Thank you for posting that!
The High efficiency Linear Amplifier type of circuit has interested me for years but I have no saturable reactor capable of QRO.
The author rightly calls it an expanding amplifier. It is just like an audio expander, the opposite of a compressor.
In another circuit, an 813 "CB leen-yar" was grid-driven and some of the RF was rectified and used to supply the screen with DC. swing, baby!
That is also an expanding amp, but it has nothing in common with the method of the article and is mentioned only as another example of a high level expander to flesh out the concept.
for an input of 1 to 5,
where carrier=1
and
4x carrier=5
outputs:
regular linear amp: 20,40,60,80,100
expander amp: 13,21,36,60,100
so it is no longer linear but some function, however, the kind of distortion causing clipping and crossover are not created. The degree of expansion depends on the circuit but eventually too much will mess up the AGC on the other end.
regular linear amp: 20,40,60,80,100
expander amp 1: 13,21,36,60,100
expander amp 2: 1,3,9,30,100 (too much swing, baby!)
There were a few 'articles' by UTC published in their ham manuals and the like, and I'd asked in 2009 if using a 3-phase plate transformer might work as a saturable reactor if the center HV winding was fed the DC.
http://amfone.net/Amforum/index.php?topic=20047.0It didn't seem to be of too much interest maybe because no one does it, but I hope this will resurrect interest in the system shown here in the article.
The only real disadvantage that comes to mind if the system is 'perfect' is the delay (in mains cycles) required to recharge the plate power supply during the start of a modulation syllable.
In the old circuits, the filter capacitances were small and even the filter inductors were smallish by today's jaded standards, so the question is, how many cycles of mains would it take?
PSUD II helps, and some of those things can charge in 2-8 cycles. For a cap input it's a 2-cycle job if inrush is disregarded. -and it can be, because the saturable reactor is an inrush limiter.
The 'assist' from the saturable reactor scheme is not a total modulation of the plate supply at all, but a partial one where it is kept rather low and progressively unleashed following the syllable audio envelope.
Some schemes have a boost autoformer after the saturable reactor because the sat. react. output may never get all the way to mains voltage due to losses and practical issues.
Other schemes eliminate the step-up autoformer by using a plate transformer with a low input voltage, such as a 150V unit operated on a 240V mains with a saturable reactor in series.
how to prevent overmodulation during that time of increasing B+, and whether linearity is a concern.
I tried with a 3-ph. plate unit from a 1500W RF generator and the results were poor but I learned these things:
If using a 3-ph transformer as a sat reactor, it must be rated a minimum of 3-4x the KVA of the plate supply or the supply might not reach full output due to its primary windings losses as they must be put in series. It is better for the 3-ph unit to have lower voltage primaries than the mains voltage; one with 120V primaries will control a 240V plate unit.
The linearity is not good from 0 to max., that is transfer function of magnetizing current through the 'secondary' winding to B+ output is not linear, so feedback is necessary if a large degree of control is wanted. If all one wishes is to turn up or down the HV, it is OK. If a strict following of the audio amplitude envelope is desired, much feedback is needed. In the experiments, I could not hear much difference but a sloppy envelope following could be seen on the scope. I don't think it matters too much in practice because it is not to strictly modulate the linear amp, only to follow the general strength of the modulation part of the signal, and that signal is really supplied by the transmitter.
The turns ratio of the 3-ph unit forced into this service is part of the problem. The transformer primary has too many turns thus too much inductance and I2R loss. The result was that a lot of current in the secondary was necessary at carrier level.
The secondary leg in use has a hard time saturating the core by itself and it is better to use all 3 of them. This makes for a series or parallel connection, but in any case it ia a lot of DC current to control with a tube. I was using 300-900mA. I blew up several transistor banks before giving that up, but I think that could be overcome. A bank of tubes might be better, like a bunch of the old octal sweep tubes or HV pulse regulators with odd filament voltages that no one uses. those are good because they conduct large currents at low plate voltages.
Beware I admit I did not finish the experiment to its conclusion at all and therefore was not successful, but I believe it would have been very satisfactory if the 3-ph unit was rated 10KVA driving a regular 2KW amplifier, 1500V transistor bank (series transistors with diode and snubber protection) were used to control the saturating winding, all 3 secondaries were connected in parallel for DC-aiding, and feedback amounting to about a 200W audio signal was applied to the saturating winding. That could be done by driving the modulating element, whether tube or transistor.
Looking back the 3-4 sweep tubes like 6CD6's in parallel (200ma each) would probably have been better, but I had a pile of old 450V TO-3 transistors on heat sinks so that is what I played with. Now I have a pile of heat sinks.
the right way, as in the article, would be to find the proper UTC or other iron or a commercial part such as from an electroplating or other big nasty power supply like an old Nobatron 28V 50A unit which consumed about 3KW.
