EEVblog #1325 - OPAMP Shootout

[graybeard]

The OPs in dual OPs don't necessary match in offset of drift. They share the same process, but the OP designs itself are made in a way that with a perfectly symmetric process there would be no offset. So the offset of the individual OPs  is already from things like gradients or local defects and also thermal effects at the time when the offset was trimmed at the factory.

The effects of thermal EMF is surprisingly small, at least with things like CMOS switch chips.

Still thermals can be important. So for high precision it helps if the power is low. So the OPA189 may have more trouble at higher supply (and thus more heat).

The transistors in integrated circuits do not match perfectly.   They match much better than discrete devices ever have a hope of doing, but they always differ a bit.  The differences are typically larger as the devices are spaced further apart.  The amount variation will vary wafer to wafer and lot to lot.  Even devices that are side by side still have small differences.

 However you are correct that sometimes the biggest differences are due to thermal gradients in the operating circuit, not the usually small variations of close by device properties.

[David Hess]

Precision matched transistors on a monolithic process use an interleaved layout to cancel out the effects of process variation.  For duals and quads which are advertised to have matched parameters, I think this means that a symmetrical and mirrored layout is used with all of the matched transistors interleaved with each other.

Matched transistors which use interleaving are not suitable for high frequency applications because of capacitive coupling between the transistors, and I suspect this affects short term settling time as well although long term setting time will be improved by better thermal balancing.