Leon Heller wrote:

"David J Windisch" wrote in message
...

Hi, all concerned:

Ponderings while listening to a dead 40M c-w frequency ...

What artifacts does latency in a DSP implementation cause?

I'm thinking of eg a "DSP rcvr" OOH and an "analog receiver" OTOH, both
tuned to WWV, and the noticeable delay of the DSP audio from the analog
audio. You know, ti-tick ti-tick instead of tick-tick-tick, and "blurred"
voice announcements.

I've noticed a similar effect in add-on DSP by putting speakers at the
input and at the output of the add-on device. Stereo cans make the
effect downright strange.

Fast c-w QSK would seem to be a challenge, especially if latency varies
with selectivity.

Makes me wonder how the commercial "DSP rigs" perform. I'm mostly
satisfied with an Elecraft K2 (wish it had 200W barefoot, instead of
100W), and have fond memories of my TS-850's. Never met a commercial
radio I couldn't make better ;o)


Delay is inherent in most DSP systems used for filtering, because they
generally use a circular buffer for the convolution operation. It acts like
a delay line.

73, Leon


The DSP allows you to implement a type of filter called "finite impulse
response". These have some strong advantages over the type of filters
(infinite impulse response) used in analog systems, mostly sharper
filtering for the amount of ringing, and the capability to make filters
adaptable (like adaptable notching). One price you pay, however, is
latency.

Fast QSK would be interesting, but if you were willing to go without
sidetone and perhaps hear the break request while you're keying I think
it would be doable (I wouldn't know from personal experience -- I do DSP
work professionally but don't have any DSP rigs).

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com