carrier is always the greatest power anywhee in teh
passband, so I retract any such statement.
However, that statement does hold true for BCAM, so I've
captured a video of SDRC's S-meter reporting on a BCAM
Several key observations:
1) The carrier is very steady.
2) The modulation varies greatly.
3) the passband is intentionally varied between 2.5k and
4) The S-meter never varies at all (at least not more than
.1dB in any direction).
5) The S-meter is reporting ONLY the peak signal detected
anywhere in the passband, nothing more and nothing less.
So tell me where Simon gets it wrong.
Nevermind, that's a loaded question...Simon gets it right.
I'm done here,