Apache Labs Community Forum

From time to time I receive inquiries about an ALC board I designed for the Anan transceivers some years back. Here is the full response:

I have been involved in the design, testing and use of Commercial and Military SSB exciters and amplifiers since 1966, so there is no part of this concept of which I am unaware.

ALC is fully depreciated and should not be used with Anan transceivers. Why?

ALC is a form of AGC where the output level of the amplifier is detected and fed back as a control voltage to the exciter to adjust its gain. Its genesis was in the era of vacuum tube systems where gain varied wildly as frequency was changed, due to resonant circuit loss changes with frequency and temperature – nothing was “broadband” back then. It was the only workable system because per-frequency gain adjustment was not possible – we lacked a gain adjustment capability that had any sort of time or temperature stability.

There always remained some overshoot, which, in addition to the initial transient, depended on the amount of excess gain and the time constants chosen, but vacuum tube PA’s had sufficient thermal inertia in the plates and grids to tolerate the very brief overloads. And IM expectations were not high, nor was transient IM a thing of observation or measurement. Contrast this with solid state PA’s where small overloads can be instantaneously catastrophic.

The advent of broadband, digitally controlled exciters like the Anan units removed the issue of gain variability over frequency, but some amplifier manufacturers continued to insist that their warranty was void unless ALC was used, and there is some small overshoot on initial transients due to the step and impulse response of band-limited systems.

The ALC board for Anan was created solely to address the concerns of owners of expensive amplifiers worried about their warranties. It worked as intended, and well, but the problem was the warrantees, not an actual operational concern.

With the advent of CESSB, which eliminates the initial transients noted above, there is now no rational reason to employ ALC, in fact it will re-introduce initial transients!

Set your power level per-band and enjoy the stability and predictability of your Anan transceiver.

The forum operators should consider making this message a sticky, and those who edit the Anan manuals should remove the references to ALC as archaic, with my blessing.

Clyde Washburn, K2UE

CESSB and RF gain variations over frequency have nothing to do with the issue. And while gain of the openHPSDR design is nearly flat over frequency at the DAC output, RF gain varies just as badly as any other design in both the exciter board RF amplifier stage and the power RF amplifier stages.

The real requirement for external ALC has to do with crest factor (peak-to-average ratio, or PAPR) in the exciter signal. On most amateur radio transceivers, the RF drive or RF output power control attempts to set a given RF gain condition. However, the designs do nothing to limit the effects of crest factor in the exciter signal. Thus multi-tone signals such as speech or certain digital modes can cause drive levels well beyond that set by the RF drive control and tested using a CW signal. These excursions can overdrive an external amplifier quite easily. Hence the use of an external ALC signal from the amplifier to limit these excursions.

The design feature that makes an external ALC input unnecessary in the openHPSDR architecture is that the internal, software ALC absolutely prevents exceeding the commanded maximum RF output power level. This is not a capability found on most other amateur transceivers. Therefore the RF drive level set and checked using CW or two-tone test signals will not vary or be exceeded under any conditions. And since modern, solid state RF amplifier designs exhibit little overshoot on transmit, that is also not a concern.

The openHPSDR software RF ALC is a look-ahead, soft limiter very analogous to what one might find in high end music studio applications. One can easily drive 3 or 4 dB into limiting with very little if any audible effects or additional IMD products (digital mode users might choose to remain at or slightly below 0dB limiting). This provides plenty of margin to allow drive levels to obtain 0dB ALC levels internally in order to support PureSignal linearization operation. No matter what, the DAC output will be maintained below the maximum, and thus crest factor effects may still be evident in terms of distortion of the DAC output (and its attendant IMD), RF drive will never exceed the limit set by the operator. Since that limit is never exceeded there is no requirement for an external ALC signal.

It is worth noting that the software ALC implementation in the WDSP software library was developed by Warren Pratt, NR0V, and is light years beyond the hard limiting ALC (a software clipper, really) that existed in the legacy, Flex version of PowerSDR. The latter, although adequate to eliminate the necessity for external ALC, nevertheless had horribly bad effects on IMD, as one would expect from a clipper. Many legacy Flex operators used external hardware or software look-ahead, soft-limiters to solve this problem in that context.

73,

Scott

"CESSB and RF gain variations over frequency have nothing to do with the issue". Huh? It was the very reason ALC came into being, but write the history any way you want. . . All of my comments were in the context of the Anan radios, which provide all of the functionality you noted. WRT CESSB, it prevents the internal AGC you noted from having to drop the gain a bit to accommodate the initial overshoots, so it gives you a bit more average level at the same regulated peak power. In other systems without such sophisticated AGC it is an actual overshoot, typically too fast for the ALC.

It's not history, it's physics and math.

The clipper style ALC in the legacy, Flex version of PowerSDR, which predates CESSB, negated the requirement for external ALC at the time of its introduction.

If CESSB had anything to do with the requirement for external ALC then external ALC would be required for modes of operation that cannot support CESSB, i.e. the "digital modes". Of course this is not true. The software ALC eliminates the crest factor problem entirely in this context.

On a related note, the combination of the leveler and the multiband Continuous Frequency Compressor (CFC) is a far superior solution to controlling crest factor in speech signals, with performance that goes well beyond that of CESSB and with substantially less audible artifacts.