On Sat, 21 Nov 2009 13:54:53 +0000 (GMT), Jim Lesurf
wrote:

In article , Don Pearce
wrote:
On Sat, 21 Nov 2009 13:20:40 +0000 (GMT), "Dave Plowman (News)"
wrote:

Just to say thanks to all for the tips - a couple of 10µF
electrolytics mounted on the PCB sorted it.

Only thing is what to do with it now - it's a great lump of a thing.
Ebay?

Sounds like a good idea. It was interesting to see how a real consumer
manufacturer approaches the tricky business of cost-cutting.

There may be swings and roundabouts. Many years ago I decided to avoid
using rail-ground-rail caps on the amp board as they tend to inject
distortion from half wave ripple into the amp via any ground imperfections.
But you then need to ensure the amplifier has an inherently high ability to
ignore rail variations.


That is where really well designed grounding is vital. Stars rule!

And I wonder how many died through oscillation. Warrantee repairs are
horribly expensive things to deal with.

Given the above was the cause of the problem I'm curious as to why the
oscillations waited until after warranty. :-) Gives me the feeling that
something else has degraded.


If oscillation started because of the inevitable degradation of a
component (a cap losing value somewhere, maybe?) then they didn't go
through the design centering stage of development. A "what happens
if..." scenario chase is really quite important.

Here's a little stability thought I've been pondering - and modelling
in Spice. Cdom goes around the voltage amplifier, then the output
stage follows. Now, if the output stage is simply a voltage follower
with no gain or inversion, why would I not connect Cdom to the output
of that stage, rather than just the one transistor's collector? Would
that not give an advantage for HF distortion suppression, since it
would not matter that the overall feedback was degraded by lost HF
gain in the V amp...


d