Amps
[Top] [All Lists]

Re: [Amps] ALC

To: amps@contesting.com
Subject: Re: [Amps] ALC
From: Ian White GM3SEK <gm3sek@ifwtech.co.uk>
Reply-to: Ian White GM3SEK <gm3sek@ifwtech.co.uk>
Date: Fri, 24 Feb 2006 08:33:24 +0000
List-post: <mailto:amps@contesting.com>
k7fm wrote:
>There are many on this reflector who think ALC is a bad idea.  I was one of
>them.
>
>But, I have changed my mind.  ALC is a good idea, it is just carried out
>poorly.

The real problem is that ALC is badly misused. In a typical transceiver, 
the ALC loop is used for:

1. Power output control, typically from 100W down to <5W

2. Compensating for band-to-band variations in exciter and PA power gain

3. Increasing the peak-to-average ratio on SSB (which is a form of 
speech processing)

4. Holding the output level inside the linear range of the transceiver's 
PA

5. Controlling occasional transient overshoots in the drive and output 
level.

Only *one* of these is a correct use of ALC - the last in the list. All 
the others are MISuses and ABuses of ALC, trying to make it do the 
impossible.

Even #4 is a misuse of ALC, if the ALC loop has to be in action 100% of 
the time.

The basic problem with ALC is that it amplitude-modulates the output of 
the transmitter, creating unfiltered modulation sidebands. If ALC is 
used correctly, as a back-stop to control occasional transients, that 
requires only 3-4dB of control range and a moderate slew rate. We can 
live with the modest levels of transient sidebands that creates.

The fundamentals are set out in the 2nd edition of 'SSB Principles and 
Circuits' by Sabin & Schoenike (of the Collins company). The maximum 
3-4dB of control range is their recommendation.

The misuses require a much wider control range (typically 15dB for power 
output control alone) and also a faster slew rate. In addition, ALC 
loops are often very badly designed and may even oscillate, causing 
wideband transient interference whenever the output level changes. The 
basic design fault is making too much RF drive available, all the time, 
and then trying to use ALC to reduce it on-the-fly.

Worst of all is that when there's no modulation, the ALC level typically 
'relaxes', so when the output starts up again there is a transient 
overshoot of full RF drive. This is because there's too much drive 
available, and then the ALC needs time to shut it down... but the faster 
it does that, the worse the transients it creates.

SM5BSZ has tested many transceivers for ALC faults - much more 
thoroughly than the magazine reviews do - and has come to some chilling 
conclusions:
http://www.sm5bsz.com/dynrange/alc.htm

There are better and clearer versions of SM5BSZ's work in the VHF/UHF 
DXer's magazine DUBUS (www.dubus.org) but these are not on the web.

SM5BSZ has tested both the receivers and the transmitters in a wide 
range of modern transceivers. He concludes that the receivers have got 
much better in recent years but transmitters have not - if anything, 
transmitters have slid back.

Increasingly, it's our transmitters that are becoming the limiting 
factor in inter-station interference. And a lot of that is due to the 
misuse of ALC that has been designed into them.

ALC in a transceiver-amplifier combination is even worse. You are 
wrapping an additional gain control loop around the outside of the one 
inside the transceiver, totally subverting the designer's original 
intentions (however dumb those might have been) so nobody knows what the 
result will be. The problem is made even worse by the fact that there 
are *no* industry standards for the ALC interface between the amp and 
the transceiver.  Just don't go there!


>The problem with all of the ALC systems to date is that they reduce
>power after the spike has occurred.  That is something like spanking the dog
>and putting him back in the house after he has spread his joy to the
>neighbor dog in heat.
>
>But, ALC can work if it reduces the gain before it attains escape velocity.
>That means it must decide to shut down while the voltage is building.  I am
>not sure what that would take.  If the sampling could occur quick enough to
>make a decision to shut down, based upon the rate of increase and before the
>peaks were reached, that would be the simplest.
>
It doesn't need ESP... about 10% of the effort that has gone into 
receiver design would solve the problem. Just little things like 
controlling the power level by a 'static' reduction in the level of 
available RF drive. pre-setting the gain in a low-level driver stage (or 
even in the DSP SSB generator) and storing the band-to-band variations 
in an EEPROM table.

The whole intention is to give the ALC loop almost nothing to do, and to 
use it only as a back-stop.

For the transceivers and amplifiers we already have, a partial solution 
is to use speech processing to control the RF drive level, so that the 
transceiver ALC hardly ever operates. Accept that you'll have to change 
it from band to band.

Then use a meaty amplifier that can handle occasional transients without 
generating splatter of its own.

>
>In the other hand, keeping the gain down and watching the scope is sure
>simple.

Yes - and if you can see any trace of overshoots or flat-topping, you 
can guarantee the rest of us can hear it.



-- 
73 from Ian GM3SEK

_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps

<Prev in Thread] Current Thread [Next in Thread>