[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Linrad] Re: MAP65-not-IQ display oddities and doesn't decode

Hi all,

Leif Asbrink wrote:
> I can not give the answers about map65, but you might press 'T'
> to get timing info in Linrad. On my laptop I can ask for 96000 Hz
> and Linrad evaluates it to 95350. When I ask for anything between
> 96100 and 97000 I get 96650 Hz. The fractional resampler does not
> allow finetuning of the hardware. Somewhere in the system settings you
> can change the quality of the resampler. Trade quality vs cpu load.
> I do not know if that could allow a finer stepping in the resampling
> ratio. In any case, MAP65 will require a reasonably correct sampling
> speed and not all soundcards deliver that.

JT65 signals have a specified keying rate of 
11025/4096=2.69165 baud.  This number also gives the tone 
spacings in Hz for JT65A; the B and C modes use spacings 
twice and four times as large, about 5.38 and 10.77 Hz.

WSJT has been coded to expect signals to digitized at a 
11025 Hz rate.  MAP65 uses 96000 Hz and MAP65-IQ uses 
95238.1 Hz.

How close to their nominal values do all of these numbers 
need to be, for the system to work?  If the sample rate is 
off by, say, 1%, will JT65 signals still decode?  Here's how 
to find the answer.

Suppose the MAP65 center frequency has been set to 144.130. 
and the sample rate, nominally 96000 Hz, is actually 1% 
high, i.e., 96960 Hz.  A JT65 signal is present with its 
sync tone at 144.150.  Because the sample rate is higher 
than the nominal value, the sync tone will appear to be 
present not 20000 Hz above the center frequency, but rather 
at 20000/1.01 = 19802 Hz above center.  There is no real 
problem, yet; we are probably not calibrated well enough to 
notice a 200 Hz offset at 144 MHz, anyway.  We can simply 
tune in the signal where we find it.

However... in JT65B the highest data tone is above the sync 
tone by 66 x 2 x 2.69165 = 355.3 Hz.  The assumed 
sample-rate error of 1% compresses the received spectrum so 
as to make the highest data tone appear to be only 
355.3/1.01 = 351.8 Hz above the sync tone.  The frequency 
error on this tone (relative to the sync tone) is thus 355.3 
- 351.8 = 3.5 Hz.  This error is more than half of the JT65B 
tone spacing, so the decoder will have a significant 
problem.  Tones near the upper end of the range will be 
mis-identified by one full bin.

Worse yet... a JT65 signal consists of 126 tone intervals or 
"channel symbols".  Their spacing in time is the inverse of 
the keying rate, 1/2.69165=0.371519 s.  If the sample rate 
is off by 1% and the decoder has been properly 
time-synchronized at the start of a transmission, the 
assumed starting and ending times of the last channel symbol 
will be off by 0.01*126 = 1.26 symbols.  This is very bad! 
... and will almost certainly prevent correct decoding of 
the message.

The bottom line is that sample rates into WSJT, MAP65, or 
MAP65-IQ must be within about 0.1% (preferably within 0.05%) 
of the nominal values, or decoding performance will suffer.

	-- 73, Joe, K1JT

You received this message because you are subscribed to the Google Groups "Linrad" group.
To post to this group, send email to linrad@xxxxxxxxxxxxxxxx
To unsubscribe from this group, send email to linrad+unsubscribe@xxxxxxxxxxxxxxxx
For more options, visit this group at http://groups.google.com/group/linrad?hl=en