[hpsdr] FFT latency

[G3XJP]

FFT latency, inherent or not inherent? Who would have imagined such a 
simple question would generate such a diversity of replies? Many thanks 
for all of them. I conclude the answer is "YES, inherent in the use of 
an FFT". And so I will continue to avoid lots of narrow FFT bins because 
although they might make the Rx better, they would also make the Tx/Rx 
worse for what I want it for.

Which is not CW contesting where the format, sequence and a significant 
part of the content is predictable. Which is not "take it in turns to 
speak" QSOs or worse, net controllers. On the complete opposite end of 
the spectrum (pardon pun), a genuine conversational near-duplex VOX QSO 
often has more than one person talking at the same time - but controlled 
and managed and for very brief durations. Where the "brief duration" is 
as long as it takes you to recognise it is happening.

This morning it took me 5 mins of DSP'ing to add a variable test delay 
line in my Rx path. This is on top of what is there already. Anything 
more than abt 30-40 msecs total Rx delay feels very uncomfortable - and 
leads to definite and uncontrolled doubles. Maybe over time I could get 
used to it because this is an operating style thing as much as anything 
else. But I don't need to. For actual use purposes, you need to add in 
your MIC-to-ANT Tx delay/latency as well.

BTW a barely controlled double is when somebody has to say "Sorry, I 
just doubled with you. Say again." An uncontrolled (disaster) double is 
when they did not even notice it happened - and you hear plenty of those 
on eg 40m SSB nets. Horrible! Are the plethora of cheepo SDRs out there 
making it worse? ANS = NO - because an inherent delay is not a function 
of how much you paid for it. It applies to them all.

Peter G3XJP

 1428659697.0