Let's back up and take this in order, like we usually do. B+ enters through a 45-degree 4-gauge brass terminal block, goes through the two parallel 30A fuses and then through the two smaller transformers. These are being used as inductors rather than transformers and, in conjunction with the three 470f capacitors, should do a good job cleaning up the incoming 12VDC. The power supply is controlled by a surface-mount TL494 on the underside of the main board running at 48.5kHz. This frequency is then doubled, switching the transformer at 97kHz by way of four IRF3205 FETs (each rated at 200 watt dissipation). The output from the transformer is rectified and filtered by two 8,200f caps, resulting in rails of + and - 43VDC available to the output devices, wherever they are.
Access to the bottom side of the PCB is the key here. Tracing the circuits back from the speaker connectors and forward from the rails results in a convergence underneath a dark grey silicon pad (actually four of them) out in the middle of the PCB. Removing the pad reveals four sets of "Direct FETs," basically FETs without the plastic package. They are tiny, about 1/4" square (too small to read the numbers and find datasheets), but this is what you'd see if you could peel away the plastic from a TO-220-sized package and take the big leads off.
Now it all makes sense--there are four corresponding aluminum blocks bolted to the top plate that match up perfectly with the four sets of direct FETs. The silicon pad is to thermally connect the FETs to the heatsink blocks, and the 11 screws holding the main board in are there to insure even pressure across the devices for maximum heat transfer. Cool. The outputs are controlled by sets of proprietary ICs--one 44-pin and two 16-pin chips for each pair of channels. The purpose of these devices are to pulse the outputs on and off in accordance with the size and frequency of the input signal. That's the Class D part--and the reason some people call them "Digital" amplifiers. The actual output signal is a series of full-amplitude (43V) on and offs, like ones and zeros. The signal goes through a calibrated capacitor/inductor set, which rounds off the square edges, resulting in sine waves to the speakers rather than square waves. The output device sets are scattered around in between the low-level circuitry, but with a stated (and measured) SNR of 100dB, this doesn't appear to be a problem.
I promised more on the speaker connectors, here it is: The speaker connector jacks on the amplifier resemble a USB jack on your computer, only bigger and not keyed. That means you can plug the connector in right-side-up or upside-down. The actual contacts are an outer ring and an inner blade in the jack. The connector itself will accept anything from 16-gauge to 8-gauge cable, so the plastic molding looks a little out of proportion. The molding is marked with "+" and "-", which must be observed for proper speaker polarity. However, it doesn't matter which way you plug them into the amp. My single biggest beef with this whole idea is that I found it difficult to determine how to bridge the channel pairs from the instructions. You will connect the positive wire to only the positive side of one connector, and the negative speaker wire to only the negative side of the other connector for the channel pair. But now, which jacks do you plug them into? For the record, the odd channels (1 and 3) will use the positives, but I determined this by finding out which were the active outputs from the amp. It's not mentioned in the owner's manual. If you plug yours in and get no output, just reverse your connectors.