Hi, may i ask about current-mode bandgap reference.
The current-mode bandgap reference is based on the same theory as conventional bandgap reference. A PTAT (proportional to absolute temperature) voltage is added to a base-emitter voltage which exhibits a negative temperature coefficient, making the output independent from temperature in first-order. The core circuit is illustrated in the figure attached.


Question is: R2a and R2b are supposed to be identical, but in practical implementation, mismatch between R2a and R2b is 2.5% on purpose, why?

yes, I agree with ywguo that I wouldn't introduce mismatch in those two resistors on purpose by design.. unless you want to start playing with second or higher order temp-co compenstation.

The other possibility is that the whole structure may have systemetic imbalance (for eg. due to bad opamp design or just bad layout), the designer may start tweaking in that direction.. but I doubt it.

Cheers!  :)

The resistors steal current from the bipolars and that will affect the tempco.  I could understand that one would tweak these resistors to get the tempco they want.

However, mismatch will be a killer and would probably remove any benefit of the tweaking.

Are the resistors even necessary?  If not, remove them.

The tempco of these resistors also impacts the temperature curve, so if you can get rid of them you should.

Regards,
Dan

Oops!  I looked at this when I was tired.  I thought there was one additional resistor that determined the ptat current.

Yes, the tempco of the resistors will cancel but only if the layout is drawn well and the effects of mismatch are minimized (duh, right?).

Thanks for pointing out my mistake .

Best Regards,
Dan

Which one has larger resistance? R2a or R2b?