OPA2277: OPA2277, Details on Input Bias Current Cancellation

Link below has an overview of the input Bias current cancellation circuit.  There is no IC level information I can share.  If you post a detailed schematic and scope photos of what you are seeing we can perform more analysis. Also to eliminate external noise effects consider the test technique detailed in the training below:

2.1 TI Precision Labs - Op Amps: Vos and Ib

8.7 TI Precision Labs - Op Amps: Noise 7

This problem appears to be unique to the OPA2277 and again, I suspect has something to do with how the bias current cancellation circuit works.  It's an interesting problem.   I noticed in the original post I referenced, there were a few recommendations for a replacement part, however both were choppers.    I am not too surprised to learn that TI wouldn't release information about the BurrBrown part and and started testing an Analog Devices AD8676 as possible replacement.  So far, it looks pretty good.

As far as a circuit, its easy enough to replicate but is part dependent.  While I suspect all OPA2277's behave this way, if you attempt to replicate it, you may find you need to test several devices of different lot codes.

Just run the part as a follower and supply the input with a clean stable voltage using a source of around 100K .  The voltage level is not important.  I am not concerned with the absolute output voltage (offset errors).  Now just monitor the output.  The problem does not appear to be temperature or time dependent.   When these shifts occur, they are in the order of sub 50uV.   You will find the part always returns to it's baseline value.   The time the part shifts for will vary a fair amount, from seconds to minutes.    The shift is not gradual.  It appears digital in nature.   The time required to see a shift will very as well.  This could take from a few minutes to several hours.   Lowering the value of the input resistance will minimize this effect.   I am running the parts from +/-15V.  I have not looked to see if there was a dependency on the supply voltage or not. 

Again, I am not asking TI to solve the problem but it would have been good to understand what was going on inside this part.   The first video, at 1:24 we hear it is a "digital" circuit which does appear to be the nature of what I am seeing. 

Hi,

Marek already explained it in the other thread as being popcorn noise. This noise appears to be "digital". Also, in the other thread the noise is said to be ultra low, only 500nV to 1µV.

You say that lowering the 100k source resistance will minimize the effect. Well, it's rather unusual to run the OPA2277 with high source resistances, because with high input resistances the input bias current and input offset current will produce an offset voltage which can be much higher than the input offset voltage of OPA2277. Also, a high source impedance would ruin the proper noise performance of OPA2277. So, with the usual low source resistances used in combination with the OPA2277, it's very unlikely that popcorn noise becomes a serious problem in a typical OPA2277 application.

Kai

Obviously I read the other thread or would not have linked it. I suspect the cancellation circuit and not burst noise is the shift can last for several minutes. There are other features to how it behaves that are not random as well. Beyond the one video mentioning the cancellation circuit in two places and pointing out that it is indeed digital and one paragraph in the datasheet, I really have little to go on. Maybe there is a patent for it.