Chopper opamp specs look fabulous... microvolt offsets, nanovolts per degree C, pA bias currents, no 1/f noise. But we have found some strange behaviors.
Some seem to have some sort of front-end charge injection that makes offset a function of the impedances seen by the input pins. This acts like an input bias current but actually isn't.
One one board we're working on, we couldn't get consistant calibrations. Turns out that one chopper amp is showing offsets that drift around 10s of microvolts. It's on a floating analog channel, and there's a bit of dc/dc inverter feedthrough on the channel common, and we're in a high EMI environment, so one of those is probably the problem. Finger-prods in the neighborhood of the amp will change the offset.
We replaced the offending AD8628 with an OPA376 and the problems disappeared. The datasheet doesn't suggest what technology the '376 uses, but it's a stunning part.
John