Floating vs Diffamp

[ezalys]

Why is it that instruments with differential input stages use instrumentation amps, and instruments with “differential” output stages float their outputs?

Why not the reverse? Why do we not insist upon using an optoisolated input stage and a THAT1606 or similar differential driving output stage?

Note, I would like to ignore all safety-reasons or “large common mode voltage” situations. Let’s suppose we’re only in situations where the common mode voltage is small relative to the rails. None of this “industrial environment” stuff.

[ejeffrey]

An instrumentation amplifier has high differential and common mode impedance and is cheap and easy and will work basically the same as a floating output.

A fully differential opamp output normally has low common mode impedance.  To get high impedance like a floating driver you need to use feedback to set the common mode level for zero current.  This works but is hard to get high bandwidth and tends to have stability problems.

Overall I would reject your premise. Plenty of instruments produce differential outputs using just a fully differential opamp or line driver.  They just have low common mode impedance.  Also for higher frequency isolating baluns are common on inputs and outputs.

[ezalys]

Ok that makes sense. Suppose I want low diff impedance and high common mode impedance. Is there a more clever way to do it than floating the DAC and reference? Is there a good way to transfer a precise voltage across an isolation barrier? Flying cap? Transformer plus rectification plus insn amplifier feedback loop?

[Wolfram]

It's difficult to recommend specific solutions without knowing the required bandwidth, linearity and dynamic range.