Error-correction power amp: running the amp at a lower supply voltage.
A few people have asked for the necessary changes to run the amp at a lower supply voltage. The following changes need to be made for a supply voltage of +/-30VDC. That will give you an output power in 8 ohms of between 40 and 50 Watts.
Input buffer. Here, resistors R47, R48 need to be changed to keep sufficient current into the shunt zeners. Change the resistors to 1.2k ohms.
Vas stage. No changes necessary.
Output stage. A few changes are needed to keep enough current into the shunt zeners. Resistors R35, R36, R37, R38 need to be lowered. But, if they become too low, the time constant formed by these resistors and the bootstrap capacitors may become to small. So, to partially compensate for that, we lower D10, D11 to 12V zeners. Then, change R35, R36, R37, R38 to 499 ohms each. As a consequence, R44 need to be lowered to 6k.
Note that I have not tested this except for a quick DC verification. Still, I'd like to hear from you if you implement these changes.
If you run the amp at +/-30VDC in an active system, and with a speaker driver without a passive xover not lower than 4 ohms resistive, you can run it with a single output pair, if you want.
To leave out the outermost output devices (Q12 & Q13), the following components are not needed:
Q12, Q13, Q14, Q17, RV2;
R22, R24, R26, R27, R30, R31, R52, R53;
C2, C6, C13
If you use the XL spreadsheet to calculate the protection system values, note that with a single output pair you must use the single pair SOA
This comes as usual with the caveat that I have not tested this, but it should work. Let me know your experience with this configuration.
Higher power in complex loads:
Changes to the SOA worksheet: I did some tests with the amp running full power in various loads, clipping in about 50% of the time. The protection system seems too conservative, in allowing only power output up to the DC SOA curve. In complex loads, the phase shift between voltage and current causes the Vce to increase above the single supply voltage, and there power is very limited at DC. But, the complex load excursion will by definition not be a DC but an AC event. So I changed the SOA curve to be limited to DC at Vce's of about the single supply voltage, and then gradually crossing over to the 100mS SOA. This requires one zener voltage change (D2 can be 18V) and a few resistors, but I think it is worthwhile.