[Radio Wrangler]

Let me think that through...

+/- 45 v rails 2.5A quiescent current is 90*2.5 = 225W

For class A, we aren't forced to have a resistor or even a constant current source opposing an output device, we can modulate the current source with the opposite of the signal.

So for peak positive output, we swing to +45v and the top device passes 2*2.5A =5A
The bottom device has linearly diminished to 0 Amps as we got there. Power consumption is 5A*45v on one rail and nothing on the other = 225W still

For peak negative output the top device has diminished to 0 Amps, and the bottom device is working hard now at 5A. Still 225W, but now on the other rail.

We've lost the constant current drain effect on the power supply usually associated with class A, but we still have constant total power drain, it just moves betweem + and - rails.

So what's the output power?

We swing between +/-45v applied to the load. And we swing between +/-5A into the load.

So the optimum load impedance is 45/5 Ohms = 9 Ohms close enough for government work, as minimum voltage drops haven't been accounted for.

So 45V peak is 31.8VRMS
And 5A peak is 3.535A RMS

So that gives an output of 112.5W absolute max in a perfect class A amplifier where all devices conduct over 100% of the cycle (definition of class A) and are always in linear mode. 112.5/225 means 50% efficiency under full power conditions. Reduce power and the efficiency slides proportionately right down to zero.

2W output and the input power is still 225W which remains constant is indeed 0.89% efficiency which is right and indeed dire.

However, by having the both top and bottom transistors in the output configuration being active participants in this way, we have a 100W ampligfier, making some allowances for losses. Yes, Self's arithmetic is fine. No surprise there!

But, if you look at some of the class A designs around, one direction of the output stage is sometimes set up as a fixed current source and this halves the efficiency of the above because it reduces the output power and therefore you need to scale the amplifier up to get back to the 100W design intent.

Having a single output transistor working against a pullup resistor would be unbelievably inefficient. But there have been designs!

So Self's 1% is a good figure, calculated by legitimate means.... but it actually represents the best, and some designs are a lot worse than that.

Things go really wrong when people eschew mathematics and 'design' things on religious principles.

Given transistor amplifier's overdrive characteristics (irrespective of class) 100W/Channel is a good output rating so clipped peaks are not too common. For a stereo pair, we're looking at 450W of heat into the room, plus power supply losses.

OK in winter, I suppose.

David