How to minimize output offset drift w/o servo??

Leolabs said:

Many fully push-pull(from input to output) amplifiers do rely on servo circuits to minimize offset drift.
Is there another way to do so(ie the duo beta by Luxman)???

Click to expand...

not quite on topic, but almost

with good thermal contact between the involved transistors,
in such an Active bias current Compensation circuit
as I show in topic I started,
it would really be minimal drifting

but best of all,
by using separate active bias current compensation
there will be near to ZERO current flowing in input resistor
as well as in feedback resistors
----------

this makes
- you are free to select any Resistor Values, without any DC-offset issues
- you can do well without blocking capacitor from feedback divider to ground
- there will be almost zero Current Noise in those resistors at input, at least when no signal
- even if you use symmetrical differential input,
NPN and PNP pairs can have considerable difference in gain,
when you have matched the Vbe for so called ' complementary ' pairs
this technique can compensate for such bias current offset, that can be often more than is pleasant

================================================
Here is my topic:
Offset Correction using Bias Compensation

================================================

By the very low interest,
it appears to me, not many are clever enough to understand what great benefits there is.
I use it very often in my Spice Simulations.
And so I trim until there is less than 1uV offset at all 3 points in amplifier:
+input, -input and output


lineup not much compensation for sharing good ideas around here

Hi Leolabs,
Most older amplifiers run happily and stable by using a constant current source that is temperature compensated as well. This may be as simple as a regulated voltage source and resistor. A CCS operates over a much wider range.

Also, match and thermally couple the diff pair transistors. Dual diff pairs need all four matched.

Couple the feedback ground side through a capacitor so the DC gain is 1 (unity).

After doing this, your DC offset should not drift much. If you are using a single diff pair you can inject a correction current into one side to adjust the DC offset (look at 70's Marantz for hints). Marantz temperature compensated the offset correction circuit some times too.

-Chris

OT:
I am still living happy with enormous drift (>100mV).

Hi darkfenriz,
You're not OT, but you don't have to suffer needlessly.

-Chris

there is some talk about designing for low offset without servos in borbely's article on his dc102 amp: http://www.borbelyaudio.com/adobe/dc102.pdf

mlloyd1

Leolabs said:

1 more questionoes this circuit of your suitable for a input stage with
base as input and emitter as feedback???

Click to expand...

Yes, of course.


It can be used for provide input small bias compensation current
in any transistor
that is working using some current going into Base !!


If you mean transistors operating with Base connected to ground
then it wont be needed,
because there will be no input offset voltage
cause by Base current.

Hi Leolabs,
Do you mean complimentary diff pair?

All those transistors have to be matched together for the lowest distortion and DC drift.

-Chris

Hi Leolabs,
I'm sorry. I'm having trouble visualizing that. It's just me.

Do you have an example?

-Chris

Hi Leolabs,
Thanks, Tandberg style. Never liked it much because it isn't naturally balancing.

You must match the fets closely and couple them thermally. A capacitor in series with R4 would greatly help with offset control. Conversely, you can use an op amp integrator (filter it's output) and use that DC signal to control your offset. It does complicate this circuit but may offer much better control. I would match the fets and couple them in any case.

-Chris