OpAmp DC Gain/DC Gain Linearity Tester

[Samuel Groner]

Hi

I need a test jig to measure opamps for their DC open-loop gain and DC open-loop gain linearity. This is usually done by amplifying the error voltage at the inverting input by means of a voltage divider as shown e.g. in OP177.pdf (page 9) or AN-1485.pdf (20 MB).

What I need:

* measure gains from 60 dB to 180 dB (in four decadic ranges)
* trim opamp offsets up to +/-20 mV (in three decadic ranges)
* output loads of 10 kOhm, 2 kOhm 200 kOhm as well as external load
* supply voltages up to +/-40 V
* optional low-pass filter to reduce noise

And here's what I've come up with: DC_gain_linearity_tester_r1.pdf

The +/-15 V supplies to bias the offset voltage trim circuitry are only used when testing at supply voltages below +/-12 V.

Any comments, additions, corrections? Thanks.

Samuel

 

[Svart]

I had ideas to offer but then I opened your PDF and saw that you pretty much nailed everything I would have suggested with the core of the circuit.

I would suggest some ferrites/inductors on the rails and possibly a lot of bulk decoupling, you are building a test jig so don't skimp on it.

Layout will be somewhat crucial here as in any test circuit, you don't want to measure inconsistencies/parasitics in your layout.

 

[Samuel Groner]

Thanks for the input. I'll give the decoupling another thought, although I'd say it ain't that critical here as long as stable operation is achieved. The frequencies of interest are all way down (< 100 Hz) where no reasonable cap has sufficiently low impedance, so the silicon in the PSU needs to take care of this. Layout will indeed need to ensure low noise pickup, and perhaps even a magnetic/static shield might be appropriate to keep the Y signal clean.

Samuel

 

[Samuel Groner]

I got around populating and testing the board today. Works fine, and resolving 140 dB o/l gain is a piece of cake. Troublesome are just opamps with discrete input transistors; they tend to have high short-term drifts which makes the oscilloscope trace shift into nirvana if the offset trim is not continuously adjusted. I'm thinking about using a servo instead.

Samuel

 

[Svart]

Servos drift slowly as well don't they? 

 

[JohnRoberts]

A servo based on a decent IC opamp should keep him adequately centered.. The servo combined with an initial manual trim could be a little easier.

I never tried to servo 140dB but I servo'd a phono preamp with 60+ dB gain at LF with no problemo...

JR

 

[Samuel Groner]

Servos drift slowly as well don't they?

But the drift of the servo amplifier is related to the output of the DUT and not to the input. If the DUT drifts 1 mV in the 1000x setting the X output wanders 1 V; if the servo drifts 1 mV the X output shifts just that one mV which you won't even see on the scope. In addition to this of course the servo amp may be chosen to have low drift, while the DUT is just what it is.

Samuel