locked Feature request: LOWER FFT resolution options in SDR#


Bob Dengler
 

I'm using the SNR logger to log peak power of NBFM stations using an Airspy Mini. However, the peak power is derived from the FFT spectrum, not the total channel power. This is fine for CW signals, but when there is modulation on the NBFM signals the peak power drops because the width of emission exceeds the FFT bin size. I have the FFT resolution set to 512, which is the lowest it can be set to. The only way to get the FFT bin size large enough to adequately capture a NBFM signal is to disable decimation, which leaves me with only 12 bits of dynamic range. In the crowded SoCal VHF/UHF RF environment I often find that I need a couple more bits of resolution to maintain decent sensitivity without occasional overloading, so I prefer to use a decimation of 16 or 32. I could still effectively log NBFM channel power if I could further reduce the FFT resolution to, say 16. At a decimation of 32, the bin size would be then 150 kHz / 16 = 9.38 kHz, just enough. The FFT display would be kind of useless, but it would allow using the SNR logger to record channel power of wider emissions like NBFM, & my guess is that this would be very easy to implement since fewer resources would be needed to implement a smaller FFT.

Thanks

Bob NO6B


jdow
 

And if you get that you will whine that the bins do not fall in the right place.

Why might you be logging peaks rather than power. The latter simply challenges amplifiers the latter transfers information.

{o.o}

On 20211020 14:31:31, Bob Dengler wrote:

I'm using the SNR logger to log peak power of NBFM stations using an Airspy Mini.  However, the peak power is derived from the FFT spectrum, not the total channel power.  This is fine for CW signals, but when there is modulation on the NBFM signals the peak power drops because the width of emission exceeds the FFT bin size.  I have the FFT resolution set to 512, which is the lowest it can be set to.  The only way to get the FFT bin size large enough to adequately capture a NBFM signal is to disable decimation, which leaves me with only 12 bits of dynamic range.  In the crowded SoCal VHF/UHF RF environment I often find that I need a couple more bits of resolution to maintain decent sensitivity without occasional overloading, so I prefer to use a decimation of 16 or 32.  I could still effectively log NBFM channel power if I could further reduce the FFT resolution to, say 16.  At a decimation of 32, the bin size would be then 150 kHz / 16 = 9.38 kHz, just enough.  The FFT display would be kind of useless, but it would allow using the SNR logger to record channel power of wider emissions like NBFM, & my guess is that this would be very easy to implement since fewer resources would be needed to implement a smaller FFT.

Thanks

Bob NO6B







Bob Dengler
 

At 10/20/2021 05:00 PM, you wrote:
And if you get that you will whine that the bins do not fall in the right place.
Well if they aren't centered I'd think that could be fixed by tweaking the LO freq.

Why might you be logging peaks rather than power.
I'd like to record total channel power like HDSDR does, but peak power is all the SNR logger plugin will do. Since one application is simply to compare signal strength over time, peak vs. total doesn't matter so long as it's consistent.

Bob NO6B


jdow
 



On 20211020 21:47:04, Bob Dengler wrote:
At 10/20/2021 05:00 PM, you wrote:
And if you get that you will whine that the bins do not fall in the right place.
Well if they aren't centered I'd think that could be fixed by tweaking the LO freq.

That is the LAST thing I want to do. It is what I have to do with SDRC.

Why might you be logging peaks rather than power.
I'd like to record total channel power like HDSDR does, but peak power is all the SNR logger plugin will do.  Since one application is simply to compare signal strength over time, peak vs. total doesn't matter so long as it's consistent.

Bob NO6B 

It sounds like the SNR plugin is as broken as I expected. Good luck with it. I don't think your FFT approach will solve the issue properly; but, I could be wrong.

{^_^}


Simon Brown
 

I suggest *always* using a cubic spline to find the local peak value when looking for the peak in an array of bins.

 

Simon Brown, G4ELI

https://www.sdr-radio.com

 

From: airspy@groups.io <airspy@groups.io> On Behalf Of jdow

On 20211020 21:47:04, Bob Dengler wrote:

At 10/20/2021 05:00 PM, you wrote:
And if you get that you will whine that the bins do not fall in the right place.
 
Well if they aren't centered I'd think that could be fixed by tweaking the LO freq.


That is the LAST thing I want to do. It is what I have to do with SDRC.


Bob Dengler
 

At 10/20/2021 10:35 PM, you wrote:

I suggest *always* using a cubic spline to find the local peak value when looking for the peak in an array of bins.



Simon Brown, G4ELI
That's great advice, Simon. Unfortunately I'm not a programmer, at least one that's capable of authoring plugins. So I can only hope that a simple change could be made that will let me make use of the SNR logger without having to turn off decimation altogether, tanking my dynamic range.

Then again, if SDRC had noise-based squelch like SDR# I could just use your app & be done with it.

;) ;)

Bob NO6B