In the attached schematic, there are two branches. The branch on the left has a pmos + nmos transistor. The branch on the right has two nmos transistors. The sizes of the devices were selected such that the current through each branch is almost identical.
Each branch sets the reference current for a current mirror, where a MOSFET is used instead of a resistor to set the bias current (i.e. M3 and M4 can be thought of as bias resistors).
My question is this - if you can use two nmos transistors to achieve the same result as a nmos+pmos configuration, why choose the option that uses the pmos? The textbook I'm reading through uses the example of pmos+nmos, without really justifying why. It's a non intuitive answer to me. I suppose more generally, if it's possible to arrange a circuit configuration using only nmos transistors, why/would you ever choose to design in a way using pmos?
