Implementing current mirrors with MOS or BJTs in a BiCMOS ?

[aryajur]

Re: Current Mirrors

It is reported that in subthreshold Vth mismatch rises exponentially. I don't know what the reason is. Maybe the inversion charge is so less and at the surface due to surface states there may be variation in the fermi level position, causing the channel charge to vary which might be quite a lot percentage of the already less channel charge, thus causing a lot of mismatch.

 

[lete]

Current Mirrors

another reason is offset
BIP: gm=deltaI/deltalV=qI/kT-->deltalV=(kT/q)*(deltaI/I)
MOS: gm/I = deltaI/deltaV = 2/Vov --> deltaV = (Vov/2)*(deltaI/I)
Vov/2 >> kT/q --> MOS has big offset

 

[jinnose]

Re: Current Mirrors

another reason is offset
BIP: gm=deltaI/deltalV=qI/kT-->deltalV=(kT/q)*(deltaI/I)
MOS: gm/I = deltaI/deltaV = 2/Vov --> deltaV = (Vov/2)*(deltaI/I)
Vov/2 >> kT/q --> MOS has big offset

Well, I am not sure this reasoning makes sense. The above equation just tells BJT gm is naturally much larger than CMOS gm. (gm=deltaI/deltaV)

there must be a physical explanation.