On Tue, 10 Jul 2007 11:08:42 -0700, Telstar Electronics
wrote in
. com:

On Jul 10, 12:28 pm, Frank Gilliland
wrote:
This 85% is by no means the average modulation... which is really
around 40% with voice signals.

It varies as much as people have different voices and speaking styles,
but it is generally accepted to be in the 20 to 25% range (peak to
average ratio from 5:1 to 4:1).

Since you have stopped your personal attack (for now), I will
certainly respond to your comments.
You are correct... I said "around 40%". You say 20-25%.
Either way... the number is far from the optimum 100% value the
VoiceMax can deliver.


See below.


The VoiceMax brings the average
modulation to 100%! That means that every voice... high or low... soft
or loud... will modulate the radio at a constant 100%.

Impossible, and you prove it yourself with your implementation of a
noise gate. You can push 40 to 50% with heavy filtering, and that's
about the best you can do, but it still causes moderate distortion.

Very possible... and the distortion is less than 1% THD.


Not possible. Older power amps were lucky to get as low as 1%
distortion -without- compression. Compression, by any other name, -IS-
distortion, whether it's by clipping or by AGC. You can filter out
some of the harmonics but the distortion remains -- and is evident by
a change in the output as you readily admit. As far as harmonics are
concerned, if you have a brick-wall filter at 3kHz you still pass
harmonics generated from fundamentals below 1.5kHz, which includes
most of the intelligible speech.

(Anyone who wants a primer in the audio spectrum should find a .wav
file of a standard 1kHz tone and try to duplicate it with their own
voice. Unless you are Mariah Carey, good luck!)

So even if you -think- you have filtered out 99% of the harmonics
generated by compression, think again. Only a distortion analyzer will
tell you how much THD remains.


Even if your "constant 100%" modulation applies only to unmuted voice,
it would be completely unintelligible. You can pick up almost any book
on radio communications and it will say the same thing.

You are correct that the 100% average applies to the unmuted signal...
Why would we be talking about anything else?
The audio signal is perfectly clear and has less than 1% THD.


Can't happen. If your peak-to-average 1:1 (which is required for 100%
modulation) then all ...=-ALL-=... the audio peaks are at the maximum.
With a 3kHz spectral limit, this would require an AGC response of less
than 20 uS, assuming you want the output to at least vaguely resemble
the input (that's 1/6000 with a very generous 10% response delay). The
chip you selected simply doesn't work that fast. You would need an AGC
amp with a slew rate at -least- 100 times faster than what you have.

Now assuming you have such a chip, and you did get 1:1 peak-to-average
voice power while muting everything else, your output would sound much
like dried ****. It's not hard to duplicate; just run a 10Vp-p audio
signal through a couple 1N34A's and filter at 3kHz. It will achieve
100% average modulation all the time, but it sounds like crap and is
completely unintelligible. And it ain't "less than 1% THD" -- the DA
says it's closer to 80% -after- the 3kHz filter.

If your processor has "perfectly clear" audio while at maximum
compression then I can guarantee you that it -isn't- running anywhere
close to 100% average modulation. 40% at best. And that's no better
than the AGC already built into the radio. On the other hand, if you
built what I already told you was a 'constant volume amplifier' (that
you confused with a 'voltage controlled amplifier') then your average
modulation is going to be closer to 25%.

Either way, your performance claims are contradictory and unrealistic
(i.e, impossible). At best they show ignorance of the principles used;
at worst, they show deception in marketing.