Re: Comparison: K3 and SDR 5000A

Posted by Richard Smith-23 on Oct 11, 2007; 4:14am
URL: http://elecraft.85.s1.nabble.com/Comparison-K3-and-SDR-5000A-tp454268p454283.html



>> ...  The sample-in/sample-out may work at steady state, but if you use an
>> FIR filter, it still takes a number of samples and time to fill the
>> filter pipeline, as well as empty it...
>
> Exactly.  As I stated, you can't get away from the delay in the filter. I
> was describing *additional* delay beyond that of the filter.
>
But the architectures are different, and can't be compared in that way.  In
each case, the sytstem requres delay to capture the data.  In the case of
the 5000A, once the data is collected, then the "filter" (which is an FFT)
is free to run all its calculations at the full bore speed of the processor
unconstrained by the sampling rate.  In the K3, the "filter" is a serially
stepping FIR filter very much constrained in its calculation speed by the
sampling rate so must wait between samples for the next input point to
arrive before it can pop out a new output point calculation.  In the 5000A
the main delay is up in front of the filter, as you point out, but in the K3
the main delay is in the filter itself caused by the need to push all that
data through the FIR filter at the sample input rate.

> If the system is sampling at 48 kHz, a sample-by-sample architecture will
> have 20.8 *micro*seconds delay before the subsequent filters, etc., see
> the signal.

Correct, but it will also take 10.6 *milli* seconds to push that data
through a 512 tap FIR filter to detect the edge of a CW signal.

>
> If it is block based, and the block size is 1024 samples, then at the same
> 48 kHz sample rate you'd have 22 *milli*seconds of delay before the
> subsequent filters, etc., see the signal.

Actually a more realistic sample rate for the 5000A is 192 kHz.  At that
rate, you have 5.3 milliseconds of delay, not 22.
Looks like we are in the same ballpark after all.

>
>
> Enjoy!

I am.  Thanks for exercising my grey matter.

73,

Rich W1EZ

_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com