Hello, all

Please check the schematic in attachment. It’s one way to generate PTAT current. For example, Q2’s area is N times of that of Q1. The two Vbe difference will dump onto R to produce the PTAT current. Obviously, the quality of the PATA current has a lot to do with how well i1 can match i2.

It looks like that the negative feedback loop (Q1, Q0, M0, M2) can help to force vx equal to vy (i.e., i1 = i2). But I just wonder how the feedback loop works on this kind of structure. This feedback loop is quite different than the conventional structure (e.g., two explicit input terminals and one output terminals of an op-amp. output signal somehow sensed and returned to negative input terminal to be compared with positive input). In my feeling:

1.      It’s very hard to identify where the negative/positive inputs and outputs of the op-amp are in the ckt of this kind.

2.      Even if the current of Q1 and current of M1 can be regarded as positive and negative inputs to an op-amp, the vx=vy (or i1 = i2) can’t be guaranteed.

Please comment on this topic and feel free to correct me. Thanks a lot!

Hello,
this has -ve and +ve feedback, they will make sure that Vx and Vy are set at rite common mode, vx=vy is wrong. Since this circuit is single ended it wont make two potential equal rather it makes two potential difference constant. coming to the fig,  -ve fb get you vy=vdd-vbe(Q1) and +ve fb get you vx=vdd-vbe(Q6) .

Thanks,
Raj.