On 21 Dec 2006 16:54:24 -0800, "Telstar Electronics"
wrote in
. com:


Frank Gilliland wrote:
The distortion specs for the chip are given so the designer can decide
whether the chip is good enough for the studio or stage; for CB radio
it's overkill. Regardless, you missed the point once again: audio
compression -IS- distortion. IOW, hook it up to a distortion meter,
crank it up to 15:1, start talking and that needle will bounce all
over the place. The chip also has some other issues that you really
need to research first.

Want a decent compression circuit? Start with a bucket-brigade chip
like the old SAD-1024. That way the control signal can be adjusted for
a peak -before- it slams into the VCA. This eliminates the need for
frequency compensation, which is a common problem with real-time and
feedback types of compression circuits, including the SSM2166. It also
allows you to get higher compression with no distortion from spurious
harmonics that would otherwise need filtering. And the circuit has a
couple other benefits: it reduces feedback which is great when some
dope modifies his radio for talkback, and you can use the chip to make
some really cool sound effects such as echo and reverb.

But I suppose it's easier to hack someone else's design than learn
things for yourself, isn't it, Brian?

Now Frank... how could they allow a chip for studio or stage (as you
put it) that had audible distortion. The answer... they couldn't.


The answer... the chip is most often used for gates and CVAs, not
compressors. Didn't I tell you to do your research?


I'm
talking harmonic distortion... not compression distortion...


Like I tried to tell you before, harmonic distortion is a consequent
of compression and doesn't go away with a magic chip.


which you
can't hear anyway.


You most certainly -can- hear compression distortion. It saps the
dynamics of the sound. How much dynamics is lost depends on the amount
of compression.


As far as your frequency compensation... I have no
idea what you're talking about...


Imagine my suprise.


but that's not surprising with all
your techno-babble.


Well, here's some more techno-babble:

Real-time or feedback type compressors such as the SMM2166 use a
comparator to create the control signal and results in a "knee" in the
response curve. But any curve that has a knee is non-linear and
results in harmonic distortion (same as "compression distortion" in
your RF amps). Because the harmonics extend beyond the limit of your
audio bandwidth you must therefore limit that distortion with filters.
Since filters are non-linear with respect to frequency, and since
audio is an extremely wideband application, some part of the loop
(usually the preamp) must compensate for the diminishing high
frequency response (yes, even when the audio is limited to 3kHz).
That's called "frequency compensation" and is something that must be
included in the design of any compression circuit (or amplifier). You
can try to soften the knee by slowing the response time of the control
signal, but that results in spikes at the output because some of the
signal sneaks through during the time delay. So no matter what you do,
distortion is just an inescapable limitation of these circuits. If you
want to learn more on the subject, get an education.

And just to confuse you even more, there is another type of real-time
compression that works -=MUCH=- better than anything else. The audio
is modulated to RF, clipped, filtered, and then down-converted back to
audio. The result is hard clipping with no harmonic distortion. Pretty
slick, huh?