From a9663b5b47df9ab70c0509eca3323eb5ff1a1761 Mon Sep 17 00:00:00 2001 From: c vw Date: Tue, 27 Oct 2020 13:54:06 +0100 Subject: [PATCH] Added comment to the changed NB/NB2 parameters. --- receiver.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+) diff --git a/receiver.c b/receiver.c index ebbdc62..8562d2d 100644 --- a/receiver.c +++ b/receiver.c @@ -1107,6 +1107,20 @@ fprintf(stderr,"create_receiver: OpenChannel id=%d buffer_size=%d fft_size=%d sa 1, // run 0.010, 0.025, 0.0, 0.010, 0); +// +// It has been reported that the piHPSDR noise blankers do not function +// satisfactorily. I could reproduce this after building an "impulse noise source" +// into the HPSDR simulator, and also confirmed that a popular Windows SDR program +// has much better NB/NB2 performance. +// +// Digging into it, I found the Windows SDR program used NB default parameters *very* +// different from those recommended in the WDSP manual: slewtime, hangtime and advtime +// default to 0.01 msec, and the threshold to 30 (which is internally multiplied with 0.165 +// to obtain the WDSP threshold parameter). +// +// Since there is currently no GUI in piHPSDR to change these values, they are now hard- +// coded here (0.01 msec ==> 0.00001 sec, 30 ==> 4.95). +// create_anbEXT(rx->id,1, rx->buffer_size,rx->sample_rate,0.00001,0.00001,0.00001,0.05, 4.95); create_nobEXT(rx->id,1,0,rx->buffer_size,rx->sample_rate,0.00001,0.00001,0.00001,0.05, 4.95); -- 2.45.2