What's the benefit for bootstrapping op amp output for driving loads?

[ELS122]

What's the advantage over just using the bootstrapping amp alone?
Even if they had chosen a better spec'd but low current output op amp to pass the signal, wouldn't slew rate, noise, etc. still be limited by the bootstrapping op amp?

[David Hess]

Do you mean IC807(1/2)?  That drives the bridge excitation to produce a singled ended output and since it operates in inverting mode, contributes no common mode error.  The alternative is to use an instrumentation amplifier to convert the bridge's differential output to single ended.

A similar configuration is used to lower the distortion of Wien bridge oscillators.

[ELS122]

Do you mean IC807(1/2)?  That drives the bridge excitation to produce a singled ended output and since it operates in inverting mode, contributes no common mode error.  The alternative is to use an instrumentation amplifier to convert the bridge's differential output to single ended.

A similar configuration is used to lower the distortion of Wien bridge oscillators.

Yeah well what's the benefit? IC805 will still produce whatever common mode error there is, regardless if the bootstrapping/servo op amp (IC807) has any common mode errors, if it had, it would be added onto the already existing common mode errors of IC805.

[David Hess]

Never mind what I said; I read the connections of IC807(1/2) wrong.

It is adding to the current output of IC805 to drive R837.

[ELS122]

Never mind what I said; I read the connections of IC807(1/2) wrong.

It is adding to the current output of IC805 to drive R837.

Yeah but why? they could've just paralleled the op amps if they lacked current

Besides, Technics didn't just use it for the headphone output amp, they used it everywhere:

[Alex Nikitin]

It is the stupid "Class AA" circuit from Technics, with the idea of improving the sound quality. In reality (at least in this particular implementation) it sounds like crap and the best you can do to this output amplifier is to disable this "AA", by cutting R839 and R841.

Cheers

Alex

P.S. - the same for the all other "AA" circuits, just replace the horrible 5238 with something better, perhaps even the old but still decent NE5534, only for the main opamp of each pair.

[jwet]

Thanks learned something new today!  That is a dumb circuit, one of many Hi-Fi BS ones.  I thought Technics was better than that.

[Kleinstein]

The circuit is in a way that the 2nd amplifier drives nearly all the current. To some degree this could improve the linearity, but this would need OP-amps in deparate cases to avoid thermal effects and some biasing current for the 1st op amp.

[ELS122]

It is the stupid "Class AA" circuit from Technics, with the idea of improving the sound quality. In reality (at least in this particular implementation) it sounds like crap and the best you can do to this output amplifier is to disable this "AA", by cutting R839 and R841.

Cheers

Alex

P.S. - the same for the all other "AA" circuits, just replace the horrible 5238 with something better, perhaps even the old but still decent NE5534, only for the main opamp of each pair.

Maybe explain what it's meant to do instead of just saying "OH HIFI SHIT, DONT BOTHER LEARNING ABOUT IT"

[magic]

Never mind what I said; I read the connections of IC807(1/2) wrong.

It is adding to the current output of IC805 to drive R837.

Yeah but why? they could've just paralleled the op amps if they lacked current

An unloaded opamp has less distortion than a loaded one, so it may be better not to share load equally.

That being said, it's not clear if this scheme is any better than the old trick of putting the "power" opamp in the feedback loop of the "driver" opamp as a voltage follower.

[ELS122]

Never mind what I said; I read the connections of IC807(1/2) wrong.

It is adding to the current output of IC805 to drive R837.

Yeah but why? they could've just paralleled the op amps if they lacked current

An unloaded opamp has less distortion than a loaded one, so it may be better not to share load equally.

That being said, it's not clear if this scheme is any better than the old trick of putting the "power" opamp in the feedback loop of the "driver" opamp as a voltage follower.

So the unloaded op amp would sort of "fine tune" the waveform while the bootstrapping op amp drives most of it?

[Kleinstein]

So the unloaded op amp would sort of "fine tune" the waveform while the bootstrapping op amp drives most of it?

Yes that is likely the idea.

