mirror of
https://github.com/lwvmobile/dsd-fme.git
synced 2026-11-16 18:57:24 -05:00
Sample/Symbol Tweaks;
This commit is contained in:
@@ -479,7 +479,8 @@ getSymbol (dsd_opts * opts, dsd_state * state, int have_sync)
|
||||
|
||||
if (state->samplesPerSymbol == 20) //nxdn 4800 baud 2400 symbol rate
|
||||
{
|
||||
if ((i >= 9) && (i <= 11))
|
||||
// if ((i >= 9) && (i <= 11))
|
||||
if ((i >= 7) && (i <= 13)) //7, 13 working good on multiple nxdn48, fewer random errors
|
||||
{
|
||||
sum += sample;
|
||||
count++;
|
||||
@@ -498,12 +499,23 @@ getSymbol (dsd_opts * opts, dsd_state * state, int have_sync)
|
||||
if (state->rf_mod == 0)
|
||||
{
|
||||
// 0: C4FM modulation
|
||||
|
||||
if ((i >= state->symbolCenter - 1) && (i <= state->symbolCenter + 2))
|
||||
{
|
||||
sum += sample;
|
||||
count++;
|
||||
}
|
||||
if ((opts->frame_p25p1 == 1) && (i >= state->symbolCenter - 2) && (i <= state->symbolCenter + 2))
|
||||
{
|
||||
sum += sample;
|
||||
count++;
|
||||
}
|
||||
//testing dmr may be slightly off compared to P25, particularly wav recording at narrower bandwidth?
|
||||
else if ((opts->frame_dmr == 1) && (i >= state->symbolCenter - 1) && (i <= state->symbolCenter + 2))
|
||||
{
|
||||
sum += sample;
|
||||
count++;
|
||||
}
|
||||
//anything else
|
||||
else if ((i >= state->symbolCenter - 2) && (i <= state->symbolCenter + 2))
|
||||
{
|
||||
sum += sample;
|
||||
count++;
|
||||
}
|
||||
|
||||
#ifdef TRACE_DSD
|
||||
if (i == state->symbolCenter - 1) {
|
||||
@@ -514,7 +526,18 @@ getSymbol (dsd_opts * opts, dsd_state * state, int have_sync)
|
||||
}
|
||||
#endif
|
||||
}
|
||||
else
|
||||
else if (state->rf_mod == 1) //QPSK
|
||||
{
|
||||
//going one left, two on the right for QPSK, local system seems to favor that
|
||||
//with same PPM setting used on C4FM, unsure if that works on other systems, just
|
||||
//testing things now, not like QPSK works well here anyways
|
||||
if ((i == state->symbolCenter - 1) || (i == state->symbolCenter + 2)) //1,2
|
||||
{
|
||||
sum += sample;
|
||||
count++;
|
||||
}
|
||||
}
|
||||
else //GFSK
|
||||
{
|
||||
// 1: QPSK modulation
|
||||
// 2: GFSK modulation
|
||||
@@ -523,12 +546,23 @@ getSymbol (dsd_opts * opts, dsd_state * state, int have_sync)
|
||||
// comes one sample too late.
|
||||
// This change makes a significant improvement in the BER, at least for
|
||||
// this file.
|
||||
//if ((i == state->symbolCenter) || (i == state->symbolCenter + 1))
|
||||
// if ((i == state->symbolCenter) || (i == state->symbolCenter + 1))
|
||||
//GFSK should ideally be the same on the left and right, I would think
|
||||
//I have not observed a system that would benefit from this alignment
|
||||
if ((i == state->symbolCenter - 1) || (i == state->symbolCenter + 1))
|
||||
{
|
||||
sum += sample;
|
||||
count++;
|
||||
}
|
||||
{
|
||||
sum += sample;
|
||||
count++;
|
||||
}
|
||||
|
||||
//MISC Notes: When using RTL input, NXDN96 at BW:16 NXDN48 at BW:12
|
||||
//I am beginning to suspect that RTL input is halfing the bandwidth, which
|
||||
//is my original consclusion back when I was working on it last, unsure
|
||||
//or the best course on that, could just be a rtl_fm flaw
|
||||
//P25 may also work better at 16 as well now, hard to tell the difference since both are good
|
||||
//DMR may not be any different on 12 or 16
|
||||
//most likely though, this all will just depend on signal stregth more than anything
|
||||
//as to how much BW you should set
|
||||
|
||||
#ifdef TRACE_DSD
|
||||
//if (i == state->symbolCenter) {
|
||||
|
||||
Reference in New Issue
Block a user