Class AB amplifier circuit not working correctly

[Chryseus]

I recently got a pair of 0-30V 0-5A supplies so I decided to start experimenting with split supply circuits which I've been wanting to do for a while now, I made a quick class B amplifier which seems to work just fine then I decided to make it class AB which is where my trouble starts.



I quickly put together this circuit but I'm getting a very high current draw from my supplies, I set the current limit on both to 200mA but then the output voltage is not split evenly with the negative rail sinking down to 0.5V, so I'm not sure what the problem is, I've checked the breadboard layout multiple times and it seems exactly as it should be. 

[madires]

I think you need a proper quiescent current control, an input stage and a feedback loop.

For example: http://electronics-diy.com/electronic_schematic.php?id=715

[Chryseus]

Putting a 0.47 ohm resistor in each emitter seems to fix the problem, I guess the original problem was thermal runaway due to error in the base bias and variations in the transistors, well at least I've learned something new today 

[madires]

Yep, emitter resistors are also a good idea :-) Have you checked the output with or without a load (speaker/power resistor) attached? The hFE of the transistors varies. That means the top darlington may have a higher or lower hFE than the bottom one which would cause an unsymmetrical amplification.

[Chryseus]

I've tried it with both power resistors and speakers and it seems to work fairly well, this is my current design:

[madires]

The input decoupling caps are quite large, 1.0 or 2.2µF should be more than sufficient. Why have you added the output decoupling cap? Is there too much DC?

[Chryseus]

The output cap is not needed ?, yeah now that I think about it that makes sense, I'm still getting used to building circuits with a positive and negative rail 
Also I'm not exactly sure how the 120 ohm resistors actually work, all I know is it stops one side from saturating.
Probably should crack open the art of electronics and have a read.

[madires]

If the top darlington conducts the voltage at the base of the bottom one is raised via the 120 ohm resistor and the two emitter resistors in series. So the quiescent current for the bottom darlington decreases and it will conduct less. And if the bottom darlington conducts the voltage at the base of the top one is decreased via the same way, and the top darlington will get less base current. That reduces the power loss caused by the non active darlington. Otherwise the non active darlington would conduct partly based on the quiescent current (and input) and would become a load for the active darlington.

[qno]

I would rather call this a Buffer iso Amplifier.
There is no feedback.

To have a true class AB stage you should have at least 4 diodes as quiesient current source to compensate for the 4 PN junctions in the output.

The max output is 9V pp max.
For better results check out the output stages of modern rail to rail opamps for ideas.

[ElectroIrradiator]

The output cap is not needed ?

I would keep it. Without a DC control loop the output voltage will in all likelihood not be exactly zero. If you ground this point, you are likely to force one side of the amp transistors to draw excessive resting current, pouring some amount of DC power into the speaker coil.

[Zero999]

How about using Sziklai pairs?

[Chryseus]

I've been making some progress on my design, although once again I've run into some problems, the circuit (ignoring the pre-driver which I'm still working on) works fine going into a 1k load but as soon as I connect it to a speaker the whole waveform turns ugly, the best I've got so far is a relatively normal looking wave on one half cycle and a U shaped wave on the other, this is with me injecting the signal into C10. 
I'll try dig up some different power BJTs and see if that helps.

[KJDS]

What decoupling have you used on the power supplies?

Also, I suspect that if you want good output power without a lot of distortion, you'll need a far better biasing scheme. 1A out will drop 1V on your 1R resistors, so the base of the input transistor would need to be at 1.7V to be turned on rather than the present 0.7V

[Chryseus]

Ahh good point, I'm using two 0.47 ohms at the moment, would I be better reducing this value or doing something with the base bias?, I know the AoE has some good circuits for that.
The decoupling is about 2000uF from a linear power supply, the DC current draw seems to be around 20mA at +-6V although I'd need to get my meter out for an accurate measurement.
I'll try increase the supply decoupling and go around checking the currents and voltages in the morning.