Why doesn't chopper stabilization remove 1/f noise completely?
Why doesn't chopper stabilization remove 1/f noise completely?
I'll preface this by saying I'm not an IC designer, and someone may have a more complete answer to this, but here's my understanding. At the very least, you have an output stage which is outside the chopper stabilized loop, so the 1/f noise of that stage is not modulated by the chopping. There is finite gain from the input to that output stage, so the 1/f noise generated is reduced by that gain level. It is not usually a very big effect, the 1/f noise corner in chopper amplifiers has often been demonstrated at 1milliHertz or less. Here is a block diagram of the ADA4051 chopping scheme that I copied from a zero-drift article to show what I mean:
You can see that the output Gm, Gm2, is outside the loop. There was a 2009 paper in IEEE by Wu, Makinwa, and Huijsing about a chopper in-amp that showed this well, which you might be interested to read.
