But a "step controller" postitive or negative, is not an audio amplifier. Thus not a topic for this forum.
The number of output transistors indicates it drives some sort of massive actuator like a steel rolling mill roller or an aircraft aerleon. Probably via a stepper motor.
It appears to be a rather extreme Class G "rail switching" amplifier. Generally these are used with DC motor servos. As an audio amplifier, if the distortion problems could be addressed, it would be one real beast... somewhere in the 800 to 900 watt region.
I just realized who the author was. He will claim it is good for silly power at low load impedance's. Generally speaking, it wouldn't be useful for any home environment. For a pro sound installation, I would use a number of amplifiers driving a couple speakers each (max). Even supplying power to this creation would pose as a loss of power due to the very high currents he intends. Best to keep the IR losses down by distributing the load (in multiple amplifiers).
I have direct experience with high power campus paging installations. I have one that uses three racks of amplifiers delivering in excess of 3,000 watts. Occasionally one might fail ('cause they sometimes do), and being rated for 250 watts, they can be carried by one person and plug in the outlet. You can imagine how many circuits this installation uses. Try that trick with one amplifier. Then, when it fails, the entire system will be down all at once. Then try to carry that one out for service, and hopefully you had a spare set up to swap in.
I've worked in clubs and all kinds of PA sound setups. You really do not want all that power concentrated in one amplifier. Even the amount of heat you have to deal with would be substantial.
So, build it if you want. But, ask yourself this, how will it be used? How much damage can it do to all those speakers. And finally, you could always just buy a Carver TFM45 or TFM75 if you want high power and beauty at the same time. The best part about these is that they are commercially successful designs (they are debugged and reliable). The TFM75 isn't light (being a PT-2400 in consumer dress), but it would be much, much lighter while sounding a lot better than the amp in that schematic. That and Carver amplifiers have very good protection circuits. The TFM45 (and PM-1.5t) have regulated power supplies measuring +/- 125 VDC! I generally set them for 118 VDC as they drift higher with age. The Carver Lightstar also uses 125 VDC rails, but uses down-converters instead of commutators like the amp you show, and the normal Carver amplifiers.
Its just the output section of a big class G amp. The dense set of 4 voltage rails is clearly designed to get the best efficiency, ie compete with class-D. At max voltage the SOA of the output transistors is pretty close to the bone, hence the need for 24 pairs to limit the current per device. For these power levels a bridged amp probably makes more sense for class B from an SOA perspective, but for class G the extra complexity of duplicating all the rail-switching circuitry would make bridged significantly more expensive than adding extra output devices.
Gee, Carver made that work using 4 outputs maximum. Keeping Vce low does massive things for the SOA curve. He uses 3 supply rails maximum. The Lightstar is variable, so you could call that infinite I guess. It works to keep Vce constant, which is a really neat idea that works well. The Lightstar also has a whole bunch of output transistors.
There are other similar amps that don't require the number of outputs that Glenn used for this amplifier. He didn't need that many in fact, he is nowhere near the SOA limits with this design.