The Alexander amp - current feedback and opamps

[rtarbell]

I'm sorry for the esoteric question here, but I asked Analog Devices directly and they have yet to respond. My two questions are about the Alexander Power Amplifier from an Analog Devices Application Note:
AN211
http://www.analog.com/UploadedFiles/...lexander&la=en

Just to give a brief heads-up, it is a current feedback-type power amplifier that also uses the power supply pins of opamps to carry the signal in the input section of the circuit.

1) I understand why A2 was added to the circuit (to reduce the DC offset at the output of the power amp). Even though A2 is specified to be a very low offset opamp, how can we be sure that it will contribute a low offset when we are using its power supply pins as outputs (instead of its conventional output pin)? How can we be sure that if we use the output pins themselves of a low offset opamp, they will contribute to a "low offset" of the overall audio amp at its output?

2) The author stated that the two compensation caps (C6 and C7) were originally placed from their respective nodes to ground - then they were later removed from ground and tied together at the output of opamp A1. Did the author have insight and prior know-how to do this, or was it a mere guess and check approach that happened to provide better results?

[PMA]

One of the most famous here:

Current feedback - Voltage feedback, how do I see the difference?

[unclejed613]

if the op amp is low offset from output to ground, it's reciprocal is also true, that the differential from output to power supply pins is also "low offset". op amps get their "ground" reference at half the difference between supply rails, so what "low offset" actually means is that the output pin is very close to the midpoint between the supply rails. the op amp still behaves in the same manner internally, but it's externally measured behavior is opposite, since it's the reference that's being changed rather than the "signal". in a way it's very similar to the functioning of a QSC amp, where the speaker output is taken from the midpoint of the power supply, and the amp output "drives" ground (actually it floats the power supply rails using ground as a "leverage" point).

[Tim__x]

Opamps don't care about ground, and they don't get their ground reference from half the supply voltage. Have you noticed that opmaps don't have ground pins?

Opamps care about the differential voltage between input pins. The offset of an opamp is determined almost entirely by the input stage, all other sources of offset are within the feedback loop and as such are negligible.

[unclejed613]

op amps however do stabilize their output voltage based on one of a few references, one of them being the midpoint between the power supply rails, and that's what i was referring to.

[Tim__x]

They don't though.

The output stage has no natural output voltage, no voltage towards which it tends. Within its operating limits, an opamp's output voltage is defined by its input, the opamp doesn't care if one of those inputs is labeled ground by the designer.

Think about PSSR.

Once again, offset voltage is overwhelmingly a function of the input stage. Edit: Even when the output is taken from the supply current.