I'm facing again some spark noise that is generated ~500m from my RX antenna and baseband noise blanker most of the time heal it.
Best figures at baseband NB are 5dB threshold, pulse widht 1us and lookup window from 50 to 100us.
Normally I use 128KHz sample rate which result in 100KHz bandwidth.
But on broadcast bands, strong signals disable NB; On 40 meter CW segment the same occur when OTHR chirps below 7MHz but within this 100KHz passband and so on.
I did a simple test: I decreased sample rate to 16KHz (10KHz BW) and choose any free segment between two strong stations on 31m BC band.
Noise blanker was able to clean spark pulses completely. This makes me think that 10KHz BW is enough to detect fast pulses. Like an old fashion noise blanker where a secondary wide IF chain is used to sample noise, detect it, drive sort of fast switch and so on.
Have the lookup window more or less the function of this "virtual noise chain IF"? I've playing with this slider control but does not see much difference.
In another aproach a "NB version" where we can drag and drop noise window / fiter to a free spectrum segment, close enough to desired signal to keep noise pulses phase usefull to drive the "NB switch".
I'm having lots of fun with SDR# + Discovery. The Black friday HF+ Dual will arrive in some days. This one will be assembled on a remote location with the AirSpy server.
PY2PLL + PY3CRX