Class AB amp problem

[pete20r2]

Hey guys, just signed up to the forum, long time follower.
I have just undertaken a little project to build myself a class AB amplifier, simple/cheap as possible.
I have a working model (on a breadboard) but I have run into a problem.
The design uses some chunky transistors (TIP41/42) but I cannot utilise their full potential because I cannot increase the voltage due to the presence of an op-amp.
I cannot figure out why I am getting no voltage gain in either the class A or class B sections of the amp, just the base-emitter voltage drop.
I have a power supplies lined up to supply ±24 volts but no op-amp will be capable of handling this.
I realise also that this circuit has no DC rejection, I'll fix that later.
Any suggestions as to how to remedy the circuit, do I eliminate the opamp and use a tong tail pair? If so how would I implement negative feedback?

The voltages graph shows the voltage at the base of Q3(green) vs the voltage at the base of Q2(blue).
Thanks

[Zero999]

Low gain is the least of your problems.

Have you measured the quiescent current? The output transistors will smoke!

[pete20r2]

Have you measured the quiescent current? The output transistors will smoke!

Where is the error in the circuit?
I mean, I have this thing built, I am listening to it now, it's not hot.
I don't mean to be rude, but do you have any suggestions for the problem I described?

[Paul Price]

I am trying to be positive but you really need some negative feedback!

Negative feedback is even good for a T-Shirt.

Be generous, take some your output and turn it upside-down and feed it back to the input.

You have nothing to gain with this circuit. AC/DC, better start with the former.

Look, if you can't get voltage gain, isn't current gain important? Isn't that gain enough for you?

Waste of a good op-amp, output is perverted is the inputs are inverted.

A grounded-base transistor stage allows for voltage level translation.

Learn how to use Google, then amp design.

Look at how others have solved this problem by looking at tons of available internet information, such as tutorials, AB amp schematics and amplifier IC datasheets and compare their circuits with your own ideas.

[pete20r2]

I am trying to be positive but you really need some negative feedback!

You have nothing to gain with this circuit.

Look, if you can't get voltage gain, isn't current gain important? Isn't that gain enough for you?

Waste of a good op-amp, output is perverted is the inputs are inverted.

As funny as it is, you got me all excited when my phone stated beeping, only to come and see this. Maybe if you wrote at least something useful.

[fcb]

Try and take the feedback from the actual output of the circuit - you may hit some other problems (loop stability etc..), but try it.

As for long-tailed pair etc... probably best to look at a few dozen schematics for PA's for hints.

[fcb]

I cannot figure out why I am getting no voltage gain in either the class A or class B sections of the amp, just the base-emitter voltage drop.

If your referring to the transistor stages - they are setup to provide current-gain in your schematic.  Typically a PA will have a front-end (long-tailed pair, provides voltage gain and NFB), followed by a voltage follower, followed by a pre-driver (current-gain) and then an output stage (current-gain).

[Andy Watson]

The base current to drive the output transistors is being limited by R3 and R4. The current gain of the TIP4x is probably only of the order of 20-40. You will need significant amounts of base current to drive them. Once you've sorted that you will have a much better chance of seeing the predicted smoke release.

[Zero999]

Where is the error in the circuit?
I mean, I have this thing built, I am listening to it now, it's not hot.
I don't mean to be rude, but do you have any suggestions for the problem I described?

Oh yes, sorry, you're right, the output stage is an emitter follower, I misread it.

As funny as it is, you got me all excited when my phone stated beeping, only to come and see this. Maybe if you wrote at least something useful.

No, he was trying to be helpful, you need the output transistors in the op-amp negative feed back loop.

[c4757p]

Why's it always beginners making snide comments about "maybe if you wrote something useful"? You know - the ones who need help the most and are least in a position to pick and choose what they get?

Take off your electronic leash and you won't have an excitement fit every time it beeps...

[Bored@Work]

Why's it always beginners making snide comments about "maybe if you wrote something useful"?

That is just the first part of the mystery. The answer to that first part could be spoiled rich kid used to push mom, dad or servants around.

The second part of the mystery is why are there always people who try to spoon feed the rich spoiled kid even after it insulted him or one of the community? Some even try extra hard to satisfy the spoiled rich kid when it gets agitated. As if they love to be pushed around by spoiled rich kids.

