Hi Jason,

If you wanted to calculate the input referred offset of an opamp, ideally you sould consider the max mismatch you could get at any gain stage and differential load that comprise your opamp. Then you should "reflect" such mismatch at each stage back to the input by dividing it by the corresponding gain.

For esxample consider just one MOS diff pair stage with a resistive load. Mismatch on your resistors will show up as a delta V on the drains of your input MOS transistors. Such delta V will generate a delta I according to the output impedance of such transistors and at the same time such delta I will be reflected to the inputs as a certain delta VGS (depending on the gm value of the diff pair).
Such delta VGS will add up to the delta VGS that already comes from the MOS mismatching.

You should do the same with each successive stage in your opamp, but it ends up being a tedious job.

If you have access to Cadence, AMS has available one analysis called "dcmatch". If your models include the corresponding mismatch parameters, by running this analysis you will readily get your input (or output) referred offset in your opamp. This analysis simply calculates any single offset contributor the way I described in an obviously much faster way than you!

Hope this helps. Regards
Tosei

Hi manodipan,

It should give you good results assuming:
1) the correct matching parameters for resistors and transistors are included in the model
2) the circuit is properly biased around the operating point.

It is not strictly necessary to have a 0 differential signal for running this analysis as long as the circuit operating conditions are not changed. Also please check the common mode and the general operating conditions are properly set.

I'm not sure what you mean with the threshold parameter you are referring to. May be the threshold mismatch of your MOS transistors?
If you could post the list you got from the analysis that would help.

Regards
Tosei