I've been working on this problem on and off for almost 4 months and starting to wonder if I am overlooking something obvious because I've been at it too long. It was my daily driver up until this engine issue and honestly, I would rather drive this to work than my 2003 WRX so as much as I hate to ask, I need some help to review my problem.

My engine had been running like a champ for well over 2 years since I originally bought this wagon home and repaired a leaking head gasket. I put a least 50 miles on it daily with both freeway and city driving. About 4 months ago, it fired right up as always but then stumbled for a bit and died. What? It has never, ever, done that since the first day I fired up after the rebuild and that was probably 20,000 miles ago. It is the SPFI engine so I never touch the gas pedal when starting. It will restart instantly over and over and runs great for the first 2-3 seconds before the stumbling begins and barely stays running or dies. Sometimes, it will continue to idle very poorly in the stumbling condition for 3-5 minutes, right on the edge of dying. If I step on the gas, it will instantly increase engine speed very quickly and strongly until the desired RPM is reached, but then it begins to stumble again as soon as the RPM is held steady. If I had to guess, I would say it seems to be running very rich during the stumbling phases but I don’t know for sure. It runs fine as long as the engine is increasing in RPM. If I hold the throttle steady at any set RPM, it begins to stumble. I can race the engine over and over by blipping the throttle over and over and it seems okay until I try to let it idle or hold a steady throttle position.

• I checked and cleaned the MAF. No change. I then replaced the MAF with a known good unit just in case. No change. The MAF was not dirty and I do get a rising and falling voltage signal from the sensor when I back-probe the signal wire at the connector. Wiring harness checks out per the shop manual procedures. I'm working from an 1989 dealer repair manual but pretty sure it's the same for my 87.

• I don’t have a check engine light and the only code I get from the little red LED hidden inside the ECU is the 5 or 6 short repeating flashes indicating the ECU build series.

• The 2-wire temp sensor used by the ECU is fairly new and reads the proper resistance range for the engine temp.

• The IAC seems fine. I took it apart, cleaned it and confirmed it opens and closes when power is applied. When it is closed, it is sealed tightly and flows air easily when it is open. It is a 100% open or 100% shut, although I don’t know if it is modulated by a PWM signal from the ECU. It seems to be active all the time during my testing. It is not a bi-metal heated type valve that closes slowly over a 5-6 minute period after a cold start. If I unplug the IAC connector the engine won’t run at all so it seems to need that bypass air for sure, at least when cold.

• I thought that maybe I had a large vacuum leak but a smoke machine only showed a very tiny amount of leakage around the throttle shaft at the TPS.

• The TPS is set correctly with the idle contact switch inside closing only at idle per the FSM. The resistance of the TPS matches the factory manual for all positions. The resistance changes are smooth throughout the rotation from idle to full throttle.

• Fuel is fresh, filters are clean. Constant 23 PSI measured under the hood with ample flow when I open the line for a test.

What causes an SPFI EA82 engine to start perfectly with no issues but then immediately run poorly, except when under acceleration? The fact that the engine starts normally at every attempt but seems to only run well when accelerating makes me think of an ECU that is going in and out of a closed loop/open loop program but I know very little about that possibility with this OBD1 technology.

• I recently tried a new Bosch 1-wire O2 sensor but it made no difference. Figured that was a long shot but it was cheap and easy to change.

At this point, I’m wondering if I can’t see the forest for the trees. Am I missing something obvious? She's to pretty to just be sitting in my driveway.

The first thoughts that come to my mind are injector, fuel pressure regulator, or the IAC valve.

While I don't think the IAC is the main culprit, those valves can be very tricky. I've heard (and this is conjecture) that they can be bad but still test good. Because they are a heater, their resistance changes between cold and warm. So they could test okay cold but are out of spec when warm.

Again, this is conjecture so don't quote me on this, but the IAC isn't just for idle. Allegedly it'll also "catch" to maintain air fuel mixture when the throttle snaps shut. It essentially acts as a buffer to compensate for a rapid change in air entering the intake.

I've been chasing a mixture issue myself for months as well so I feel your pain.

But if I had to guess, I'd say it's fuel delivery related OR some vacuum issue (be thankful you don't have to mess with the vacuum madness that is the factory carb setup).

:]

Okay. Sketchy repair maybe (because I don't know 100% what I did) but it is running as well as it ever has. I set out to shoot a short video of the issue last night and decided to check the IAC one more time. It is NOT a b-metal control. It is an on-off solenoid that has the flow metered by a small adjustment screw that is in the throttle body, not in IAC itself. That screw allows minor adjustments to the idle speed while the IAC is active. I turned it in a few revolutions and then backed out to the same spot where it was. I went to start the engine and then do the same adjustment while it was idling but nothing else was required. The engine started as well as ever but never stuttered or bogged down. It idled perfectly and I've now driven it about 45 miles with zero issues. Maybe the port was blocked inside and turning the screw opened up something that has been blocked or too restricted. It runs like a champ now so we'll see if it ever repeats again.

The really cool part about all of this is trouble that I figured out a way to use those reversed TPS sensors (SERA484-10) that others have experimented with. I'll detail that in another thread.