AW: opamp compensation in integrator application (was:Re: Descret OTA )

[Haible Juergen]

	>Yes, you have to care about the compensation of the opamp for an
	>integrator.  Textbooks say the opamp must be compensated for unity
gain. 
	>This is contrary to what is in the ARP 4023, where the opamps are
	>de-compensated.  Normally in a case like this, stability can only
be
	>guaranteed if the closed loop gain is at least 9, which is
impossible for
	>the usual type of integrator. However, this one has a 100 PF
integration
	>capacitor, and the inverting node of the opamp has a 1000 PF cap to
ground.
	> This attenuates the feedback by a factor of 11, which is
sufficient for
	>stability.  Now it makes sense.  The decompensation speeds up the
LM301 by
	>9 times, and the two capacitors together ensure stability by
attenuating
	>the feedback around the opamp.  As you pointed out in an earlier
post, this
	>is the noise gain approach to stability.
	>
	>Terry Michaels

Hi Terry,

that's pretty much how I see it, too. Make the opamp as fast as possible,
and care
for stability by increasing the noise gain. 
But what if they would have compensated the 301 (i.e. made it slower) for
unity gain
stability, and then built a "normal" integrator around it (without the cap
to ground) ?
My "feeling" says it would be an inferior solution, but I haven't done the
calculations
to back it up. Would you say that the difference is still neglegible,
because of the
overall integrator function ?

JH.