Rich wrote: "Would that 100% modulation refer to the increased carrier power the AM100 produces when the modulation input level is high, as well as when the modulation input is normal and the d-c input power to the final r-f stage is 100 mW?"
From viewing the schematic it appears that this is what happens. If this is the case then the carrier power will be running above 100 mW, and by implication, the "300%" number would apply as the 100% modulation of the increased power signal as if it were 300% modulation of the 100 mW carrier judged by the peak amplitude of the envelope. By increasing the carrier power the balance between the carrier power and the sidebands power would be maintained. The RC time constant of the boost circuit is 35 milliseconds so the increase in carrier power would not persist for long after the audio peak is reduced.
Neil
Searching on Internet I found something interesting about Carrier Control, just read it.
http://hraunfoss.fcc.gov/edocs_public/attachmatch/DOC-309538A1.doc
Neil wrote in post 17 of this thread: If this is the case then the carrier power will be running above 100 mW, and by implication, the "300%" number would apply as the 100% modulation of the increased power signal as if it were 300% modulation of the 100 mW carrier judged by the peak amplitude of the envelope.
Thanks for your analysis and comments, Neil.
Developing all of this, physics shows that the useful coverage area of a conventional, DSB/full-carrier AM broadcast transmit system is a function of its fully-modulated carrier power (plus other conditions).
If 100%-modulated AM carrier power increases with audio input level compared to the unmodulated value of that carrier, then so will the useful coverage area of that transmit system -- other things equal.
This apparent Panaxis design is different than MDCL (Modulation Dependent Carrier Level) techniques implemented in some licensed AM broadcast transmitters made by Harris Corp, Nautel and others, and recently permitted by the FCC as an a-c power saving measure for licensed AM broadcast stations in the US.
Some MDCL techniques actually reduce carrier power during high modulation percentages from the licensed value of the unmodulated carrier, with no practical difference in coverage area compared to a system not using such techniques. This approach is the opposite of that described so far for the Panaxis AM100.
Of course, it would be important to the FCC that AM carrier control techniques would not increase coverage areas, so as to degrade the interference ratios defined for co- and adjacent-channel licensed AM broadcast stations in the US.
Sorry that this thread is getting rather technical, but hopefully such comments will be useful.
The technical details discussed have been very enlightening as to the methods and reasons for the energy saving and modulation handling schemes, but an important distinction, already made, but worth emphasizing, is that Part 15 transmitters may not benefit in any legal way from anything so exotic.
Part 15 operation has no need of reducing energy consumption, and anything that would fly too high on the police radar is risky.
This is only a summarizing of what has already been said.

What is the average carrier power when referring to 300% modulation? Is that 300% relative to 100 mW of d-c input power to the output element of the final r-f amplifier?
And how does that reconcile with the text about increasing the carrier power for modulation "peaks" and reducing it when the modulation input drops, together with trying to "keep an overall 100% modulation at all times?"
Would that 100% modulation refer to the increased carrier power the AM100 produces when the modulation input level is high, as well as when the modulation input is normal and the d-c input power to the final r-f stage is 100 mW?
If so, then for high modulation (audio) input levels the transmitter is not being modulated to 300%, but to 100% of the increased carrier power produced by the AM100 under those conditions.
However the AM100 will be non-compliant with FCC §15.219(a) during those times.
Anybody have any insights or further information about this?