Oliver Offset Optimiser

[Zero D]

I noticed the Oliver Offset discussions in here http://www.diyaudio.com/forums/solid-state/171159-bob-cordells-power-amplifier-book.html & here http://www.diyaudio.com/forums/solid-state/274064-oliver-bias-set-point-class-ab-power-amps.html

I must confess i hadn't heard of if it before ! It got me thinking as to how a particular mV value could be controlled etc. So i present the following circuit for evaluation etc.

For convenience, the input stage is shown as an OpAmp, but this could be any suitable input stage, non inverting or inverting. I've then only shown the OPS with the Oliver Offset Optimiser circuit connected. The output devices could be any suitable. The missing stage/s could be any suitable circuit.

My idea is to monitor the voltages across the Re's, then filter out the audio, & feed the combined voltages to the - input. The mixing resistors would need to be scaled to take account of the required offset.

 

[Zero D]

Edit

Just noticed the Re's are not what i planned. Please see my updated screenie

 

[AndrewT]

does the measuring circuit work when the output current exceeds the bias current?
i.e. when one Vre becomes near zero, or maybe even becomes a reverse voltage?

 

[Zero D]

Originally Posted by AndrewT

does the measuring circuit work when the output current exceeds the bias current?
i.e. when one Vre becomes near zero, or maybe even becomes a reverse voltage?

I don't know, as i havn't built it. I did sim a very simple amp though with my circuit connected, & it definately was doing something positive. "Possibly" only reducing DC offset, i'm not sure. By varying the values of the mixing resistors i could bring the original output offset down to around 5mV from 360mV.

I was hoping others might examine/sim it, & futher critique it, & explore any possibilities it may have.

 

[sreten]

My idea is to monitor the voltages across the Re's, then filter out the audio,
& feed the combined voltages to the - input. The mixing resistors would
need to be scaled to take account of the required offset.

Hi,

I fail to see how this does anything useful for proper bias settings.

rgds, sreten.

 

[Ian Finch]

You may also like to check out Edmond Stuart's extensive work on dynamic bias control, which sheds some light on the principles and problems too. Auto Bias part II

 

[Zero D]

@ sreten

I was trying to come up with a circuit that auto compensated for any changes in the desired mV target of the Re's.

@ Ian Finch

Thanx i'll take a look

 

[sreten]

@ sreten
I was trying to come up with a circuit that auto compensated
for any changes in the desired mV target of the Re's.

Hi,

Bias is a local loop within the amplifier and can't
be affected by the overall loop conditions.

rgds, sreten.

 

[franzm]

Designing a low-distortion audio output stage - Part 1: Introduction, the problem with push-pull
Linear Audio | your tech audio resource

 

[sgrossklass]

Interesting circuit. As shown here I'd say it does much the same as a DC servo, i.e. put a zero in the transfer function at 0 Hz, only that the reference is the current difference between Res rather than output DC offset. If you think about it, the only condition under which the currents in both Res are perfectly equal at DC is no current at all flowing out of (or into) the output - and that can only occur with either the output floating or at exactly ground potential.

 

[jan.didden]

I noticed the Oliver Offset discussions in here http://www.diyaudio.com/forums/solid-state/171159-bob-cordells-power-amplifier-book.html & here http://www.diyaudio.com/forums/solid-state/274064-oliver-bias-set-point-class-ab-power-amps.html

I must confess i hadn't heard of if it before ! It got me thinking as to how a particular mV value could be controlled etc. So i present the following circuit for evaluation etc.

For convenience, the input stage is shown as an OpAmp, but this could be any suitable input stage, non inverting or inverting. I've then only shown the OPS with the Oliver Offset Optimiser circuit connected. The output devices could be any suitable. The missing stage/s could be any suitable circuit.

My idea is to monitor the voltages across the Re's, then filter out the audio, & feed the combined voltages to the - input. The mixing resistors would need to be scaled to take account of the required offset.

The problem here is that seen at an Re, the audio is unipolar, so you can't 'filter it out'. This is the main problem ALL similar circuits have to face, and have been facing, the last 50+ years ;-)

Jan

 

[Zero D]

@ sreten

Maybe instead of the mixing resistors going to the FB, it could be "somehow" introduced into the driver transistors. Possibly each Re monitor feeding each transistor seperately ?

@ franzm

Thanx i'll take a look 🙂

Originally Posted by sgrossklass

Interesting circuit

Thanx 🙂 & for the info. Well it's just another of my what if & maybe ideas. I thought i'ld post it instead of keeping quiet about it, as you never know what might come from it/them

@ jan.didden

Thanx for chiming in anyway. Better to post & have it critiqued etc i think.

 

[jan.didden]

Of course. BTW it would work in a class A amp...

There's also an interesting solution by Edmond Stuart using opto-isoators, 'The autobias amplifier'.

Jan

 

[Zero D]

Originally Posted jan.didden

BTW it would work in a class A amp.

..

Really ! Why class A & not others, & how, connected to the FB or driver transistors ?

There's also an interesting solution by Edmond Stuart using opto-isoators, 'The autobias amplifier'

Yes, Thanx someone kindly posted the link earlier 😉

 

[Ketje]

Offset upsetter ??
Mona

 

[sreten]

@ sreten

Maybe instead of the mixing resistors going to the FB, it could be "somehow" introduced
into the driver transistors. Possibly each Re monitor feeding each transistor seperately ?

Hi,

As it stands its just an overcomplicated DC servo effectively.

The voltage sum across the Re's is only constant
in Class A, independent of the signal to the load.

Class AB requires heavy filtering to see the DC conditions,
which are addressed very adequately by the standard bias
practises, without the extra complications of filtering, the
op-amp supplies and all the extra gubbins.

This is a solution looking for a problem it doesn't understand.

rgds, sreten.

 

[jan.didden]

..

Really ! Why class A & not others, & how, connected to the FB or driver transistors ?

In class A the voltage across Re is a symmetrical ac signal. In class AB it is single-ended, so filtering gives a varying DC which does not represent Ibias.

Jan

 

[Zero D]

@ Ketje

I did also suggest that "Maybe instead of the mixing resistors going to the FB, it could be "somehow" introduced into the driver transistors. Possibly each Re monitor feeding each transistor seperately ?"

If the seperate outputs went to the opposite polarity driver transistors, with the appropiate scaled R14/15 resistors, that should cancel any deviations ?

I don't see how you got 0V out of U2 ?

@ sreten & jan.didden

OK, Thanx. Nice to know it would work, with Class A anyway 🙂

 

[jan.didden]

You can make it work in AB but your circuit which is basically a passive integrator should be changed into a 'minimum detector'.
The voltage across an Re in AB is half-sines (assuming sine signal) on top of a DC level from Ibias. Detect the minimum of that waveform and you can use that to control Ibias via feedback.

Jan

 

[Ketje]

If the seperate outputs went to the opposite polarity driver transistors, with the appropiate scaled R14/15 resistors, that should cancel any deviations ?

I don't see how you got 0V out of U2 ?

The upper OPA gives an output referenced to the output of the amp.
But the lower one takes the emitter of the negative side transistor as zeropoint, gives a shifted result.
As for U2 , like this 😉
Mona