It's not a real spec. Searching online found a couple of schematics
from people trying to guess what it means.
One design claims that it is a highly biased push-pull amp that limits
the opposing transistor's bias at high amplitude. This seems like a bad
Another design has two amps working together, with one amp providing
current boosting and another amp providing precision. This sounds a lot
like an obfuscated version of a typical high bias push-pull amp with
some booster transistors.
Class AA OP Preamp /Headphone Amplifier
snip clip clip ---
The "current booster" simply measures the current into the
load and injects the same amount of current into the output
op-amp's node, so that it experiences zero load current.
That's a pretty common technique, but it's simpler for the booster
(U2) to have a gain of 1 (kill R6, make R7=0), and have R5=R3.
The other trick, for serious power, is to put helper transistors into
the opamp supply rails.
You gotta load the output of the opamp and disconnect it from the
collectors to get that to work. R to ground such that Iout from the
chip reaches ~50-60% of max. A few pF in shunt with that resistor will
speed things up a bit. You can get high slew rates out of garden
variety opamps (35v/us with a 741 and 2N2907/2N2219). This is a classic
circuit published in a Fairchild App note around 1969.
The load loads the opamp. This circuit works.
R to ground such that Iout from the
A few pF in shunt with that resistor will
Of course, frequency stability has to be considered. My sketch just
shows the basic topology. There are variations.
Well, I suppose. I prefer stuff that works without/doesn't care about a
load. Save maybe a current pump.
I introduced that circuit into an undergraduate 'Instrumentation for
Physics Kids' in 1970 as a power amplifier for driving dc servo motors
and whatnot. About an hour into the 3rd lab session, somebody connected
a speaker and in about 2 hours there were turntables and speakers hauled
in from dorm rooms on every bench. They grabbed every 2N3053 and 2N4037
Unity gain is a tough problem but anything over 5 was pretty tractable
with a 10 pF across a 10k feedback resistor