Hi all, been a while since I've posted, but everything has been running OK. :P I searched and could not find this specific issue.

Car is fully maintained, this problem has shown up immediately. Cap, rotor, wires, plugs, air and fuel filters, cam and crank sensors, O2 sensor, MAF, and fuel pump relay have all been replaced recently.

1996 NA 850, AT

The car will crank up easily when cold. However, after idling for ~1 minute, it will stumble and then shut off. It will spin and try to crank but usually won't (Pulling the plugs, they are wet with fuel). If I then pull the fuel pump fuse, it will start up after ~5 seconds and run on the limited fuel available for a few seconds. I pulled the fuel regulator vacuum line and it is dry. I can read the coolant temp with my OBDII reader and the temp sensor appears to be operating correctly. It does not throw a CEL or have any codes stored.

If I crank it and hold the throttle open with the RPMs at ~2500 until the engine warms (doesn't take long, the thermostat appears to be operating correctly) it will then go to idle and run as well as it ever has. I can then shut it off and it cranks back up just fine. Acceleration, idle, and coming to a stop, the engine runs normally.

As I understand it, the ECU has different injector timing for starting, for a cold engine, and for a hot engine. It's almost like the injectors are being held in the "startup" timing mode until the coolant temp sensor reads the engine is hot, then it goes into the correct fuel trim and runs perfectly.

I just can't figure any likely mechanical reason for this. I changed the plugs just for kicks and it did not help. Looking at the live data on my reader when it is trying to cold idle, it appears that the O2 sensor is reading a rich condition, but the ECU is not adjusting the fuel trim. Obviously it is getting plenty of fuel, and spark quality and timing seem to be ok, being that it starts so easily when not flooded or already hot. I would think that a leaking or stuck injector would give problems all the time, not just under specific conditions.

I found a used ECU on eBay for $40, so I'm gonna try that. I'm open to any suggestions.

 

Is your ICV working? Also, try unplugging the MAF to see if that helps. It will toss a code but you'll know if it's effecting your cold idle.

 

Just as a guess I'd pull the idle air control motor and clean it out. Remember to keep the motor up if you're going to spray it with a cleaner. Carb cleaner works pretty good for that. Might be that the motor isn't reacting as fast as it should or could and it's not letting in enough air with that rich mixture. The computer should modify the injector dwell (on) time as the engine warms up. When it's cold it injects a ton of fuel and until the heated o2 sensors get heated they can't give the computer any help on mixture. The other thing to check is that the temperature sensor actually does change as the engine is getting warmer.

 

I forgot to mention, I did try switching the ICV with a known good one (off my 850 turbo), to no effect. Same with the TPS.

I unplugged the MAF and it did manage to run until the engine got hot, but it missed and stumbled the whole time. Tomorrow I will remove the MAF, clean it, clear codes and try again. I could also borrow the MAF from the other car and try it.

I still doubt this is the problem however, because the MAF on both cars is reading the same at idle (about .7xxx @850 RPM if i remember right). I would think that if the MAF had failed the reading would be off.

 

Scratching my head ........

Any chance it's the coolant temp sensor taking too long to report the correct temp. Like the end of it being covered with crud that insulates it and makes it slower to respond to the actual temp ??

Did you replace that list of stuff to try and correct this problem ??

 

I can view temp sensor on my OBDII reader and it appears to be working. The sensor shows temp rising constantly, and it reaches operating temp pretty quickly, 3.5 minutes or so with me holding it @ 2500 RPM, even with the air temps here at 30-50 F. I think that's normal for both the temp sensor and the thermostat.

The problem occurs at <1 minute or so, when the coolant temperature is actually still low.

Thankfully, no! That stuff has been done over the past 4 years at different intervals. I had an similar problem to this one a couple years ago, that my sort-of shotgun sensor replacement seemed to help (at that time, it occurred when the engine was warmed up, but was also a flooding issue). I never was 100% certain that I had fixed it, but the issue disappeared at some point. I'm wondering if this is just the ECU acting up again, maybe stuck on a bad fuel trim. I've tried leaving the battery disconnected for over 12 hours, and that didn't help either.

 

What you still need to find out is if the coolant temperature sensor is sending the right temperature. You need to physically take the coolants temperature and compare it with what the scanner says the temp is so you know if the computer is flooding the engine because the sensor is out of range and telling the computer the engine is colder than it really is.

 

Sorry, didn't see your reply until just now. You're probably right.

I was pretty sick all weekend (food poisoning sucks) so I didn't mess with it until Sunday evening. I removed the fuel rail and tested for leaks just to be sure. Key cycled the fuel pump every couple minutes for about 15 minutes, lots of pressure at the schrader valve, but not a drip from the injectors or regulator.

My new ECU had arrived Saturday, so I installed that and cranked it up. No flooding, but it did idle up to ~2K RPM after about two minutes at idle, then slowly worked its way back down to 850 RPM. It did this again this morning, and when I went to lunch, and after work.

That 4th time, it set a coolant temp sensor out of range code at the end of this "idle up" cycle just before it dropped back down to normal idle.

I have a remote temperature monitor (it's actually for cooking, but it has found several non-culinary uses already :P ) that I can stick in the expansion tank. I'll try it tomorrow and see what it looks like.

 

That would work. Just to find out if it's right on or possibly 10-20-30+ degrees off. The thing about temperature sensors is that they can be off enough to mess up your hot or cold start but still be within their operating range and not toss a code.

 

Took two tries, but I finally got it to fail while I was watching.

Cranked it at lunch, and idled all the way up to operating temp, no funny idle.

Tried it again after work. The temp moved linearly up to about 90F, and while I was looking at the expansion tank's temp monitor, it idled up to 1500 RPM. I looked down at my live data, and the coolant temp was showing -14 F. It gave very low readings for about 20 seconds, then jumped up to 120F and appeared to work normally up to 180F, when the thermostat opened and I could verify via its reading via the other temp monitor. Did not throw a code, like you said, this time it did not move out of range.

It appears to just give bad readings in the range of 90-120F or so. Or maybe it's not necessarily the temp, but the speed of the change in temp, since it is rising quickly in that period. It seems to give a good reading when the temp is stable, at any rate.

I still think my old ECU had an issue; this new one seems to be able to handle the problem, and did recognize an "out of range" sensor at one point. Maybe the old ECU increased the injector dwell time, but did not increase the IAV correspondingly as this one seems to do?

 

Wow, good catch.

 

New coolant temp sensor installed for about 10 days, no flooding since then. I changed back to my original ECU a few days ago - the new unit did not like my lack of emissions equipment, and also didn't handle tip-in acceleration (especially when cold) very well.