ICs that provide the first stage?

[edbarx]

It is difficult to find a schematic on line that does not have something wrong with it. But this one has the input arrangement I had in mind.
http://circuitscheme.com/wp-content/uploads/2015/04/2000W-Power-Amplifier-Circuit-Diagram.jpg

The cascode's bias if from OUTSIDE the differential input. According to Douglas Self, this can somehow inject mains hum into the signal path.

I will have a look at that amplifier. Maybe, I can adapt it to what I need. The voltages are what I want.

I understand that high voltage coupled with high currents can literally melt metal and silicon chips before even realising. Therefore, I am exercising a lot of caution. The circuit has to first work using a 20V supply, then after that, I will use the higher voltages with current limiting series resistors.

 

[edbarx]

To steveu:
Could you please, post the same circuit. I cannot clearly read the component values.

 

[steveu]

2000W Class AB Power Amplifier - Circuit Schematic
http://circuitscheme.com/2000w-class-ab-power-amplifier.html#

Go to the bottom to skip the ads.

 

[edbarx]

I think, I can easily adapt that amplifier to my already made driver + power stage, which I know works, and is designed to handle 1500W of continous power. Probably, using an amplified diode across D18 and replacing R20A and 20B is enough. I must make sure the voltage across D18 is around 1.2V and stable. Using too much bias current in a Class AB amplifier seems to give very little benefits according to Douglas Self.

Thanks you for the circuit. Now, I have to find a decent seller to get the parts.

 

[steveu]

As I said, there is always something wrong with schematics on the web. I would divide the 90 Volts in half with with 120K+10uF instead of the stack of 1n4148s, or maybe a ~24V Zener if the cap causes turn-on transients. This one has only 2.95 Volts on the diff amp. That means the maximum input is about 3V peak, which only works because the gain is super high, ie 120k/33= 3636 = 71dB. Now that high gain (reduced feedback) probably also makes it stable given the extra driver stages. The driver circuit is incapable of getting very close to the rail which means it clips far sooner than it should. There is no output protection: not recommended. D18 is not a voltage reference. It is a clamp to prevent reverse biasing the two sides at clipping. There are several two-resistor tweaks, which means the value is critical and probably theoretical, ie this is not a proven commercial design. etc, etc,... You may want to download a schematic of a real amplifier from a service manual from Crown, BGW, QSC, Yamaha, etc. And for this power range, I would seriously consider class-D.