op amp boot strap ?

[3roomlab]

while browsing some of lymex threads, I tried to simulate this 1 particular circuit about bootstrapping. it didnt quite work out in LTspice. pic 1 (lymex did his in multisim)

but I found an alternative. pic 2
https://how-to.fandom.com/wiki/How_to_boost_the_output_voltage_swing_of_an_operational_amplifier

the 2nd version works in LTspice. but while playing with it, I find that Q2 and Q4 (of pic2) could be snipped off and the circuit still works (in LTspice pic 3).

Q1 : so what is the purpose of those 2 BJT?
Q2 : w/o the 2 BJT, is this a class AB ?

[Kleinstein]

The first circuit is more or less classical bootstrapping. However the transistors give are not good for so much voltge. It's more like 400 V max. The transistors have to withstand the maximum voltage. The current limiting for the output is only good for some 100 V.
Appart from these detail the circuit may work, but could still oscillate.

The 2 nd circuit is not bootstrapping. Here it is just a more of less classical audio amplifier to boost the OPs output. Q2 and Q4 are there ti get symmetric drive from both side - without it the output current and speed are limited. They kind to are the change from class A to class AB for the transistor.  For an amplifier one usually calls it class B or AB depending on the mode the output transistors work. In the current circuit is depends on the diodes: with 1N4001 it tends to be class B, with small diodes (higher forward voltage) like 1N4148 it could be a crude class AB.

[David Hess]

Q2 and Q4 provide push-pull drive to the base of Q6.  Without them, drive to the base of Q6 is class-A.  Either way can work depending on the performance requirements.

I have designed similar audio amplifiers with that push-pull common base input stage.  But I used a resistor between the bases instead of tying them to ground with a controlled current to make the input class-AB and to control the idle current through the following stages.  The output stage used complementary current mirrors and direct feedback between the output and common-base input stage controlled the voltage gain of the output stage.

I always worried that this design would have poor power supply rejection but it actually turned out really good.  Open loop bandwidth was about 1/3rd of the output transistor Ft so 2 MHz output transistors provided fine performance for audio applications.

A more clever design than shown might mirror Q5 to the opposite side and drive Q5 at its emitter instead of base turning Q5 into a folded cascode.

[Marco]

The Nr 2 circuit isn't exactly elegant, a later design from Jim Williams uses less transistors, but the transistor stages are class AB instead of class B and it has current limiting.

[David Hess]

There are all kinds of ways to do it.  Walter Jung gave a more sophisticated design in his book which is shown below.  A cascode is used instead of bootstrapping the amplifier.  The feedback current into the output controls the voltage gain of the cascode stage so this is effectively a voltage feedback operational amplifier followed by a fixed gain current feedback power stage.

This configuration is essentially an operational amplifier with two high impedance inputs, one low impedance input (the output), and two current outputs (the supply pins).  Feedback into the output controls the transimpedance of the discrete output stage.

[Marco]

That doesn't mean there aren't wrong ways. Bootstrap starts having problems at higher voltage gains, you can do partial positive feedback to get a little headroom but it's killer for noise.

The nr. 2 circuit was one Williams presented in an older article for pretty much the same exact niche as the newer one, it has few real positives. Slightly higher bandwidth than his newer circuit, but that could be solved by partially bypassing the resistors driving the first stage transistors.