measuring offset voltage using montecarlo

[gayu]

Hi,

I'm trying to do the mismatch analysis using montecarlo simulations for a simple diff pair..I selected "mismatch only" option and tried plotting the drain current for the diff pair transistors. I expected a change in the two currents so that I can divide the difference by corresponding gm to get the offset voltage..BUT I didn't see any difference in the current!!! Though I have mismatch block in my model mc file and correlated the two transistors. Is this due to the correlation coeff I'm using??? Please clarify how to go about this

 

[tshankar501]

did you make the difference between the transistor models so high? just as a trial......make sure the difference is so high just as a check at first and proceed......if the difference between the model values of the i/p transistors are so high and still if you dont see any change in the current values on i/p transistors.....then there is some other error that you are doing......

 

[amitpjag]

hi
what Mr. shankar is saying, its correct.
well the current difference will also depend on which parameter you are making mismatch and obviously on gain of your Differential amplifier.

 

[gayu]

I don't understand what do you mean by the difference in the models. Say I'm testing for a diff pair, I'm going to attach the same model lib to both the transistors. Where do you ask me to bring the difference. Please clarify.

Thanks,
Gayathri

 

[tshankar501]

I do not know what would happen if you select "mismatch only" option or if you attach "mismatch block in your model mc file"...............I do not know what would be taking place in the background of the tool that you are using......

If you are going to attach the same model lib file to both the i/p transistors, then there would be no difference in the current between the two transistors.....

The current in the transistor is determined by the model parameters that you are specifying in the model lib file for that particular transistor.......

If you are going to have the same model file for both the transistors, there I would expect the same current on both......You wouldnt see the change......

So try giving different model lib files for the two transistors and you would see different current......

The offset voltage is caused due to mismatch present in the two input transistors......In simulation, this mismatch could be introduced only by giving different model files.....

 

[gayu]

Then what is the advantage of going to Montecarlo mismatch analysis?? I thought ,depending on the correlation between the transistors the simulator would change the parameters statistically and so forth.. Is my belief arguable??

 

[tshankar501]

Sorry I get what you are telling now. It is supposed to be varied statiscally. Take a look at your mismatch block again and get clearly verified whether your parameters are actually varying.

Check carefully what all parameters are getting varied and by how much are they getting varied. I guess there should be an option to look at the values of each parameter every time it cycles thro. I hope this change in the parameter values for every cycle will be written into a text file.

When you tell that the transistor currents are not getting varied, I hope that you must have noticed the currents at two different times. Check all the parameters at those two different times. Check how much did they vary. Calculate manually and try seeing if there is any small difference.

Check thro the parameters values and currents for every cycle. Calculate manually for upto 10 cycles. Thats the best I know out it.

 

[avt]

could someone clarify me how to set up the correlation values - they can be between -0.9999 and 0.9999 - so what to choose - ad why negative values ?