[hfsig] Performance of 24/192 soundcards? USB Soundcards
tom.mcdermott4 at verizon.net
Tue Sep 21 21:55:59 CDT 2004
Hi Bob (et al).
The DC coupling problem is caused by the moving-average variation
(usually called digital sum variation, or DSV). Scrambling an arbitrary
sequence does not reduce the DSV, since any pattern can be found to
counteract the scrambler (and statistically, will be found). I spent a good
part of a chapter in the TAPR Digital book on this problem, and the
statistics involved. The coupling problem is far worse than initially
imagined, and I ran eye patterns with the pole at 0.015 baud with very
significantly detrimental effects.
Coding the data (adding exra bits to compensate the spectrum) can
remove the LF spectrum. Ethernet does it with 8B10B coding. Mode efficient
codes (less overhead) are available. Coding can also add pilot carriers,
sync carriers, etc. if those are helpful. For Texnet 9600B modems, we DC
coupled everything. This then just moved the problem over to the slicer.
But it remains a problem with the soundcards. Alternatives are to
shift the received signal away from DC to a low IF. But it sure would be
nice to have flat DC to almost Fs/2 response. It would be nice to hear what
other alternatives might be.
-- Tom, N5EG
From: hfsig-bounces at lists.tapr.org [mailto:hfsig-bounces at lists.tapr.org] On
Behalf Of Robert McGwier
Sent: Tuesday, September 21, 2004 8:44 PM
To: TAPR HF Modes SIG Mailing List
Subject: RE: [hfsig] Performance of 24/192 soundcards? USB Soundcards
You analysis is correct. This is a self synchronizing descrambler. It is
essentially the anhilating polynomial for the scrambling polynomial. As
such, since it is a polynomial division, it is an "infinite impulse
response" type of configuration. This means that there is error
propagation. The SYSTEM BER does indeed degrade with a hole at zero.
I delivered James paper at the TAPR/ARRL DCC where the work was introduced
(Columbia, Md. a million years ago). He has indeed done a careful analysis
and I believe the orignal paper is still available on his web site as well
as other analysis.
More information about the hfsig