[sdiy] Op Amp Buffer stability

[David G. Dixon]

> >I generally put an additional cap directly downstream of an on-board
> voltage
> >regulator (such as 78L05).  Are you saying that I should do the same with
> a
> >voltage reference (such as REF02)?
> 
> I say I don't know, but the data sheet of that specific reference will
> tell you!

The datasheet figures do not show any filter capacitors on the output in the
application examples.  Only a 0.1uF cap between +V and GND on the input.
What it does tell me, however, is that the typical REF02 generates
peak-to-peak noise voltages of about 4 to 10uV between 0.1 and 10Hz.  The
noise voltage increases with frequency, but I'm not sure how one figures the
frequency for this calculation when the thing is connected to a couple of
rails which, strictly speaking, should not be oscillating.  In any case,
this does not sound like it justifies a capacitor to me.  I suppose it
wouldn't necessarily hurt, but not having a degree in electrical
engineering, I'm not sure I understand how to interpret the datasheet in
this case well enough to judge.  All I know is that I hate cluttering up my
boards with unnecessary components.

I also think we generally over-decouple our synth circuits, using caps which
are often unnecessarily large, and which put too much strain on power
supplies at power-up.  I use 10uF caps on each rail on each module, but I
know people who use 22uF, and I'd bet that some people even use 100uF.  When
you stack up 20 modules on a PSU, each with a big ol' cap on each rail,
unpleasant things can happen at power-up (so I've been told -- I don't have
enough modules to see it myself -- yet).  I'm beginning to wonder whether we
shouldn't be decoupling the rails at the power distribution board within
each cabinet, and leaving some of these caps off of the individual module
boards...?