[c4757p]

Oh come on... where'd you get "spoiled rich kid" from...?

[pete20r2]

Guys, chill, I'm not having a go at anyone but it makes it difficult when the answers are so cryptic.
I'd also appreciate if you'd stop making assumptions about me, I'm trying to be nice.
I have had a look at some of Rod Elliott's designs and he seems to use the long-tailed pair, then a voltage follower, then what's in my circuit.
His designs look a bit more complex, I don't really want to change my design as I like the simplicity of it.
I tried to simulate a long-tailed pair to no avail, I will try and get some screenshots of it.
I have also adjusted the bias resistors, they are 2 * 1/2W 1k and the transistors are heat sunk.
There is still a lot of bias current required to get a decent voltage out though.

[DaveW]

There are example circuits wih some reasonable explanation here, http://educypedia.karadimov.info/library/Chapter5.pdf; have a look at the suspended supply amplifier

[madshaman]

Any reason you're opposed to introducing a voltage gain stage?

[Andy Watson]

I have also adjusted the bias resistors, they are 2 * 1/2W 1k and the transistors are heat sunk.
There is still a lot of bias current required to get a decent voltage out though.

Why don't you put away the spice and the breadboard for half an hour and sit down with a pencil, paper, calculator, transistor data sheet and a bit of ohms law. Calculate the currents required in the output stage at the full excursion of the signal voltage - it should become obvious why you're unlikely to see more than half-supply voltage driven into the 4 ohm load.

I'll save you the effort of dragging up the data sheet and tell you that the large signal current gain of a TIP4x is at best 105.

[fcb]

Any reason you're opposed to introducing a voltage gain stage?

The OP already has a voltage gain stage - the opamp with the 560/50K NFB network.

[c4757p]

Yep, I agree - put down SPICE, get out a pencil and do the math, I think you'll see your problem.

A couple points, not 100% relevant to solving your problem...
1) If you are going to use SPICE, could you post the actual SPICE file as well as a screenshot? Sometimes it's nice to be able to have a quick probe around and try a couple changes.
2) You might want to reduce R5 and R6, very high resistances tend to be more subject to noise. I'd use 10k and 560.
3) Are C1 and C2 really supposed to be 100mF? As in 0.1 Farad?
4) Just for simplicity - an ideal capacitor in parallel with an ideal voltage source is redundant, so if you're going to use ideal voltage sources (not filling in the series resistance), you might as well omit the capacitors to clean up the schematic a bit.
5) On that same note... it's not entirely uncommon for amplifiers to become unstable through feedback coupled in the power supply. You might want to introduce some series resistance to the voltage sources to help catch that. In general, using ideal anything is a recipe for a happy simulation that runs perfectly.... when the real circuit does not.

[pete20r2]

Yep, I agree - put down SPICE, get out a pencil and do the math, I think you'll see your problem.

A couple points, not 100% relevant to solving your problem...
1) If you are going to use SPICE, could you post the actual SPICE file as well as a screenshot? Sometimes it's nice to be able to have a quick probe around and try a couple changes.
2) You might want to reduce R5 and R6, very high resistances tend to be more subject to noise. I'd use 10k and 560.
3) Are C1 and C2 really supposed to be 100mF? As in 0.1 Farad?
4) Just for simplicity - an ideal capacitor in parallel with an ideal voltage source is redundant, so if you're going to use ideal voltage sources (not filling in the series resistance), you might as well omit the capacitors to clean up the schematic a bit.
5) On that same note... it's not entirely uncommon for amplifiers to become unstable through feedback coupled in the power supply. You might want to introduce some series resistance to the voltage sources to help catch that. In general, using ideal anything is a recipe for a happy simulation that runs perfectly.... when the real circuit does not.

Similar to what Mr (Det.?) Watson said.
I'll do the calc's soon, I am busy with work/uni atm.
The goal of the amp was to make a simple amp with +-12 supply to drive a minimum 4 ohm load.
If I was using different transistors with higher gain I might able able to drive the output stage directly from the op-amp, which would compensate for the crossover distortion.
I used my sneak moves and got some "sample" LM4532NA op-amps. Unfortunately I have not got the Spice model of it working, should be no different to the current op-amp within it's limitations of course.
http://www.ti.com/lit/ds/symlink/lm4562.pdf
SIM files attached.

