[Flexradio] CW QSK ability

[Robert Dennison]

Hi guys,

I was born and raised on Telegraph Rd (true,) I never bought a mic for my
SDR, don't know psk from rtty and I have to agree with your comments but
consider...

So okay until you get that first receive buffer filled you aren't going
to hear anything on switch to receive:
 
512/48,000 = 0.01 seconds -- longest time to processing
1024/192,000 = 0.005 seconds -- shorter time to processing
512/192,000 = 0.0025 seconds -- shortest time to processing (mebbe even
reasonable)

Now consider the T/R switch has to close (mebbe 0.01sec) and  processing
has to occur after the receive buffer fills (give processing another 0.01
second, or more) and you have probably 0.0225 - 0.03 seconds minimum till
you hear something in your phones...  As computers get faster and multi
threading gets better the processing will get shorter but.. the T/R and
fill times will remain the same.

It's interesting that faster turn around is reported with 512/48,0000. 
That may be because the processing of a shorter buffer is faster and
swamps the sample time issues.  It would be very interesting for some of
you guys with good ears to try a 512/192,000 combo and see if that makes
any difference.  Of course the short buffer may be overwhelmed by the
fast sample rate.  Still I think mebbe those of you with really really
hot boxes ought to give it a try...

If the big determinant is processing time then multi-threading and more
cores may help..  Say continuously processing the receive buffer even
during transmit, then mebbe...

Otherwise only solution I can see is to use pin diodes for T/R and keep
filling the receive buffer while transmitting...................  then
throw the bad stuff in the receive buffer away and process the rest. 
That would be one heck of a trick without causing a pop!

very best regards
Rob
AB7CF


 
On Thu, 24 Apr 2008 16:58:24 +0000 Dave Blaschke <w5un at wt.net> writes:
> The best performance I can obtain for QSK is at the following 
> settings:
> GENERAL-AUDIO-Buffer: Size 512, Sample Rate: 48000
> GENERAL-OPTIONS: Process Priority: High
> DSP: TX Buffer Size: 512, RX Buffer Size: 512
> Delay (ms): 10 (minimum setting)
> 
> At these settings I see around 10% CPU utilization on the PowerSDR 
> screen.
> 
> With these settings, I am almost, but not quite there. Here is my 
> complaint. When returning from T to R there is an audible "pop" that 
> 
> tends to mask reception during Dits and Dahs. I am wondering if 
> Receive recovery is occurring quickly enough. Elimination of the 
> "pop" on transition from T to R would go a long way in smoothing out 
> 
> QSK, in my opinion.
> 
> My method of comparing QSK capability between my SDR-5000 and my 
> FT1000D is to select the frequency on a constant carrier signal, and 
> 
> then key at 20 wpm to see how well I am able to hear the carrier 
> between Dits and Dahs.
> 
> 
> Dave, W5UN
> 
> 
> _______________________________________________
> FlexRadio Systems Mailing List
> FlexRadio at flex-radio.biz
> http://mail.flex-radio.biz/mailman/listinfo/flexradio_flex-radio.biz
> Archives: http://www.mail-archive.com/flexradio%40flex-radio.biz/
> Knowledge Base: http://kb.flex-radio.com/  Homepage: 
> http://www.flex-radio.com/
> 
> 
>