I'm not familiar with Linrad, so on that I cannot comment.
DSP work is not done with general purpose microcomputers running the huge
overhead of an operating system and everything else. I can well believe
that under those conditions, lots of latency results.
Rather, DSP is done with specialized DSP processors. Their instruction sets
are optimized to DSP work, and so is their hardware. In the cases I
mentioned, using the processors we had at work, the latency encountered was
trivial. Certainly nothing you could hear with your ear, and nothing
approaching even 1 second.
Maybe in the rigs you mention, they have that level of latency. Maybe they
don't use DSP signal processors.
All I was stating is that with readily available DSP processors, and doing
DSP processing, at low (audio) frequencies, the processing is effective and
produces results.
Perhaps I am mistaken in believing that there is actually a DSP processor
employed in the Orion, or that they do actual DSP processing with it. Ten
Tec's deafening silence on this whole DSP issue is a glaring failure of
communication, IMHO. Hopefully we can get some factual feedback on these
points (from Ten Tec that is).
73 de Gary, AA2IZ
----- Original Message -----
From: "Bill Tippett" <btippett@alum.mit.edu>
To: <tentec@contesting.com>
Sent: Thursday, December 07, 2006 8:03 AM
Subject: [TenTec] Noise Reduction Setting
> AA2IZ wrote:
>
> However narrow the bandwidth, there is still a mixture of noise and signal
> left inside the pass band. Furthermore, properly implemented algorithms
can
> identify much of this noise and eliminate it. This is done by digitizing
> the signal, then doing Digital Signal Processing (DSP) such as looking for
> correlations for example (there are other methods)...then retaining that
> portion of the digital content and actually deleting the noise portion.
> Then we go from digital back to analog, and you have the de-noised signal.
>
> This takes a lot of processing. That is why only the military can afford
to
> do it directly on HF frequency signals. So we do it at audio signals -
like
> the 14 khz stage of the Orion. They call it IF DSP, because its before
the
> detector, but 14 kilohertz is an audio frequency no matter what we call
it.
>
> Bottom line - this should work very well. If done well, it will
outperform
> bandwidth reduction alone. Timewave does a great job of this in their add
> on boxes. I've used really expensive equipment at work to do it at much
> higher frequencies.
>
> All of the above is true (maybe with exception to Timewave).
> The problem is that all of the above takes a LOT of processing
> which adds a LOT of latency (i.e. delay before the CPU is done
> processing). Short of having a Cray computer, there is simply
> NO WAY any so-called DSP rig (IC-7800 included) can do real noise
> reduction (i.e. not simply BW reduction) without at the same time
> dramatically increasing latency (e.g. Linrad adds 6-7 seconds
> using much faster CPU's than in any rig CPU).
>
> Bottom line - If you see a QSK rig claim to have DSP
> Noise Reduction, it is little more than a fancy name for
> noise bandwidth reduction. Anything else simply does not
> compute (pun intended).
>
> 73, Bill W4ZV
>
>
>
>
----------------------------------------------------------------------------
----
> _______________________________________________
> TenTec mailing list
> TenTec@contesting.com
> http://lists.contesting.com/mailman/listinfo/tentec
>
_______________________________________________
TenTec mailing list
TenTec@contesting.com
http://lists.contesting.com/mailman/listinfo/tentec
|