----- Original Message -----
From: "Jim Shorney" <jshor...@inebraska.com>
To: <drakelist@zerobeat.net>
Sent: Sunday, March 02, 2014 10:40 AM
Subject: Re: [Drakelist] Tweak my Peak....
OK, My opinion:
Best practice is to operate at a lever *just below* where
the ALC light starts
to flicker. This gives you the cleanest signal. It's
easier to do if you can
use the clipping action of a good RF speech processor,
such as the SP-7 or
equivalent, to limit your peaks.
Otherwise it's probably OK to adjust so the ALC just
flickers occasionally. Too
much ALC will increase the IMD of your signal. See:
http://www.sm5bsz.com/dynrange/alc.htm
On CW, definitely operate below the ALC level. The quality
of the tranmsit
signal starts to degrade when you go into ALC.
73
-Jim
NU0C
The link above makes some good points but also makes a
couple of statements I must disagree with. One is that
two-tone testing is meaningless on transceivers. Well, it
has exactly the same significance it does with straight
transmitters, namely it is a measure of the linearity of the
amplifier given certain limits. The purpose of an ALC or AVC
in a transmitter is to prevent the drive signal from
becoming large enough to cause distortion in the amplifier.
The author is right that it should not be used to take the
place of an audio processor.
There is a very great deal of valid scientific and
technical material on speech characteristics in the
literature. The importance of understanding the
transmission of speech under less than favorable conditions
has been studied by, for instance, AT&T though Bell Labs for
many decades and has received more study by the military. A
few decades ago it was shown the "infinite" peak clipping
when used with proper filtering (differentiation- clipping-
integration) would result in speech which was still
intelligible under favorable (no noise) conditions and had
much better intelligibility under noisy conditions. The
Russians did tests of filtered and limited speech that
showed that intelligibility was maintained even when the
speech was so modified that the speaker could not be
identified. The idea is that systems like our SSB
transmitters are power limited and rely on noisy channels,
at least most of the time. The nature of speech is that most
of the power is concentrated in low frequencies that do not
contribute to intelligence and the ratio of peak to average
power is very large (on the order of 10 to 15 db, some tests
show more). So, when transmitted without modification the
average power in a system is very low compared to the power
the system is capable of. So, we find the wide use of
microphones or filters that roll off the bass and peak up
the treble and filters to limit the total bandwidth along
with various mechanisms to try to reduce the ratio of peak
to average power. These can make an enormous difference in
the readability (intelligibility) of a voice signal in
practice where it is transmitted over a power-limited and
noisy channel.
Effective processing is not trivial since the processing
itself can generate spurious signals, as the article cited
points out. Also, if processing is done improperly the
intelligibility can be destroyed. For instance, in the paper
on infinite peak clipping the reverse filtering was tried.
That is integration followed by clipping followed by
differention, the results were awful. The filters for
differentiation and integration are simple RC 6db/octave
filters, not anything exotic. Note that any kind of
processing is, by definition, distortion and is audible.
Some distortion is useful some is to be avoided. In any
case, the ALC or AVC in a transmitter of whatever kind
should be used only to prevent occasional excursions beyond
the linear range of the amplifier and some external
processor used for real increase in average power and
intelligibility.
If at all possible listen to your own signal on an
independent receiver for both quality and spurious signals.
--
Richard Knoppow
Los Angeles
WB6KBL
dickb...@ix.netcom.com
_______________________________________________
Drakelist mailing list
Drakelist@zerobeat.net
http://mailman.zerobeat.net/mailman/listinfo/drakelist
