[TenTec] NR - Fact or Fiction?

[Bill Tippett]

K4IA wrote:

 > However, the software still ain't right after 2 1/2 years.

         Buck I cannot argue with you.  Unfortunately it may
be the case that it is never completely "right".  That
said, the other benefits of Orion far outweigh most of
the firmware issues for my operating.  I operate contests
seriously and simply would not tolerate frequent crashes
(which don't happen often in my usage and experience).

 > I am a hold-out for the Noise Reduction of 1.371 - it stopped working in
1.372 and hasn't been right since.  NR in Ver2 just distorts and there are
plenty of posts still complaining about problems with the audio and 
other issues
in 2.055.  As good as TT is, the software is a serious shortcoming and  there
is nothing wrong with admitting it.

         This brings up a discussion we had on the beta
list and maybe it's time to make it public here.  IMHO,
NR on any rig (or outboard DSP of any flavor) is little more
than marketing hype.  Think about how NR actually works.  It
automatically builds a narrow DSP filter around the signal
of interest.  Guess what, you can do exactly the same by
directly turning Orion's BW knob to 100 Hz!  There is no magic
in DSP of any flavor.  They work by simply reducing noise
bandwidth to improve the S/N ratio.  There is NOT some
magical algorithm that extracts the signal from noise.  This
is the way all DSP's work including Linrad (the Linux-based
software EME guys use).  Don't get me started on JT65 which
actually cheats by using data exchanged by means other than
on-the-air in real-time...I can also find creative ways to
cheat via the Internet if I got a thrill out of that.

         In one of the V2 beta releases, one of the team felt
that version 2.XXX's NR was a major breakthrough in S/N
(Minimum Discernible Signal for a given noise bandwidth).
ARRL chooses 500 Hz BW for consistency across all rigs they
test but you can compare MDS at other BW's as well.  After
someone made the MDS comparison between the two firmware
versions in question, they discovered that MDS was exactly the
same!  The problem is that NR compensates the gain to adjust
for the lower noise bandwidth, so you can't really do an A
vs B comparison instantly.  The gain change fools our brain
into thinking it is helping, but when you make the actual S/N
measurement (MDS) you discover NR actually does nothing that
simply reducing Orion's bandwidth will not do as well (and
much faster to boot).

         I work lots of weak DX signals on 160m and also
hold the current record (27 uW) for copying the N2XE weak
signal beacon on 80m.  FWIW I never use NR.

                                         73,  Bill  W4ZV