Compared to the driver OP-amp insider the loop, the auxiliary OP-amp to drive nearly all the current may have an advantage with the stability. At least in the simple form an extra amplifier in the loop needs to be significant faster than the outer loop - not sure in how far this also applies to the current driver.

[David Hess]

So the unloaded op amp would sort of "fine tune" the waveform while the bootstrapping op amp drives most of it?

That is sort of why this is done.  When an integrated operational amplifier drives a heavy load, thermal feedback from the output transistors to the input transistors limits open loop gain and can cause low frequency distortions.  Adding the extra operational amplifier to help drive the load reduces this effect.

As Kleinstein points out, doing it this way rather than by adding a current booster within the feedback loop has some advantages for stability, but I still would have added a current booster within the feedback loop which only requires a couple of bipolar transistors.

[Kleinstein]

Using a simple driver stage with discrete transistors may introduce addition cross over error. Besides the thermal effect an extra output driver could (with additional bias to operate the OP as class A) also reduce / avoid the output cross over error from the OP-amp.

[magic]

So the unloaded op amp would sort of "fine tune" the waveform while the bootstrapping op amp drives most of it?

That is sort of why this is done.  When an integrated operational amplifier drives a heavy load, thermal feedback from the output transistors to the input transistors limits open loop gain and can cause low frequency distortions.  Adding the extra operational amplifier to help drive the load reduces this effect.

It doesn't, if it's a monolithic dual such as here.

Thermal modulation is rarely an issue in audio; the real problem is class B output stage distortion. In 1980s it was only attenuated by the overall open loop gain; more modern RRO opamps seem to use "output inclusive" compensation, but whatever benefit it brings is apparently spent on further reduction of output stage bias.

[ELS122]

by adding a current booster within the feedback loop

Could you link some article on this? can't find any info on that.

[David Hess]

So the unloaded op amp would sort of "fine tune" the waveform while the bootstrapping op amp drives most of it?

That is sort of why this is done.  When an integrated operational amplifier drives a heavy load, thermal feedback from the output transistors to the input transistors limits open loop gain and can cause low frequency distortions.  Adding the extra operational amplifier to help drive the load reduces this effect.

It doesn't, if it's a monolithic dual such as here.

Then maybe they wanted the extra current, or power dissipation.

Thermal modulation is rarely an issue in audio; the real problem is class B output stage distortion. In 1980s it was only attenuated by the overall open loop gain; more modern RRO opamps seem to use "output inclusive" compensation, but whatever benefit it brings is apparently spent on further reduction of output stage bias.

I do not anticipate that the effect would be great enough to bother with, however unloading the output of a high performance operational amplifier *does* lower distortion.  This is a common change to make in Wien bridge oscillators to get the last bit of distortion performance out of them.

Even in the 1980s they had class-AB output stages in operational amplifiers, which are easy to tune when the transistors are all matched.  Fairchild even had class-AB versions of the 324/358 but they never became popular.

[magic]

I do not anticipate that the effect would be great enough to bother with, however unloading the output of a high performance operational amplifier *does* lower distortion.

I don't disagree, but your previous post seemed to suggest that thermal modulation is the cause of loading induced distortion, while in reality it appears to be mainly a matter of output stage nonlinearity and lack of loop gain to straighten it, mainly at high frequencies.

[David Hess]

I do not anticipate that the effect would be great enough to bother with, however unloading the output of a high performance operational amplifier *does* lower distortion.

I don't disagree, but your previous post seemed to suggest that thermal modulation is the cause of loading induced distortion, while in reality it appears to be mainly a matter of output stage nonlinearity and lack of loop gain to straighten it, mainly at high frequencies.

I know that thermal feedback causes low frequency distortion; I have seen and measured it because it spoils settling time.  I just doubt that it matters in audio applications.

[David Hess]

by adding a current booster within the feedback loop

Could you link some article on this? can't find any info on that.

Linear Technology application note 18 - Power Gain Stages for Monolithic Amplifiers is a good place to start.