Re: Constant Gm biasing
it's the equivalent of using a delta-vbe (ptat current) for biasing an op amp. basically your bias changes with temp inversely proportional to the diff pair, so you can keep the transconductance (and therfore GBW) constant over temp. now your room temp compensation *should* work over all temps. it's a good idea for any diff pair actually since it's pretty straightforward to build one.
the good bias ckt wil use a 1:N mosfet pair over a resistor R to generate the bias, similar to what is shown in the paper below. a crappy bias ckt will use a temp-independent voltage (vbg) minus a vgs over a resistor, but don't tell anyone i told you that.