talking about using transistors in place of capacitors and resistors, because those take a lot of space
still have to bias circuit, even without real resistors
NPN general purpose transistor, gives value maximums, thermal characteristics, 
all curves on a mosfet Vds/Id graph come off a point, come off a line for BJTs
BJT is Vce/Ic, if in sat/cutoff, working as switch
max power curve varies on operating temperature
on a mosfet, you have triode, cutoff for switch, use saturation region for amplification
triode vacuum tubes have curves a lot like mosfets
still have to set a qpoint and obey the usable space decided by power curve, cutoff, triode, Vds max, and Id max
gate length is almost too small
mu is electron mobility
Vdd is DC voltage bus
early voltage still applies for mosfet, but the physical mechanism is totally different
on page 260, table 4.1
overdrive voltage: Vgs - Vthresh

in saturation, Vgs > Vthresh, Vds >=Vgs-Vthresh
hFE: DC current gain equal to beta
can get slopes from graphs from curve tracer to find early voltage

basic MOSFET circuit used for training, no resistor used in real life
mosfet mirror, Iout should mirror Iref, constant current source
gates tied between the drain and resistor
equations for mosfet current mirror
Q1 side
Iref = Id1 = 1/2 * un*Cox*W/L*(Vgs-Vthresh)^2
Vgs + Iref*R-Vdd = 0
Iref = (Vdd-Vgs)/R

Q2 side: must still be in saturation
Iout = 1/2*un*Cox*W/L*(Vgs-Vthresh)^2
Iout/Iref = 1 if transistors are identical, perfect match, else, (W1/L1)/(W2/L2)
easy to control with mosfets
totally geometry dependent

remove R and substitute with a current source
conditions for mosfet saturation: Vgs>=Vthresh, Vds>=Vgs-Vthresh == overdrive voltage, Vov
Vds2 = Vo -> Vo>Vov
Vov is usually a few tenths of a volt

if W/L values equal, Id1 = Id2 = Iout = Iref
all this happens at a particular Vds and deltaIout
Ro2 = deltaVo/deltaIo = Va/Io