File talk:Gain-assisted current mirror.PNG

Something is wrong here

It looks like this design is confused between two approaches.

One correction accepts the error due to the base current of Q2: I_OUT = I_ref • ( R_E2 / R_E1 ) / (1 + 1 / B_Q2). In this circuit, Q1 does nothing useful. The input voltage is not constant: V_in = V_BE(Q1) + I_ref • R_E, which isn't so bad. But if Q1 is removed, V_in = I_ref • R_E, which is nicer. So just delete Q1.

The other correction roughly corrects for the base current of Q₂. Connect the + input of the op-amp to the base of Q₁ and the − input to the base of Q₂. Now I_OUT = ( I_ref • (1 + 1 / B_Q1) • R_E1 + V_BE(Q1) − V_BE(Q2) ) / R_E2 / (1 + 1 / B_Q2). Now the error only depends only on the ratio of (1 + beta) and the difference of V_BE.

(There are probably designs that have no error except for the op-amp offset and component tolerances. They probably need one more op-amp to implement a precise voltage-controlled current sink.)

-A876 (talk) 04:39, 7 May 2024 (UTC)