why common centroid for diff pair and interdigitized for current mirror?

[ms_90]

I have read in most articles that for current mirrors Interdigitisation is better and for diff pair common centroid is better. so why is it so?(one reason for avoiding common centroid is routing complexity) cant we use interdigitization for both case? or vice versa?

i have gone through other similar articles on forum but still unable to understand the concept or reason behind it?

 

[D.A.(Tony)Stewart]

I believe "common centroid" makes them immune from cross-chip gradients. Best matching performance is possible.
Common source area can reduce size.

 

[schmitt trigger]

Absolutely correct SunnySky

The late Bob Pease had an excellent explanation on the technique.

 

[D.A.(Tony)Stewart]

ah yes R.I.P. thanks Schmitt T> is your location encoded near Area 51?

 

[schmitt trigger]

"is your location encoded near Area 51?"
Of course not. I'm in South Texas.;-)

If I remember correctly how I coded my location, it is the Hex representation for the ASCII code of the name of my town.

 

[ms_90]

I believe "common centroid" makes them immune from cross-chip gradients. Best matching performance is possible.
Common source area can reduce size.

thanks sunnyskyguy,
but can you please explain the detailed concept of cross chip gradients? it would be better if you will give any example..