MOSFET Follies Clairification

[R.G.]

Constant current sources bleed in a constant current regardless of the voltage across them. So in this application, it feeds a constant current the source which is following the signal voltage, **regardless of ripple and trash on the negative supply**. The resistor will let the source current vary by I = V/R, and therefore mix in some of the trash and ripple which may exist on the negative supply.

In addition, a source- (or emitter-, or cathode-) follower has a gain nearer 1.000000 instead of nearly one as the equivalent resistance it sees as a load goes up. This is the "Rst" term in the equation I used. Constant current sources have equivalent resistances which are very, very high, so this pushes the follower to being a better, more ideal follower as well as reducing noise feedthrough.

Constant current sources were an expensive luxury back when you had to use an expensive tube to make them. But with $0.10 semiconductors, they are very, very nice additions.

[defaced]

Thank you, and as always I'm going to have to chew on this to really get it, but that's what I like about this stuff. What does the constant current do to the output impedance of the stage? Would it be characterized by simply taking R=V/I for the idle no-signal condition? This general circuit has some other applications (FX loop sends) that I'd like to play with.

[R.G.]

Thank you, and as always I'm going to have to chew on this to really get it, but that's what I like about this stuff. What does the constant current do to the output impedance of the stage? Would it be characterized by simply taking R=V/I for the idle no-signal condition? This general circuit has some other applications (FX loop sends) that I'd like to play with.

See Lectur 9: Common Drain Amplifier at Stanford's online EE courses.

Making the source resistor be a constant current source, with a difficult-to-calculate-but-very-high impedance decreases the output impedance some, although not much. The output impedance is already very close to 1/gm, and making the source resistance even higher doesn't help all that much in lowering it. It's already quite low, and usable for applications like loop sends into impedances down in the 10K range as is.

The principal value is the immunity to power supply noise.

This is not the general case for current source loading of the drain, where a good current source load can run gain way up.