Compound fully differential amplfiers

Hi, all:-

The fully differential amplifiers used to drive high res 24+ bit ADCs such as the OPA1632 have excellent AC characteristics (noise & distortion, for e xample) but horrible DC drift and resulting drift and low frequency noise- worse than a 1960s 741 amplifier.

Is there a straightforward way of combining a zero drift or ultra precision amplifier (or in-amp) with a fully differential amplifier to reduce the 'D C' issues?

I need really really good performance from DC (< 1mHz) to low audio (a few kHz and nothing weird between a few kHz and 1MHz)

--sp

Reply to
speff
Loading thread data ...

h as the OPA1632 have excellent AC characteristics (noise & distortion, for example) but horrible DC drift and resulting drift and low frequency noise

- worse than a 1960s 741 amplifier.

on amplifier (or in-amp) with a fully differential amplifier to reduce the 'DC' issues?

w kHz and nothing weird between a few kHz and 1MHz)

Can you break up the feedback path so one does HF and the other DC, and then sum the outputs? (That sounds fussy... and I have no idea how to do it with a differential amp.)

George H.

Reply to
George Herold

What sort of accuracy and frequency response flatness do you need? It's fairly easy to have a DC feed-beside path that corrects the fast gain path, but the gains of the two paths need to be matched to get precise gain flatness. I designed one of those last month, with 0.1% resistors in both paths.

It's much easier to correct the offsets and LF noise of an inverting opamp: just force the input offset of the fast amp to zero. You can do that with the fully differential amps *if* the equivalent opamp nodes are accessible, which they are on the OPA1632.

If I had more coffee, I could sketch things up.

Cascading those ADC driver type things sounds interesting. The combined GBW could be astronomical.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

I don't know of any method to handle down to DC, except, perhaps, chopping or periodic pause and execute an auto-zero. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
 Click to see the full signature
Reply to
Jim Thompson

Just measure the amp's DC/LF error, integrate, and use the output of the integrator to tweak the main amp's offset.

If you have access to the equivalent summing nodes of the fast amp, it has already computed the error for you [1]. If not, you have to look at the overall amp input and output and compute the error taking into account the forward gain; again, integrate the error and correct.

[1] except for differential bias current error in the feedback resistors.
--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

On a sunny day (Thu, 18 May 2017 09:10:02 -0700) it happened John Larkin wrote in :

Those high res 24 bit ADCs how stable is the reference at low frequencies? For audio that may not be a problem, but...

Reply to
Jan Panteltje

This has been done a lot - Tektronix in 'scopes, and there have been numerous app notes on how to build composite amplifiers with separate LF and HF paths. So long as gains match and nonlinearities are not too bad this can work well.

In particular I'm remembering some app notes from Maxim that show how they take a switched-capacitor filter (with crappy offsets) combined with a normal analog pathway to create a sharp rolloff filter with good DC characteristics.

Sorry I haven't pinpointed specific reference[s].

Reply to
Frank Miles

The "feed-beside" idea is in one of the Tek Concept Series books, "Vertical Amplifiers" I think.

The Tek thing was pure feed-forward, which was plenty good enough for an analog scope. And it had trimpots. For more precision, a computed-error+integrator thing is better.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

IIRC the trimpots were to better match gains. Without them, some serious ugliness happens in the frequencies where the paths are joined. This has not an error in the usual sense. That error calculation is between a reference and a feedback signal; this is between two sources, one which has one kind of error vs. the other with a different kind of error.

There are definitely more precise ways of doing it these days, I mentioned it only because this was an early use of the approach and seemed similar to what the OP was trying to achieve {I'm confident there were uses before that too, not meant to be exhaustive}. Tek itself used a variety of techniques developed after the Concept book series was published. I designed a split path circuit into a trigger circuit, largely to simplify switching complexity in the 7D20 (Tek got a patent out of that) which used yet another combining technique (no trimpots).

Reply to
Frank Miles

such as the OPA1632 have excellent AC characteristics

drift and low frequency noise- worse than a 1960s 741

ision amplifier (or in-amp) with a fully differential

few kHz and nothing weird between a few kHz and 1MHz)

?

Hi, Jan:-

The reference contributes a fraction of a ppm **of reading** per degree C, so the effect on the average of a noisy signal is perhaps an order of magni tude less. The offset drift is a ppm of **full scale** per degree C so it a ffects small average signals disproportionately.

Of course I could improve both with ovenization but that's no fun and air c urrents will still tend to add to the low frequency noise.

--sp

Reply to
speff

Is this a pure analog product, or one that feeds the signal into a computer of some sort.

if a computer, use two ADCs, one of which is one designed for weighing machines, then sum them with the appropriate digital filter in software

-- Kevin Aylward

formatting link
- SuperSpice
formatting link

Reply to
Kevin Aylward

h as the OPA1632 have excellent AC characteristics (noise & distortion, for example) but horrible DC drift and resulting drift and low frequency noise

- worse than a 1960s 741 amplifier.

on amplifier (or in-amp) with a fully differential amplifier to reduce the 'DC' issues?

w kHz and nothing weird between a few kHz and 1MHz)

At that low frequency what is the problem with requiring "excellent" AC cha racteristics and low distortion, anything can be made with that performance at low frequency. Forget looking for a composite and look at the high perf ormance audio OAs.

Reply to
bloggs.fredbloggs.fred

I would like drift of a maybe a couple hundred nV per degree C.

It's possible that just very, very good op-amps will do the trick. The harmonics are likely to be up above my anti-aliasing filter cutoff.

--sp

Reply to
speff

characteristics and low distortion, anything can be made with that perform ance at low frequency. Forget looking for a composite and look at the high performance audio OAs.

rmonics are likely to be up above my anti-aliasing filter cutoff.

It depends upon your signal. Third harmonics of a composite signal tend to end up in band for even narrow band, and second order also sounds problemat ic for your large bandwidth those tend to be an order of magnitude or two below third order BUT they are a major source of DC shift in analog circuit s ( square law term of the transfer function). I haven't been following tec hnology development lately but I'm pretty sure nothing exists to support 22

-bit linearity- yet, anyway.

Reply to
bloggs.fredbloggs.fred

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.