One thing people might ask, or get confused with, is why not just using R1 & R2 as a resistor divider. Why all the trouble with the current mirror? And the reason (I think) is the sensitivity to Vin that would imply. With a R1 & R2 resistor divider we would be not only measuring changes in nVo, but also in Vin all mixed up. With the current mirror, as Vin changes, Vgs1 remains the same, and so, up to some (reasonable) frequency the gate voltage of the PMOS follows Vin. Therefore the voltage drop across R1 is in 1st order not directly dependent on Vin and so is Ib2.
PS: most likely gm2*R1 >> 1 such that a change ΔV in nVo becomes approximately (ΔV / R1) * R2 change at the error amplifier. Or rearranging it ΔVε = R1/R2 * ΔV which is linear.
@@sambenyaakov I watched video again. i seen a different approach that does not needs super-precise resistors. It was called "over-the-top amplifier" and it uses some schematic magic. i was wondering if i could make one with classical components and external transistors