I'm working on a power amp where I'm trying to get 10W across an 8Ω load off of a 100 mV pk-pk input. I've attached the most recent circuit that I've tried but I'm running into a few problems. As you can see in my attachment, I'm running through a preamp stage and a voltage follower to get voltage gain and then I'm using a class AB amplifier with darlingtons to get current gain.

Now, if I attach the function generator straight up to my AB amplifier, i get a decent output but the bottom fourth of my sine wave clips for some reason and I can't figure out why. Once, it actually slowly expanded until I had a full sine wave but then I turned it off to heat sink my transistors and when I turned back on I couldn't get a full wave again. I've tried playing with the number of diodes I'm using but if I take out or add diodes nothing changes for some reason... I've also tried grounding the point halfway between my diodes but that kills my output completely.

If I go through my voltage gain stage, I get a clean output from op amp, but the output across my load is horrible. The last time I hooked it up, I just had a DC voltage of 0 V on the scope, but I had to have been getting more than 0 V because my 10W resistor started smoking. Right now, my op amp has a slew rate of 7 V/us, should I get an op amp with a higher slew rate?

Any advice or inputs are appreciated, thanks.

 

Your R5 is too low in value; increase it to 10k.
Change R4 to 820k for now.
Remove C2 and C3.
Connect the output of the voltage follower opamp to the junction of D2 and D3.
Decrease R1 and R2 to around 3k.
The MJE702 emitter and collector connections need to be swapped.
Disconnect the inverting input of the voltage follower from the output, and re-connect it to the junction of the MJE702 and MJE802 emitters. This includes the output transistors in the feedback loop.

Another of the problems is that you have a lot of gain in the 1st stage, and the input offset is amplified along with the signal. You might get around that by AC coupling the left opamp to the voltage follower using a capacitor, and connecting the noninverting side to ground using a large-value resistor; 100k or so.

[eta] See the attached. Not perfect, but a step in the right direction.
Note that you should have 0.1uF and 1uF caps across the opamps' Vcc and Vee supplies.

 

Ok, I understand the need for the caps, the need for the feedback loop to contain the class AB amplifier, and the need for a higher R5, but why do R1 and R2 need to be lower and why don't C1 and C2 need to be there? I guess I'm still trying to understand the theory behind the class AB amplifier itself. I've looked around online and in books but every place I look pretty much just has a basic circuit diagram and doesn't go into much detail about the theory.

 

Ok, I understand the need for the caps, the need for the feedback loop to contain the class AB amplifier, and the need for a higher R5, but why do R1 and R2 need to be lower...

You need a few mA's current through the D1-D4 network, otherwise you won't have enough current available to drive the Darlingtons.
3K might be a bit low; I just kind of threw the simulation together. Static current thru Q1 and Q2 is around 500mA, which is high. You may need to change the D1-D4 diode type to 1N914/1N4148 - I don't know offhand - but the static current through Q1/Q2 should be just a few mA's.

D1, D2 should be thermally coupled to Q1. D3, D4 should be thermally coupled to Q2.

... and why don't C1 and C2 need to be there?

If C1 and C2 were in the circuit, you would wind up clipping either the top or the bottom of the output waveform, as you were observing with your circuit.

There was really nothing to pull the bias back to center (0v) except relying on your R1 and R2 divider.

I guess I'm still trying to understand the theory behind the class AB amplifier itself. I've looked around online and in books but every place I look pretty much just has a basic circuit diagram and doesn't go into much detail about the theory.

Well, do you understand that including the Darlington output in the feedback loop gives a far more precise output than if there is no feedback as in your circuit?

Do you see that the D1 & D2 diode drops are basically the same as the base-emitter drops in the Q1 Darlington? And four diodes in series are the same as the four base-emitter drops in Q1 and Q2?

The diodes have to be thermally coupled to the transistors, so that they stay at the same temperature. The reason for that is because when diodes and transistors get warmed up, their forward voltage or base-emitter voltage decreases.

 

Ok, those explanations clear up a lot for me. Thank you. The only thing I'm still a bit unclear on is the reason for the feedback being placed at the emitter junction. I know that it does clear up the signal and give me more gain, but why does it affect the circuit in that way, and how would I know to put the feedback there? I guess what I'm trying to understand is if I had a voltage follower in another type of circuit, how would I know where to put the feedback?

 

Ok, those explanations clear up a lot for me. Thank you. The only thing I'm still a bit unclear on is the reason for the feedback being placed at the emitter junction.

You want the output of the amplifier to be a faithful reproduction of what is coming into opamp U2, only with a lot of current gain. Providing feedback to the opamp from the amp output to the input of U2 causes the opamp to constantly strive to make the output at the common emitters the same as the voltage level at the non-inverting (+) input. If the feedback were simply coming from the output of the opamp itself, the output at the Q1/Q2 common emitters would not be tracking the noninverting input.

I know that it does clear up the signal and give me more gain, but why does it affect the circuit in that way, and how would I know to put the feedback there? I guess what I'm trying to understand is if I had a voltage follower in another type of circuit, how would I know where to put the feedback?

The circuit I posted is using a low-power opamp to drive the Darlington stage into following the opamps' noninverting input, and providing a great deal of current gain.

You need feedback to ensure that the output faithfully reproduces the input signal. Without feedback, you would have a good deal of distortion present in the output.

 

Ok, thanks a lot. That helped a lot.