[c4757p]

You can use diodes to give a voltage offset to overcome the crossover distortion - generally preferable to relying on the op amp being fast enough. To avoid thermal problems, thermally couple them to the output transistors so that if the output transistors get too hot, the diodes will lower the bias.

If I was using different transistors with higher gain I might able able to drive the output stage directly from the op-amp, which would compensate for the crossover distortion.

Remember - you're using some pretty beefy output transistors, which usually means low gain. But the transistors driving those do not have to be the same type. You can use a smaller and higher gain transistor for that purpose.

Hint, if you need it: your Q1 and Q3, of course, are offsetting the output voltage, which should help with crossover distortion. But you have a problem with your circuit topology: Q3 can only sink current, and Q2, which it is driving, requires current to be sourced to it. This means that Q3 isn't actually driving it, R4 is - Q3's just pulling against R4 to block the drive. And R4 thus does two things: it drives Q2, but it also sets the quiescent current, and the V_BE, of Q3. And Q3's V_BE sets the quiescent current through the output transistors. Basically, the more you drive them, the more they need to be driven. They'll never be happy. You need to completely rearrange the area around Q3 and Q1.

[pete20r2]

You can use diodes to give a voltage offset to overcome the crossover distortion - generally preferable to relying on the op amp being fast enough. To avoid thermal problems, thermally couple them to the output transistors so that if the output transistors get too hot, the diodes will lower the bias.
.......You need to completely rearrange the area around Q3 and Q1.

I have only just started amp design, I was under the impression the whole purpose of complementary pairs was to eliminate crossover distortion. Surely the Base-Emitter matching is going  to be closer than to the drop across a diode?
Should I get diodes working as close as possible but keep the output with a flat spot (no cross-over) and let the op-amp clean up the scraps? That way I can remove the two 1 ohm protection resistors on the output.
Attached is a revised version, for testing, the working model would have Q2 and Q4 joined to a single 4 ohm load and the feedback for the op amp taken from that node.

[c4757p]

Looking much better.

Give the driver transistors a bit more current. Instead of R3 and R4 out to the rails, use small (220-ish) resistors from Q3 and Q4's emitters to the output. (There'll only by 0.65V across the resistor, so it won't draw too much current).

If you didn't have the driver transistors, V_BE would be 0.65, and V_D would be 0.7V, so you'd actually have a bit of extra bias, completely eliminating the crossover distortion. With driver transistors, though, the accumulated excess bias usually ends up being too much and the quiescent current will be too high (if you use the equivalent of four diodes for four transistors). I suggest using the attached circuit to give a variable voltage drop, rather than a series of diodes. Make R7 a trimmer. This will hold a reasonably constant voltage across it, continuously adjustable. If you thermally couple it to the output transistors, again, it will drop the bias in case of excess temperature.

If you're careful with your computations and use 1% resistors, you may be able to eliminate the trimmer as well.

I wouldn't remove the protection resistors if I were you. The op amp will compensate for them and they'll prevent the output from ever being a dead short.

Your output is clipping because the gain is way too high. You're trying to multiply a 0.55Vpk sine wave by 17.9, giving 9.8V, which you don't have the headroom for. Reduce the gain or the input amplitude. (Or increase the power supply if you truly want 12W output into a 4 ohm load.)

[pete20r2]

I am sorry, try as I might, I cannot get that voltage dropper circuit of yours working.
It looks like you already have it on SPICE, can you post a simple working model?
The output I get is a very low amplitude wave with a huge dc offset, the resistor values don't seem to affect it?

[madshaman]

Any reason you're opposed to introducing a voltage gain stage?

The OP already has a voltage gain stage - the opamp with the 560/50K NFB network.

Apologies, I do realise that.  I only glanced at the OPs problem and the schematic; no values, when I look at the opamp I see a preamp to provide input impedance and pull the signal off the noise floor, if I'd actually started looking at values I may have realised additional voltage gain was unnecessary.

[c4757p]

Can you show me how you connected it? I won't be at a real computer for a while, but I could look at a screenshot.