Violently cyclic idle - MC engine

[Huw Powell]

> Someone else posted a clue that checks out.  What is really happening is that idle is surging
> up to 3000 rpm, but the ECU gets pissed when it sees 1500 rpm with the idle switch closed and
> chops the fuel frequency valve.  As idle drops, it restores it and off we go again.  This is
> demonstrable if you get it doing this at rest - pull the plug off the throttle body switch and
> up you go to 3000 rpm.

Very interesting.  I'm glad you did that idle switch test, because it 
tells us so much more about what is happening "behind the curtain." This 
behaviour is completely consistent with a *metered* air leak.  You've 
checked the obvious, ie, malfunctioning ISV.  I had a similar problem on 
my coupe, a long time ago (idle stuck up around 3k), and that was a 
gummed up secondary butterfly not closing.

Of course, your car has just one big butterfly in the TB, I think (not 
really an expert on the turbo machines...).

So without further reasoning, since it is fairly easy to do (and a snap 
for a seasoned tech like you), I'd recommend opening up the intake tract 
and having a look at the guts of throttle body, if you haven't already.

> What confuses me is that it usually idles at 1050 rpm - almost at random.  I've done the full
> proper setup (ISV at 430mA, fuel pressures correctish, 1.25%, etc.) but I can't get a stable
> idle below 1050.  If I set it where it should be, it won't start from cold and often stalls.
> There are no air leaks - pulling the dipstick stalls the car INSTANTLY.  The only thing I've
> ever seen produce this sort of effect is a defective metering head - the infamous upper
> O-ring.

-- 
Huw Powell

http://www.humanspeakers.com/audi

http://www.humanthoughts.org/