FETs on the ASM-1 VCO

[Arthur Harrison]

AH:

> > I was a little concerned,
> > though, about possible instabilites in U1A resulting from C1's
>direct
> > connection
> > to its output.

JH:

>No, it's perfect the way Joachim has done it. There is a
>certain *range* of capacitance values that will make an
>opamp instable. You can either stay below that range
>(very small cap), or go above that range (very large
>cap). Think of it as a 2-pole filter, one pole being the
>internal compensation cap, and the second one being
>formed by the output resistance and the capacitive load.
>It's only when the two poles are relativly close together
>that you get high Q factors, and thus overshot or
>oscillation.
>Normally you don't have much choice, of course: You will
>use low capacitive load when the opamp output voltage
>has to change fast, and high capacitance when you want
>to bypass a constant voltage.
>It's the same principle with voltage regulators: They even
>*need* a certain capacitive load to be stable !

AH:

Yes, I've seen certain vivid examples of the stability regions for
capacitive  loading; the TL431 shunt regulator particularly comes
to mind.

As far as linear series regulators, my experience has been that the
78xx parts are more stable with minimal or no capacitance, although
a capacitor at their outputs will, of course, improve transient rejection.
The 79xx regulators, however, certainly do require a minimum
capacitance for stability, as do many low-dropout types.

Although I agree with your explanation regarding the filter poles,
oscillations
can also occur whenever a bipolar transistor's emitter is capacitively
loaded.
This type of oscillation may occur regardless of loop phase-gain criteria.

AH:

> > Can C1 be moved across R6, instead, or does the transient
> > response of Q1's source node degrade too much?


JH:

>This would give you a good reference voltage at the opamp
>input. But at the output, it would only be good at low frequencies.
>For higher frequencies, the internal gain of the opamp goes
>down, and thus the output resistance of the closed loop goes up.
>With the cap at the output, you have the precision buffer of the
>opamp at low frequencies, plus the low impedance of the 10u
>at high frequencies.


AH:

Agreed.