[Elecraft] ALC implementation

[Greg Troxel]

Don Wilhelm <donwilh at embarqmail.com> writes:

> The above method works fine for many transceivers, but NOT for
> Elecraft which measures the actual power output, compares it with the
> set power and adjusts the RF drive accordingly.  This is a fully
> closed loop system.  As far as I know, Elecraft is the only amateur
> transceiver using that type of system.
>
> With Elecraft transceivers (except for the K2), adjust the audio drive
> level to produce 4 bars solid on the ALC meter with the 5th bar
> flashing (this is the NO ALC point.  Then leave it that way and adjust
> the desired power with the power knob.  Doing otherwise will result in
> low power at the start and power hunting as the transmission goes on.

I have always been a bit unclear on what the Elecraft ALC/power was
really doing.

You mention low power then then hunting, and the notion of a set power
level, measuring, and adjusting gain.  That makes me think that there is
an initial gain that is computed/calibrated somehow, that if right will
not result in an adjustment right away.  Presumably that's a known
relationship between an audio input of an appropriate power and the
desired RF level.  Is that a fair characterization, or if not, could you
explain how it's wrong?

I am guessing also that there is some kind of silence detection or max
gain, as keying the mic without speaking doesn't lead to ramping up to
full-power transmitted noise.  I could see the power loop having a
narrow range, enough to account for how much variance there should be,
but not 40 dB worth.

I have seen repeatedly about "4 bars solid and 5th flashing" as the
right place, and the "no ALC" point.  Is there really an ALC circuit,
that reduces gain in the audio stage when a higher input level is
measured?  Does this have a sufficiently long decay time constant so
that an FT8 or PSK signal would get adjusted acceptably and still have a
low-distortion signal?  Or is there still a transfer slope where the
output ends up too high?

Overall, this seems like having two coupled control loops instead of
one, and I'm guessing that the point is to have different time constants
in each one, or to get the levels right at two points in the circuit
instead of one, by having audio gain and RF gain separately controlled.
Is that accurate?

Separately, I wonder about issues with different levels of different
audio frequencies resulting in a transmitted signal which while not
necessarily distorted is off plan and therefore less well decoded
(assuming there isn't adaptive equalization in the decoder after pilot
tones, and I don't have the impression there is).  Does this end up
being an issue in practice?  I would hope that with the relatively close
spacing within an FT8 signal that this is likely to be not a problem.

tnx es 73 de n1dam