I was recently successful in resolving a nagging problem with my VW at
cold-start and idle. I usually do a "lessons learned" write-up for my
personal files, and I thought I'd share the story with the group. I hope
it's not too lengthy...

1986 VW GTI, RD gas engine, CIS-E injection.

For quite awhile, my GTI has had a couple of problems with idling, which
seemed to happen after the last repair session at <shop name deleted>. I
recall that they had had a problem starting the car after they had done
major transmission/engine seal work. They had tried to start engine, and it
wouldn't idle, and I think the mechanic started adjusting things.
Eventually, I think they found that something was disconnected, they
connected it, and it started and ran, but I don't think they set everything
back to original settings. They also may have played with mixture, since
the sealing plug was not in place on the fuel distributor. I wasn't able to
look over their shoulder while they were doing this.

Anyway, my problems were:
- Started very easily, but after about a minute, idle got very rough. You
needed to keep foot on pedal, or pump every few seconds. Eventually, when
engine warmed, it ran pretty well, and overall, engine power and
acceleration were pretty good.
- Looked at readings for Oxygen Sensor (OXY) and current for Differential
Pressure Regulator (DPR). OXY voltage, when engine warmed up, should have
been varying between about 0.3V and 0.7V. However, it started off at about
0.5V, and then slowly went down to about 0.2V. According to Bentley manual,
this suggests the sensor is seeing a mixture that is too lean. Is the
sensor okay, or is it really reading a lean mixture? (My OXY sensor
dashboard reminder had recently come on, at around 100,000 km.)
- Current for DPR starts off around 80mA or so, and is supposed to slowly
drop to around 10mA after about 20-30 seconds. However, my current went
down to about 20mA, and never went lower.
- During troubleshooting, I discovered that the idle switch on the throttle
body was defective. The switch was broken, and would not close when the
throttle pressed it at idle position. Thus, the fuel computer did not know
that throttle was at idle position. Did this switch break at the repair
shop??

I didn't want to start adjusting the mixture too much, and wanted to
understand the sensor readings. Read over Bentley many times, and also
consulted rec.autos.makers.vw.watercooled archives - good stuff there!

To look at the idle problem, I made up a 15Kohm pot with wires to simulate
the coolant sensor. This sensor is about 3-6K when cold, and drops into the
hundreds of ohms when hot. If nothing else, I might be able to fool the
system into thinking it was colder than it was, and resolve the rough idle
at initial startup.

I set up my digital multimeter on OXY sensor, and an analog current meter on
DPR, and the pot on the temp sensor. Set temp sensor to around 3-4 K (about
10 deg C).

Started engine; it started easily. OXY was about 0.5V, and DPR was around
100mA, and slowly went down to about 30mA after about 30 seconds. Wasn't
stalling, and as I adjusted the fake temp sensor, I could see the DPR
current change. Hotter engine made current drop, which caused a lower
commanded richness. Mixture should be richer when engine is cold. I
lowered the temp sensor resistance a bit more, and now current around 20mA,
with OXY still around 0.2V.

I decided to try a mixture adjustment. Using long 3mm allen key, turned the
mixture 1/4 turn clockwise, which makes mixture richer for a given DPR
current. OXY rose slightly, and DPR current dropped to about 15mA(?).
Hmmm... this seemed good. I turned the mixture another 1/4 turn clockwise,
and now the DPR current dropped to around 8mA. More importantly, the OXY
started rising to about 0.6 and 0.7V, and then it started to waver between
about 0.7 and 0.4V. THIS IS THE CORRECT OPERATION!!!

Revved engine, and it worked fine. Temp sensor setting now around 1K or
less. I decided to shut off engine and connect real temp sensor. Did that,
and engine started and idled fine. This time, OXY starts at 0.5V, and after
30-40 seconds, starts the 0.3 to 0.7V varying, as it should. Current still
around 8mA, with some varying when running. (Couldn't actually tell DPR
varying, since this meter has quite heavy damping.) I decided to tweak
mixture back about 1/8 turn CCW (less rich), and now DPR current around 10mA
average, as Bentley manual says. CORRECT!

Took car for a drive, and it went just fine. I also reset the OXY sensor
service reminder (on firewall behind coolant overflow tank).

THEORY:
=======
I think that somehow, the auto service tech had turned the mixture screw
about 3/8 turn CCW, making mixture too lean for a given DPR current. The
OXY reading of 0.2V was telling system mixture was too lean, and the module
was driving DPR current to 20mA, trying to make the mixture richer. 20mA
was as hard as it could drive the DPR, given all other constraints, like
engine temp. When I then turned the mixture screw CW, it made mixture
richer for a given DPR current, and the OXY voltage showed that.
Eventually, when I turned the mixture screw far enough CW, the mixture
actually got richer, and OXY went above 0.5V. From that point on, the OXY
sensor/DPR current control loop was working correctly.

The real proof was the next morning, when it started up instantly, showed no
stumble or hestitation when idling, and ran smooth and easy on the road -
like a new car!

Thanks for reading, and for sharing your own repair stories!

WT, Winnipeg, Canada