I've had one or two like that, so far. Not nearly so improbable, being much
simpler circuits, I suppose you could say primarily due to my inexperience.
Take, for instance,
http://webpages.charter.net/dawill/Images/20MHz_Amp.gif
The 2N5179s oscillated at some ungodly frequency, way beyond my 200MHz scope
(unsurprisingly, as they are >1.3GHz transistors). Probably a combination
of "bad" wiring and a not-too-great design (the next revision did away with
the 3904 emitter follower drivers, which probably coupled too much
end-to-end, exacerbating the problem).

Besides shitty performance that just didn't look right, I discovered this
one by touching the circuit. Oftentimes, parasites respond to the most
useful tool, the finger: it's dextrous, a thermometer (sometimes a
voltmeter!), has a reasonably consistent capacitance and resistance.

The other prominent example was breadboarding discrete ECL gates. Great
performance, it's wonderful to see 5-10ns edges coming from mere 2N3904s (on
the breadboard, no less). Problem is, the stray capacitance and rail
inductance sucks so damn much when breadboarding that any more than about
five gates makes the thing oscillate. It only adds insult to injury that I
was running out of space, so had to patch wires in to an adjecent breadboard
to continue. Icky!

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms

"John Larkin" highNOTlandTHIStechnologyPART.com> wrote in message
news:dgp0b4d5p6dvborke4b2viiln0gvmf2rsu@4ax.com...