v1.586: FT8 improvements, "Spec AF", misc fixes
From the CHANGE_LOG file:
v1.585,586 Mar 12, 2023
The spectrum ("Spec") button on the main control panel now toggles between the existing
waterfall spectrum (renamed "Spec RF") and a new audio channel spectrum ("Spec AF").
Spec AF is identical to the spectrum function on the FFT extension (which has been removed).
This change allows the AF spectrum to be observed when other extensions (that don't use the
top display) are running. The sig gen in particular.
Set default SNR correction value to -22 dB.
Include decoder-computed SNR in spots uploaded to pskreporter.info
Added background color stripes to log messages for easier reading (thanks Martin).
Fixed bug causing incorrect reseting of admin page WSPR grid field while typing.
SNR reporting task takes twice as many measurements spread out over one minute.
Thanks for taking a look at this.
The SNR results of KiWi's on the same site are now much more closely matched, which is great.
The down side is that the averaged results generally tend to be lower than the previous un-averaged results. I guess this is to be expected as it will remove any short term spurious peaks in performance that may have been produced as a result of local interference, or any other very short term propagation enhancements to signals.
We will have to wait until the new averaged version ripples through to all on-line KiWis, before the SNR scores make sense again.
One other option is to take the average of peak values, Max / Min ?, rather than just an average, or maybe that's nonsense and would actually make no difference.
Averaging occurred previously. The waterfall capture was run in "fast" mode and 32 averages taken (over about one second). Now it's run in "1 Hz" mode and 64 averages taken (over about one minute).
Thanks for that explanation, it makes good sense.
However I'm now regretting my suggestion, as it has dramatically changed the 'pecking' order on the SNR list, and from the few KiWi's I've sampled it now seems to be even less representative of their actual performance.
Just one example
http://g8ure.ddns.net:8074/ at 10. with 41 & 37 but practically no signals on the upper HF bands, as the antenna gain is not adequate to lift the noise floor and receive signals.
and mine at Wessex which IMHO is one of the best http://wessex.zapto.org:8073 at 84. with 35 & 37, with lots of signals across the whole spectrum, even on the HF bands although it is now well after sunset.
and another of mine at Kernow http://kernow.hopto.org:8073/ at 61. with 36 & 29, even though by my own admission it is one of the worse performers due to the temporary antenna and pickup of local noise.
The ones in the cluster at Weston / G3SDH are also way down too, despite all being pretty good performers.
I'm not blaming you John, as it was my idea, but I don't know why despite being a lot more consistent when using shared antennas, the reported SNR values now seem to be even less related to the actual level of performance.
I'm genuinely puzzled.
Well, let me know if you want the change reverted. I've got an update ready to go out with an FT8 fix.
OK revert it.
Although technically it should be better, it seems to be making the derivation of any already flawed metric even worse.
Sorry, it seemed like a good idea at the time.
v1.587 Mar 13, 2023
Limit log output to 1024 lines (roughly 10 minutes of decodes). Prevents browser from
eventually going into meltdown.
Reverted SNR measurement change per request.