[Alun]

Nice circuit audioguru.

How did you calculate the values you used?

What do the extra capacitors do?

That 68ohm resistor is connected accros from the input to the output of the complementary pair, as it's non-inverting this will give posative feedback could this lead to oscilation or even latchup?

 

[audioguru2]

Hi Alun,
My circuit has proper negative feedback, the transistors don't invert, they are just followers.

I very carefully matched the 39k feedback resistor to the 33k in series with the two paralleled 10k resistors so any high input current opamp can be used without offset voltage.

I also carefully matched the 20Hz rolloff of your 1000uF output cap into the 8 ohm load with the 1.5k resistor and 4.7uF cap, and the 0.22uF input cap with the 33k reference resistor. There is a trick about matching the 3 capacitors so that the amp produces a perfect square wave without tilt at low frequencies. Now I think the 4.7uF should be about 2.2uF for that to occur. Try it, changing the cap changes the tilt (overall phase shift).

Oh, oh! Both you and I forgot to include a supply bypass cap.  ;D

The 47uF cap should fully-charge (98%) by the two 10k resistors in about 1.2 seconds for good filtering.

The gain is 1 plus (39k divided by 1.5k) so a 100mV input gives a 2.7V output (about 7.6Vp-p).

The opamp drives the load through the 68 ohm resistor until 10.3mA flowsand the resulting 0.7V turns on the transistors.

 

[Alun]

So the 68ohm resistor is to get rid of the cross-over distortion my concern was that because the complementry pair has a gain of 1 the resistor would give positive feedback and lead to latch or oscillation why won't this happen?

Yes it does need a capacitor across the supply, 100uf should do.

Thanks for explaining it so well, I'll build this when I get the time.

 

[audioguru2]

Hi Alun,
The complimentary emitter-followers won't oscillate due to positive feedback into them through the 68 ohm resistor:
1) An emitter-follower isn't perfect so will have a max voltage gain of maybe 0.95 with a low resistance load. It is less than 1.0 so it can't oscillate.
2) The open-loop output impedance of the opamp is between 10 ohms and 100 ohms depending on which opamp. Its output impedance forms a voltage divider with the 68 ohm resistor further reducing positive voltage feedback through the 68 ohm resistor.

I haven't tried it for many years but I remember my little amplifier performing very well:
1) The output transistors can't have thermal runaway since the 68 ohm resistor bypasses any base leakage current.
2) Very high output voltage swing since the transistors don't have emitter resistors.
3) Very low crossover distortion since the opamp drives the load until the current allows the output transistors to be fairly linear.
4) Extremely low idle current since the output transistors are turned off.  ;D