Instrumentation Amplifier-Monolithic or 3 individual op amps

Any thoughts on the advantages/disadvantages of using a monolithic instrumentation amplifier like the INA103 versus three individual op amps and 0.1% resistors?

I'm a big fan of using a unity gain buffer in the feedback loop of high gain circuits in order to thermally isolate the driver stage from the gain stage. Seems the 3 op amp configuration at least has that advantage.

Thomas

Well, at first consideration, it seems two advantages of the monolithic amp would be that it'd be pre-trimmed, and being on the same die, the amplifiers would better track each other thermally.

This seems more like a "Drawing Board" discussion to me.

Yeah, I thought I was posting in the Drawing Board, but somehow it popped up here? Hopefully one of the moderators will move it. -----moved -prr

You think thermal tracking is a good thing? I definitely think the driver stage should be thermally isolated from the gain stage. But I can imagine that there might be some advantage in thermally coupling the two input amps. Maybe a dual package for the inputs and a separate one for the output?

Yeah, I was thinking of thermal tracking between the inputs. Your idea sounds like a good one.

> fan of using a unity gain buffer

You can stick a buffer inside the INA103 output loop, keep the heavy load out of the INA103. B-B even show one, though it is an awful design.

I know many recent chips have much better thermal balancing than the old ones that showed significant artifacts in the bass.

[quote author="PRR"]You can stick a buffer inside the INA103 output loop, keep the heavy load out of the INA103. B-B even show one, though it is an awful design.

I know many recent chips have much better thermal balancing than the old ones that showed significant artifacts in the bass.[/quote]

So, do you think you would prefer an INA103 over a well implemented separate design made with say, OPA134s or OPA604s and 0.1%, 25ppm resistors?

> you would prefer an INA103 over a well implemented separate design made with say, OPA134s or OPA604s and 0.1%, 25ppm resistors?

I would prefer a situation where I didn't need anything as heroic as 0.1% 25ppm parts.

I glanced at your website. If you are looking for a full-floating input for an active monitor, I would not use that topology at all. The flaw is that if the power goes off, the input looks like a diode to ground, which may badly upset whatever it is connected to. And if the source has enough grunt, it will turn the INA into a puddle.

For general use, the plain 2-inverter diff-input is hard to beat. Using 22K or 47K resistors it is pretty blowup-proof. Using a 4-R thinfilm SIP resistor pack, the CMRR will be over 40dB no-trim. And you are not relying on a single-source product like INA103; any 3/4-decent dual opamp will do.

If you need more than 40dB CMRR, you have huge common-mode voltages and probably should be looking at a transformer. Use a low-Z transformer, couple input resistors, and a zero-Z amp, like shown on page 2 of this paper. A similar concept, more over-blown, is in this thread, implemented as a mix-amp. This scheme gives KV of galvanic isolation, high CMMR, very low THD and very good bandwidth. You could probably make a pro-quality input with a $2 modem transformer: the opamp does the work.

on page 2 of this paper

Click to expand...

I would love to see the rest of this paper.

I would love to see the rest of this paper.

Click to expand...

This is from:
Microphone engineering handbook
Michael Gayford
Oxford, Focal Press, 1994, 449 p.

A not too short "paper"..!

You'll find some info on this topic in the data sheets of Lundahl's zero-field input transformers LL6404 and LL7101.

Samuel