'89 80 - miss and idle instability problem

[Bryan Lally <bryan@mst.lanl.gov>]

My Audi goes vroom at idle.  No, this is a bad thing!  It also misses.
Anyone want to play a remote diagnosis game?  Here are the details to
date.

Car: 1989 Audi 80.  2.0l 4 cylinder.  Front wheel drive, not a
quattro.  72,000 well-taken-care-of miles.  KE-Motronic fuel injection
(the mechanical fuel distributor, continuous injection setup with a
Motronic computer unit).  The car was running fine until about 1 month
ago, when it started this behavior suddenly.  As we started to
investigate, the car suffered from the cappucino incident (failed
coolant pressure cap), which was not connected to the current problem,
and has been fixed.  I realize that the offering to the Audi gods for
the pressure cap may well have been insufficient.

The short story:

The car exhibits 3 symptoms.  All happen somewhat sporadically -
sometimes 2 seconds apart, sometimes 2 minutes.  The symptoms are

	1) vroom at idle,
	2) miss, most noticeable under load, but seems to occur at any
	   speed or load
	3) voltage blip at test point used to read motronic fault
	   codes.

The "vroom" (1) is a momentary increase in RPM from idle by 3-400 rpm.
The idle goes back to normal immediately.  It is caused by the idle
stabilizer responding to it's control signal - i.e., the stabilizer
works, and the computer is generating the signal to make it go vroom.
Why is it generating this signal is the big question.

The miss is a short miss, but seems fairly complete, like spark
disappeared for a short period.  Not a light stumble.

The voltage at the motronic test points is normally 1.5v (engine
running), or 12v (engine not running, key either on or off).  During
the "blip" the needle swings to ~5v.  Of course, the actual spike
could be higher (it's probably 12v), but it is short, and the needle
doesn't react that quickly.  The correlation factor for the voltage
spike and the vroom/miss is 1.0 - they _always_ happen together.

The long story:

The idle vroom is annoying, and a bit of a safety hazard in parking
lots.  The miss is bad enough that it must be fixed.  We've been over
everything we can think of, and have come up blank.  Any ideas before
I take it in to Mr. Dealer?  Or, just want to play along and predict
what Mr. Dealer is going to find?  I hate to admit defeat on this one.
Here's what we've looked at.

Alternator - we disconnected the alternator and ran the car on the
battery, in case the diodes or voltage regulator were causing voltage
spikes that were confusing the computer.  Nope.  Rule out faulty
alternator/regulator.  We also looked at the voltage across the
battery terminals, and saw no spikes.

O2 sensor - we disconnected the O2 sensor, and ran it open loop.
Nope.  Rule out faulty O2 sensor and exhaust manifold leaks.

Intake leaks - we searched for intake leaks, false air, etc.  We found
none.  All the hoses look to be in excellent shape.

Slightly temperature dependent? - the symptoms don't seem to appear
immediately after a cold start; it takes about 1.5 miles for anything
to happen.  The symptoms do occur immediately on a warm start.

Water temperature sensor/switch - we have complete gone over the
temperature sensor and switch, measuring, testing, etc.  We were
really interested in this, as the problem seems to be _slightly_
temperature dependent, and since we got temperature sensor fault codes
from the computer.  Finally, we determined that the temperature
sensor/switch should have a value of about 200 ohms when the engine is
fully warm.  So we put a 200 ohm resistor in the wiring harness, to
simulate a warm and perfect temperature signal.  Nope.  Rule out the
temperature sensor/switch.

Idle stabilizer - other than the vroom, the idle is perfect.  If we
let the clutch out slowly, feeding in no throttle, the idle stabilizer
will keep the idle speed up and we can drive away.  The stabilizer has
been cleaned.  If the stabilizer is disconnected (with the engine
warm), the vroom will (of course) go away.  The miss and voltage spike
remain, and the car idles nearly perfectly.  Rule out idle stabilizer.

It was suggested that since the Motronic computer is near the lower
heater duct in the front passenger foot area, that leaving the heater
off might prolong how long the problem takes to commence.  No, this
doesn't happen.

What to do next and possible theories:

Take car (and wallet) to the dealer.  Make a proper offering.

Check power input to computer, at the computer connector.  If this
voltage is intermittent, the computer will be constantly rebooting.

Check temperature sensor value at the computer connector, to look for
faults in the wiring harness.  Not very likely, as the problem occurs
with the temperature sensors disconnected completely in the engine
compartment.  This test says that the temperature wire is not going
open, as we created an open and the fault still occurred.  It is
perhaps slightly possible that this signal is intermittently getting
shorted to ground.  But it has no correlation with driving over bumps,
etc.  It will happen idling in the garage, not moving.

Open up the computer unit, and spray "cold" spray on the components
while they are operating and see if the problem can be made to
disappear or get worse.  Of course, squirting stuff on the operating
computer is risky and may end up with having to perform the following
item.

Replace the computer - expensive, if this isn't the problem.
Expensive, if it _is_ the problem.  Anyone know an approximate price
for one of these?

Resolder connections on computer.  I've looked at the boards, and
there is not much that can be done in there.  This isn't hand-soldered
hardware, although I've heard of someone doing this on a K-Jetronic
setup.  A screwup results in the previous item.

Because of the abruptness of the voltage spike, and the similar
performance at all speeds/loads, I have concentrated on electrical
problems.  It just doesn't seem to me that fuel flow is the problem
here.  Let me know if this is a bad conclusion.

As always, thanks for any advice.


	- Bryan

'94 BMW K75S		_red_
'94 Honda XR250L	with extra needles and thread
'89 Audi 80		I think I can, I think I can ...