[Amps] IMD

[Jim Thomson]

Date: Sat, 31 Dec 2011 22:34:17 +0000
From: Dan Mills <dmills at exponent.myzen.co.uk>
Subject: Re: [Amps] IMD


On Fri, 2011-12-30 at 09:59 -0600, donroden at hiwaay.net wrote:

> I disagree.
> 
> If the RF chain is clean, audio chain distortion will only be heard as  
> audio distorton on the main channel.

If and only if, the bandwidth of the baseband signal fed to the
modulator is band limited to the width of the main channel (Unless there
is a RF BPF further up the chain limiting the transmitted bandwidth,
often one of the RX filters is used in the transmit IF path). 

It is possible to provide all the bandwidth control at baseband, but you
have to be very careful about broadband noise, and any non linearity in
the RF chain can undo a lot of hard work with regard to envelope rise
times in SSB and data modes. 

### providing TX BW control is done all the time with ESSB.  I modified
both my MK-V xcvr's with a pair of 6 khz filter's, one at each IF. The BW
of the TX is limited by the DSP EQ set up.  That works very well.  I put
various pre-sets in the processor, such that TX BW's  of 3/3.5/4/4.5/5/5.5/6
khz can be used. 





ALC is insidious in this respect as the control is usually applied after
the channel bandwidth filters, and so the resulting mixing products are
not well controlled, particularly as the loop dynamics are not well
defined in general. 

## don't use it then.  You can get around the alc problem a couple of ways.
One is to inject a steady state dc voltage into the alc jack of the xcvr in question.
Then the pep output is limited to a precise value. No need to wait for alc Vdc to be
developed.   Another method is to do all of this externally, in the audio chain, b4 it even
gets into the xcvr.  By employing compression/limiting/distortion cancelled clipping, etc,
you can then achieve loads of 'talk power'.... and get your 1.5 kw pep output on ssb, and not have
any alc showing on the xcvr.  You can also have your po control on your 200w xcvr cranked to max
output... and limit the po to say 112.5 watt pep...and again, no alc action. 

## No rocket science here, BUT most hams are not about to employ those techniques any time soon. 
To really do this right, you start off with a real mic, not the typ ham junk, then use the mic to drive
a real pre-amp, not the typ yaesu  mic pre-amp junk.  Run it all at line level. Follow that up with a dedicated
downward expander/gate, to kill the background noise etc.  Then feed that mess into a real audio 
compressor, and not some junk you found in an old arrl handbook.  The one I use works pretty good.  The low freqs
don't  dominate the high freqs.  They run independent of each other, yet 'talk' to each other.  No background swell up,
and no hole punching.  Ok, now we finally have a constant level of audio that we can feed into downstream devices. 
Feed that mess into the digital eq, and between parametric and graphic eq sections, the audio is shaped into something we like.
Then any subtle effects can be added, like harmonics over certain ranges of the TX bw, which will add clarity and presence.
You can do all sorts of tricks like  stretch transients so they are perceived as being louder.  Then add a touch of digital reverb.
Don't laugh, I ran hundreds of tests with all of this over several summer's and several say the recovered audio went from Q3 to Q5
every time under marginal conditions.   This entire mess gets fed into a 2nd compressor..which drives a splitband limiter/peak limiter/
clipper set up.  The 3 x band limiter  can have the 3 x bands  put where I want them.  The 3 x band limiter is unique in that it's 
pep output never exceeds  a pre-set threshold.  Again , the 3 x bands of limiting all 'talk' to each other. Limiting in any of the 3 bands,
will trigger an input  threshold change in the other 2 x bands. So when u sum it all up on the output side, the peak output is not all over the map.
I can adjust the density, and change the peak to average ratio easily. Any overshoots get brickwalled  by the distortion cancelled audio clipper,
which is just a quad array of transistors.....and slightly more sophisticated than a pair of back to back diodes.  The output of all that can be
adjusted in .2 db increments.....over a 34 db range.  Ok, now you can set a precise drive level into the analog balanced modulator on a ssb xcvr.  

##  You can get limiters that will trigger off the peak, the average, or both. You can change the attack time on them. On the downward expander/gate
I can control the attack, hold, release.   Now why can't you do that with the agc on a ham RX ?   If one could lengthen the hold a bit, and shorten the rls,
then you end up with 'hang agc' .  



##  ever watch your favourite TV show,,, then when the commercial comes on, you get blown outa your chair ?  Look at the TV audio sometime on
an audio rta..b4 and during a commercial.  The peaks will all drop by exactly .5db  and the average power will increase by exactly 6db. 
 On my limiter/clipper I can increase the average from 0-15 db max.   There are various schemes out there to increase the average power with either no 
distortion, or very minimal distortion...and do it all in the audio.  Now yaesu can't even build a clean mic pre-amp that works properly,
 so I would not be holding my breath  any time soon  for detailed audio tailoring. 

## some advantages of doing all of this in the audio, outboard of the xcvr are... various configs and pre-sets can be used for different applications,
like FM/AM/USB/LSB..dx/contest mode/ragchew mode, etc etc. It's endless really.   I can put  a mic next to an air condtioning unit..and you will never hear it
over the air, even with the noise gate defeated. I set a narrow notch to take out the fan in the ac unit.  All of these various techniques that can be used on 
TX..can also be used on RX.     


IME serious CW splatter is usually squarely down to the exciter keying
envelope or hot switching the PA, unless you get really silly with drive
levels the amplifier nonlinearities have only a modest effect. 

## don't try feeding a click free  signal into a GG amp, that is biased slightly into Class C. It will click like crazy. The GG triode needs to be drawing some idle current,
IE: u have to operate it in Class AB.   Even 10-20 ma of idle current will ensure a click free signal. 

later...... Jim  VE7RF



73 2E0CHE, Dan.