Self Biasing PWM

Hey, I made an audio amp (class-AB, unity gain). Wanna offer your thoughts on it?

Also, the circuit doesn’t seem to be working. How long do I have to wait for it to start working?

It's working. A minute or so to self bias itself

I noticed the class AB amp. I have some suggestions. The differential amplifier stage is ok, it could do with some touching but generally it's fine. The output stage is something I'd reconsider. You have the output transistors going good. Good job on choosing a Sziklai Pair, I always make my amps with it instead of a Darlington Pair. I have some suggestions on the Biasing though. I'd find a way to ensure a more robust temperature compensation than the two diodes, and would work it up with a bootstrap networks instead of constant current sources. The bootstrap networks enable the amp to be driven almost to complete cut off, while maintaining a considerably good linearity. You can also use a supply rail of a little over half the supply voltage to power up a bootstrapped bias network because they work by lifting the supply rail above the rated supply voltage anyway. The only specific thing about a bootstrap network is that it typically lacks any PSRR and requires a stabilised power supply to work reliably. Also at these power levels it's good to have some protection circuitry in place - the least you can do is add an output current protection circuitry. Ah and I have an optional idea which you might like. You noticed in my class D amp circuit how I used an integrator to compare the output offset voltage with a preset reference voltage and force it to a null offset. You can do the same here, it also acts as a protection circuit in case one of the output transistor fails, it might be used to trigger a load disengaging circuit to prevent a catastrophic load condition.

http://everycircuit.com/circuit/6264753428365312

http://everycircuit.com/circuit/5520088173379584

Here's some ideas you can gather from these circuits.

I added some current protection, but I haven’t done anything about thermal increase of idle current. Perhaps if the diodes were in thermal contact with the main heatsink and the driving transistors of the Sziklai pair (not the power transistors, the ones driving the power transistors) were kept cool either with their own heatsink or by natural dissipation, but not in thermal contact with the main heatsink, then the idle current won’t increase excessively. That, or a microprocessor can sense the temperature and adjust the transistor bias accordingly.

I made an electronic breaker that can be used for the amplifier since it has a trip current of 30A, and the amplifier has a short circuit current of a few amps above 30A.

You can just thermally couple the diodes to the transistors, but usually at such high power levels a more aggressive thermal compensation techniques are used. For instance emitter degeneration for the power transistors is a must, also I'd couple a properly biased Transistor instead of the two diodes. Anyway circuit breakers are too complex, you can just use a fuse at the input of the supply rails, and you can limit the output current with the circuit I gave you. Output Transistors rarely fuse up when they blow, 90% of the times they cause an open circuit, so no supply rail current should typically pass through the load in a fault mode.

What is emitter degeneration? And I already made the breaker in another circuit, not implemented in the amplifier. What do you think about it? Unrealistic MOSFET? Looks good?

Emitter degeneration is when you put some emitter resistors on the transistors. It's usually done for many reasons - parallel Transistor matching, current limiting, temperature drift compensation etc

I can't find the breaker btw

Search “@jason9” and make sure it is set to “recent” and not “popular”. It should be at the very top.

Protection looks good, but I wouldn't use it on an amp. I noticed you are using a low side current sensing, and this has some unwanted implications on sensitive devices like amps or preamps etc. If you're driving a resistive load it's fine, but amplifiers may develop ground loops which is a highly unwanted phenomena in such devices. Imagine this scenario. Your low end sense resistor causes the ground rail to develop potential above 0V (a perfect ground). This would lead to noise and whatnot to build up on it and travel along the amp. Furthermore if different parts of the amp don't share the same ground potential, a differential common mode voltage could be picked up by the amp and amplified. This mostly leads to audible noise in the output

You need a high side circuit breaker. I'll try to design one and show you an example. They're typically harder to make, because high side sensing is not referenced to ground, and usually you need a small voltage drop across the sensing resistor. So when you try to amplify the voltage drop in order to drive a latch or whatever, you have to typically use a differential amp to re-reference the voltage drop to ground and drive the switching mechanism. One huge advantage is that your ground rail ramains a common rail with zero volt potential (ish) and won't cause you such problems. I've had some luck with such circuits, but could never touch up one quite to my taste.

http://everycircuit.com/circuit/4854856857944064

I’ve continued this discussion on the breaker circuit you made.