Class A Chip Amp: and now, the complementary version

[Elvee]

This amp is the complementary version of this one:
SE class A regulator-chip-amp madness

Now, the L-R correlation is used to modulate the current source, with L and R having opposite polarities (the loads must also have their polarities reversed).

This version no more qualifies as a true single-ended (it is a sort of hybrid), but it has more output power and even less (simulated) THD than the previous circuit.

An actual prototype will follow soon.

[stoc005]

I'd prefer the LT1115, for lower distortion. Why have blocking caps (C2) in feedback R's and use a low voltage offset op-amp? I'd bet they are not needed.

I hate cap. coupled outputs but I listened to one for years. Heathkit AA-15. Was OK....

How is the performance at clip? At Current limit? Difficult load (1uf)???

I don't believe LTSpice simulations for a minute of devices like this. Performance is so model dependent and many complex devices are modeled simplistically with many "features" omitted for ease of model generation.

[Joachim Gerhard]

I can not 100% understand. The circuit shows 2 x 8 Ohm resistors but has only one input.

[Elvee]

Originally Posted by stoc005

I'd prefer the LT1115, for lower distortion. Why have blocking caps (C2) in feedback R's and use a low voltage offset op-amp? I'd bet they are not needed.

Since you seem so competent, why don't you explain for what precise reason the caps are not needed, and what would happen without them?

How is the performance at clip? At Current limit? Difficult load (1uf)???

One advantage of class A amplifiers is that they can be made to tolerate loads ranging from +/-(0->∞)+/-(0j->∞j).
This may not be the case for this circuit in this form, but it could be achieved without too much difficulty.
The same cannot be said of your usual chip amp.

I don't believe LTSpice simulations for a minute of devices like this. Performance is so model dependent and many complex devices are modeled simplistically with many "features" omitted for ease of model generation.

I make simulations and measurements, if you care to read the previous topic.
The prototype for this particular circuit hasn't been built yet, but it's in the pipeline, and I will report the results.

[Elvee]

Originally Posted by Joachim Gerhard

I can not 100% understand. The circuit shows 2 x 8 Ohm resistors but has only one input.

The sim shows a stereo amplifier being fed by a common input signal.
This is just to illustrate how the circuit works by exploiting the correlation normally present between L and R signals.

I will include a "working" schematic, as in the previous topic, with the L and R channels fully separated.

One input signal should go to the top amplifier, and the other to the bottom one, and the phases of each speaker should be reversed wrt. one another.

[1543]

I have also problems to understand the circuit.
Wondering first why the phase reversed input is not shown, but you show a full stereo design, not a symmetrical input and bridged output.

Hope your "working" schematic will clear a little bit.

1543

[Elvee]

Originally Posted by 1543

I have also problems to understand the circuit.
Wondering first why the phase reversed input is not shown, but you show a full stereo design, not a symmetrical input and bridged output.

Hope your "working" schematic will clear a little bit.

1543

Previous schematic was for sim/test purposes.

Here is the way things are supposed to be connected in "real" use:

[AndrewT]

lt1056 is shown on the schematic.
What parameters are important if looking at a substitute opamp?

[1543]

Originally Posted by Elvee

Previous schematic was for sim/test purposes.

Here is the way things are supposed to be connected in "real" use:

Thanks!

One more question. To make this more usable (not only for headphone amp) it is necessary to multiply the number of regulators.
Is all to do to parallel them and use a 0.2Ohm resistor in each output?

1543

[Elvee]

Originally Posted by AndrewT

lt1056 is shown on the schematic.
What parameters are important if looking at a substitute opamp?

I just basically picked any all-purpose op-amp from the LTspice list.

From my prototypes, I use TL072 or LF353.

This already gives an excellent quality, because the power elements behave like near-perfect depletion-mode MOSFETs having a very high transconductance.

But the ultimate quality will be detemined by the op-amp: 1), because it sets the performance floor, and 2), because its loop gain will correct the remaining non-linearities from the regulators.
In short, the better the op-amp, the higher the performance.

Using a NE5532 would certainly improve things.

One more question. To make this more usable (not only for headphone amp) it is necessary to multiply the number of regulators.
Is all to do to parallel them and use a 0.2Ohm resistor in each output?

Yeah, something of the kind.
Of course, the 0.2R will reduce the transconductance of the regulators, but if a number of them are parallelled, this is not too important.
And because the amplifier is class A, and the 0.2R resistance is purely linear, there won't be adverse effects like in a class AB amplifier.

One note of caution:
This version hasn't been physically tested yet, and it is very likely that it also requires a stabilization network, like the previous ones.