this class a amp is good?

[tadiam1]

see

[ontoaba]

May be not, better looking for another.

[Elvee]

This amp suffers from a number of flaws, all fatal:

It uses an opamp, and with supply voltages of +/-26V, it will fry. Power supply voltage could be reduced, but then the output power will become miserable.
That's the usual dilemma when using standard linear ICs for straight amplification.

A resistor is missing between B-E of VT4: without it, VT1 will not be able to split properly the drive between the two OP transistors, and the "virtual complement" effect will essentially fail.

There are no degeneration resistors, and not even a thermal compensation. And the bias source is not even stabilized against supply variations.
That's the straightest route to thermal runaway.

In summary:

a load of

[ingenieus]

Perhaps those resistors in series with the opamp's power pins will drop the voltage to safe levels. The problem is easily solved with a high-voltage opamp like the OPA551. The circuit has some possibilities and seems elegant in its simplicity. If the thermal issues can be managed, we are onto something. How about a regular Vbe multiplier and MOSFETs on the output?

[ingenieus]

Having stated that this circuit has possibilities, I took a good look at it. Apart from the things that Elvee mentioned, there are a few more problems.

The voltage gain is only 4.5 - why?

What is that squiggle next to R7? Surely it is not earthed at both ends?

Shouldn't VT2 have an emitter resistor like VT3?

Specifying the feedback cap as 1 to 5pF is a bit of a cop-out.

That inductor had better be made of thick wire to handle the current. Is it enough of a compensation network?

Those 1k resistors bring the op-amp supply to +21V, which is still over the limit of +20V for the LT318A.

However, the whole thing does not seem to work in LTspice. It converts a sine wave input into a square wave with an amplitude of +1V.

It still needs a lot of effort.

For now, I come the same conclusion as Elvee.

[Bigun]

wow, I'm getting rusty at this, I didn't spot almost any of the issues stated.

It looks like a JLH 1969, with the phase splitter switched from npn to pnp to give unity gain, so the output is essentially a Class A buffer. JLH didn't have any emitter resistors in the output devices of his amp either and providing it has adequate heatsinking nobody who built it had thermal runaway problems that I've read about. If I were copying the JLH of course then R5 would be 8k2 - could be a simple typo.

What I don't like is that input is dc-coupled, doesn't leave any margin for error if the source has any dc on it's output. And clumsy approach to limit voltage to opamp by using 1k dropper resistors in each rail.

I know use of inverting input is popular in some circles, but then the global feedback loop is more complex. which I don't like too much. Putting the signal to the non-inverting input is safer in that regards.

I believe people have built this kind of output stage, some have even applied tubes to the front end.

[AndrewT]

Maybe it's intended as a ClassA headphone amp.
That would go some way to explaining the low supply rails voltages and the low gain.

[Bigun]

yes, that would make more sense.

but overall I didn't have such an issue with this design and for a DIY'er willing to experiment I would say it is worth looking at.

[ingenieus]

Here is another one. I found it somewhere on the interweb, can't remember where, and saved it to take another look at it. Now that I have, I can't make out how it is supposed to work. LTspice can't figure it out either.

[Bigun]

my guess: upper opamp drives the upper darlington.

the lower opamp, U2B takes the signal across the emitter resistor, R15, from the top Darlington and inverts it to drive the lower Darlington in order to operate the output in push-pull.

Conceptually, I don't believe it's a new approach (look at the various totem pole followers) although I never saw it done with opamps. But I think it will only work in Class A.