v1.8.1-beta2 - KA9Q Channel Filter, RS41/DFM Estimator BW Tweaks

[darksidelemm]

Some possible performance impacting changes in this one.

• get_sdr_iq_command function now accepts an optional channel_filter argument. This gets passed to the KA9Q channel setup, allowing bandpass filtering. We should look to see if this can be implemented for other SDR types.
• DC blocking is disabled for KA9Q setups. It doesn't look to be required.

In response to an issue observed by @argilo where a DFM and RS41 are regularly seen only 10 kHz apart, I've made the following changes:

• RS41:
  • Estimator range has been narrowed to +/- 5 kHz, and a +/- 5 kHz channel filter is enabled for KA9Q setups.
  • fsk_demod is set to use a mask estimator, with a tone spacing of 4800 Hz.
• DFM:
  • Estimator range has been narrowed to +/- 5 kHz, and a +/- 5 kHz channel filter is enabled for KA9Q setups.
  • This should be fine for DFM17 sondes. However, there is a possibility that DFM09s may drift outside of this window. Need to evaluate this in the wild.

We should look into the validity of the current estimator ranges for other sonde types, and whether the mask estimator adds any performance improvement.

Some testing data of mask estimator vs no mask estimator:

RS41 - No Mask Estimator

rs41_96k_float_08.0dB.bin, 0, 1.739
rs41_96k_float_08.5dB.bin, 0, 1.385
rs41_96k_float_09.0dB.bin, 0, 1.944
rs41_96k_float_09.5dB.bin, 21, 1.813
rs41_96k_float_10.0dB.bin, 56, 1.928
rs41_96k_float_10.5dB.bin, 78, 3.035
rs41_96k_float_11.0dB.bin, 76, 2.458
rs41_96k_float_11.5dB.bin, 94, 3.305

RS41 - 4800 Hz Mask Estimator

rs41_96k_float_08.0dB.bin, 0, 2.111
rs41_96k_float_08.5dB.bin, 0, 2.322
rs41_96k_float_09.0dB.bin, 3, 2.053
rs41_96k_float_09.5dB.bin, 54, 2.467
rs41_96k_float_10.0dB.bin, 103, 2.297
rs41_96k_float_10.5dB.bin, 115, 2.283
rs41_96k_float_11.0dB.bin, 110, 2.367
rs41_96k_float_11.5dB.bin, 116, 2.317

(Approx 0.5 dB improvement)

DFM - No Mask Estimator

dfm09_96k_float_05.0dB.bin, 0, 2.591
dfm09_96k_float_05.5dB.bin, 0, 2.957
dfm09_96k_float_06.0dB.bin, 5, 2.097
dfm09_96k_float_06.5dB.bin, 24, 2.181
dfm09_96k_float_07.0dB.bin, 59, 1.805
dfm09_96k_float_07.5dB.bin, 71, 2.033
dfm09_96k_float_08.0dB.bin, 88, 2.446
dfm09_96k_float_08.5dB.bin, 90, 2.353

DFM - 5080 Hz Mask Estimator

dfm09_96k_float_05.0dB.bin, 0, 2.073
dfm09_96k_float_05.5dB.bin, 0, 2.367
dfm09_96k_float_06.0dB.bin, 1, 2.539
dfm09_96k_float_06.5dB.bin, 14, 2.082
dfm09_96k_float_07.0dB.bin, 55, 2.229
dfm09_96k_float_07.5dB.bin, 64, 2.099
dfm09_96k_float_08.0dB.bin, 85, 2.447

(Not really any performance gain in this case, maybe a slight degradation)

[argilo]

I've got this running on my main station, and will let you know how it goes.

[argilo]

This is looking good. In the short time since applying this change on my main station, I've seen normal-looking packet counts for the launch sites around me, and I have seen simultaneous reception of 403.800 (RS41) and 403.810 (DFM17) for the first time.