How do you figure that Harry? Am I missing something? The output stages of most are AB, which would mean the class of operation and which transistors are actually being driven should be dependant purely on the output voltage and have nothing to do with the load impedance. Am I missing something?
I disagree. Idle current determines how much overlap there will be between two drive transistors, but does not determine where class A and AB operation happens. A class A-B output stage will only have overlap and essentially class-A operation for a very small region. The outputs of most op-amps are class AB. The output transistors are essentially emitter followers with a couple diode drops between the PNP and NPN assuming bipolar. Hence, the output current should not matter, only the output voltage.
I think you are thinking about a class-A output stage where the output current can be no more than the bias current.
Bob Widlar, the legendary fountainhead of IC op amp
design did in fact do at least one Class A gain circuit,
the LM110 (and 310), which is a high performance
voltage follower. Deconstructing its schematic is an
education, and is recommended to topology geeks.
You can, of course, externally bias an op amp into Class A
with a current source on the output, or even a resistor to
one of the supply lines. I did this many years ago on the
Forte preamp, and I have seen it elsewhere.
"Idle current determines how much overlap there will be between two drive transistors, but does not determine where class A and AB operation happens." Oh really....and what does?
"Hence, the output current should not matter, only the output voltage." OH REALLY! You are just making all this stuff up aren't you?
You need to go read some Data sheets for some op-amps. If the bias current in the output stage of an op-amp is greater than the peak current swing it is Class A by definition. Now if you want to learn what putting an external current sink or source on the output of an Op-Amp does, that is another matter....
Tangent and I are almost finished with our headphone amp project over on Headwize. The amp uses a FET cascode current source suggested by PPL to bias the opamp into class A. Note that the output is buffered, the opamp is not driving headphones directly:
Back to my original response to Harry.... I fully understand biasing output stages with a current source in order to ensure that only one of the transistors remains on all the time.
The problem I have found with this is that the output power that the transistor in the op-amp must supply can be quite high as the voltage drop across the transistor can now be twice the supply voltage and being in class-A, it is never less than a supply rail. Hence, you need to be careful that you don't overheat your op-amp. You could lightly bias the output to maintain class-A operation and then put a power class-A buffer after that. That is probably one reason why it is not done very often.
The other issue I have found is that with heating, etc., the output transistor that is in theory biased off, has a tendency to turn on as you move through the voltage swing. This did not happen on all op-amps and if memory serves me correctly was more a problem with FET op-amps than bipolar and was definately a factor of the output bias current. One day when I have time I will look more into the mechanism. The net effect was an increase in distortion.
But back to the original statement from Harry about an op-amp being in class A as long as the load is small. Assuming that the op-amp is running from bi-polar supplies and the load is connected to ground on one side, then the two output transistor are likely to be switching on and off dependant on what the voltage level is and hence not operating in Class-A. If there is some other mechanism in play, I would like to hear about it.
.... and see, I did not need to take any personal attack on anyone and their ability to read engineering texts, etc. to make my point.
I fully understand biasing output stages with a current source
You are as confused as a barking cat and I don't know where to start or how to explain this. I will agree that the power dissappation for one of the output transistors is increased and the complementary transistor is turned off with sufficient current from the external current source. Voltage has nothing to do with Class A bias per say and the rest of your discourse lost me completely. Does this make sense to anyone else?
"One day when I have time I will look more into the mechanism."
We too will wait for that day.
Some people around here are totally against all forms of cascodes. They complain about it affecting the sound or some such thing... probably does. Others admit to using cascodes everywhere in everything saying either they dont remember how it sounds without them (it's been so long since they built a circuit without a cascode in it) or that cascodes are the only way.
I personally would like to remain somewhere in the middle believing that cascodes have a place in the tool box but that they are not the be all and end all. I try to take this (and many other things) into consideration when designing a circuit.
Harry likes cascodes but dont mention folded cascodes to him.
Go ahead mention them. I used my first folded cascode in an audio circuit over ten years ago. AudioFreak will be happy to explain the differences in folded cascodes and regular cascodes and do a much better job than I can.