[Elecraft] K3 - SPE Expert 1K-FA and ALC

[Tom W8JI]

David,

Bandwidth might not be a good argument. I'm not convinced 
ALC transients are a major universal problem with SSB 
signals when they are compared to other reasons for 
bandwidth problems. This isn't the appropriate forum for 
that discussion.

I have looked at ALC overshoot because of my connection with 
amplifier designs and failures in semiconductors and other 
sensitive components caused by ALC problems, and the pulse 
duration is very short. That short pulse repeats only after 
the ALC has discharged significantly, so the transmitter's 
gain is high.

The problem is a great deal like automatic bias, where the 
leading edge always causes some distortion as the circuitry 
changes states. The deeper the bias, the more time it takes 
to remove it and the stronger the unwanted off-channel 
energy. ALC is a little worse because the control signal is 
sampled after some delay and has to wrap back around to the 
start. This means no matter what they do in the amplifier 
the cannot cure the leading edge problem. The one exception 
would be if they held the ALC voltage high and then pulled 
it down to a level that allowed proper drive. They may be 
doing that, but an indicator would be a slow power rise on 
the leading edges. It is a mirror of normal ALC.

As for bandwidth measurements, I'm not convinced the 
WaveNode is good at transient analysis. With FFT analysis, 
the window where waveform is sampled has to be present at 
the same time as the transient, and the processing cannot 
average the power. It has to calculate to provide bandwidth 
information of the very short duration peaks. The problem is 
much like the reason a conventional spectrum analyzer will 
miss ALC problems in all but a few really severe cases. If 
it isn't sampling the transient frequency when it is there, 
the extra bandwidth doesn't show.

I'm not convinced at all, because the overshoot is so short, 
it is the major problem we hear on the air....or even a 
significant one. I'm not convinced, and actually think it 
unlikely, the WaveNode could measure such a short burst.

All that aside, because it is a big discussion, we are left 
with the fact Elecraft has clearly warned to NOT do 100% 
power control with an external ALC source. That warning is 
100% understandable based on the unique way the K3 ALC 
works. It is absolute common sense the K3 ALC should not be 
replaced with a traditional external ALC for primary power 
level control. Even without Wayne's warning, I would not do 
it now that I understand how the K3 ALC works.

To prevent splatter, the K3's internal pre-filter ALC must 
have primary control of power limits. The amplifier should 
only provide a fail-safe control that pulls back or kills 
the exciter drive if limits are reached.

By the way, I'm testing new PA module designs with an IC706. 
It has terrible ALC overshoot. It has "100-watt plus" 
transients when I have it set at 20 watts to drive the PA 
modules. The K3 does not. Fortunately the MRF-150's are able 
to handle the 600% overdrive bursts from the IC706 without 
failure. I keep a couple ICOM's around just for that reason.

I cannot test for PA module survival with a bad ALC system 
in the exciter using the K3, unless I ran the K3 power on 
full and used an external ALC detector system. Then I could 
make the K3 emulate the IC-706. Why would people not testing 
things want to do that????

73 Tom