[sdiy] Taming an oscillating unity-gain voltage follower

[Tim Stinchcombe]

Hi list,
	I have been taking some measurements from the filter in a TB-303
clone, an Analogue Solutions TBX-303. As I'm taking the output from directly
after the filter, I wasn't that surprised to discover that apparently the
capacitive loading of all my test leads allows the filter (at high resonance
settings) to easily oscillate at about 10Hz, as it is the effects of the
lower of the filter's two resonant peaks that I am investigating. Nothing
for it but to put in a simple unity-gain buffer.

Having done so I was both dismayed and delighted at the same time to
discover that the *buffer* could easily be made to oscillate at around
760kHz. Dismayed in that I was going to have to suffer the distraction of
fixing it, but delighted (especially now that I _have_ fixed it) as it also
accounts for two similar problems I have seen, one about 6 months ago (on a
piece of commercial kit in fact!), but the other from about 6 *years* ago,
on a module I designed but left on the shelf due to this spurious
oscillation it had, that at the time I didn't have a clue about how to track
down, never mind fix!

Anyway, I turned to Bob Pease's great little book 'Troubleshooting Analog
Circuits' (it is a bit like 'Art of Electronics' - packed with many useful
hints and tips, but most of which have little to no detail as to how or why
they work: for that you need to look elsewhere). The solution is to
apparently increase the ac 'noise gain', which I did by replacing the
output-to-inverting-input short with a 10k resistor, and then placing a 2k
resistor across the inputs of the op amp (fig 8.11b on p102). 760kHz
oscillation is no more! (There is a little more detail on the other 'cures'
given in the book here http://www.national.com/ms/LB/LB-42.pdf but not the
one I've used above - darn!).

I couldn't find much on this sort of thing in the archives, which makes me
think this problem is not that frequent, and thus due mainly to my apparent
poor choice of screened cable for my test leads (it is spec'd at
320pF/meter, so I 'guestimate' I have between 0.5 to 1nF of load on the
buffer output, which is a TL082 by the way). Thus finally my question: have
others seen such oscillations from a simple unity-gain op amp voltage
follower, and if so, what is your preferred method for preventing them?

Thanks in advance for any thoughts!

Tim
__________________________________________________________
Tim Stinchcombe 

Cheltenham, Glos, UK
email: tim102 at tstinchcombe.freeserve.co.uk
www.timstinchcombe.co.uk