It doesn't want to idle after warm-up, and would rather stall when I clutch-in or in neutral.

2004 manual, 185k mi, recent plugs and coils and wires, and seafoam, new MAF, good on oil and coolant, runs great.

Please don't flame me about this, I've been searching and reading everyone's idle/stall problems for the past two weeks. There's some great stories out there. My favorite was this guy having nearly exactly my problem, checking everything I've checked, except it turned out he reversed the gas lines into his injectors. End of story! Well, I haven't touched my injectors, ever.

Except, when I'm in my driveway, I can, somehow, seduce it into an idle.

I know it sounds crazy. But when we're at the end of our journey and ready to turn it all off, if I clear my head and pause... Rev it to 4k and hold it for a second, then let it go, it will fall and nearly stop, then usually, catch itself, roll a couple times, then up to 1.2k for a couple seconds before settling into a beautiful 810 rpm idle.

But if I tap the gas, it dies.

No codes.

If I unplug the MAF it will not idle at all. This seems to matter, but I can't remember why.

I've looked for vacuum leaks, and I don't know, but I can't find one.

I've done all the various resets. No love.

I'm really so depressed about this. If anyone has an idea, please offer it up (except everyone's favorite: low compression?).

 

Can you please rephrase your question?

You seem to be having a complaint that your engine idles, which doesn't sound like a complaint at all. I am guessing that it stalls at idle, but you haven't actually included that comment, or any details around how/when/where it stalls, so I'm a bit confused.

 

Sorry RIWWP. I fixed my phrasing above.

 

Ok, I've got an opinion, and a suggestion. But since you noted that you want ideas except it, I guess I can't say it

 

lol okay okay, fair enough.

But can you tell me _why_ low compression would cause these symptoms? And if that's the case, why can I coax it into a perfect idle?

 

It's pretty simple, and common.

Point 1: Thermal expansion
When metal heats up, it expands. In piston powered cars, the pistons expand faster than the block, so the engine generally makes more compression as it heats up, as the piston rings seal tighter and tighter. Some exceptions to this in unique engines.

For a rotary, the opposite is occuring. The housings expand away from the rotors faster, and farther than the rotors expand to keep up. With a healthy engine, the apex seals are more than up to the task of bridging that gap and maintaining a seal. When everything starts wearing down, the apex seals slowly lose their ability to bridge that gap, and compression loss starts. Eventually, it's too much for the seals to adequately bridge, and compression loss reaches critical levels.


Point 2: Rotary compression is RPM based
A rotary makes more compression the faster it spins. It's a side effect of poor sealing design methods (even when healthy), geometry of the engine, inertia of the air/fuel charge, and speed of loss vs speed of compression. This is why compression tests need to be normalized to a specific RPM, because the RPM does make a dramatic difference in compression. In fact, if you assembled a rotary without any apex seals and installed it in a car, that engine could still run, as long as you kep the RPM high enough. It would be nearly impossible to start (but doable), run with extremely high EGTs (I saw one member catch his bumper on fire trying to keep his blown engine alive), but it is entirely doable.


So these two points as they apply to your situation show how the car wants to die at idle, but can be kept alive by throttle usage, and getting the ECU to compensate with the correct AFRs needed to keep it running.



The reason why I point to compression first is your engine mileage. While not unheard of, that is ALOT of mileage on our engine.


The other cheaper and simpiler option is a vacuum leak. Vacuum leaks make a bigger difference to the ECU's ability to run the engine at low RPM than at high RPM, since it's air that your engine is consuming that the ECU doesn't know is there. This can also lead to stalls, which can also be something that you can 'trick' the ECU into not stumbling over.

 

A simple obd reader and torque app would help diagnose the problem. You can check fuel trims. If they are getting high or the afr is off it will point to a leak.

If the engine load is high it will point to compression. I think a healthy engine should idle around 25 pct load.

 

RIWWP, thanks for the info, and I think I basically understand the mechanics there. The reason I don't like compression for this problem is because, as I said before, I can coax it to a nice 810 idle without my foot on the pedal. Vacuum leak is something that still haunts my thoughts. I just, apparently, don't know how to find it. Please don't get me wrong: I'm not saying I understand this thing in any kind of a real way, lol. There are soooo many moving parts in these things...

logalinipoo, I do have an ODB reader (app). When I get it into an idle, the calculated engine load is around 45%, long term trim is -22%, short term trim is 24%. Those numbers mean anything to you?

 

holy crap, those numbers are out of wack. With numbers like that, no wonder you are having idle problems.

Load should be more like 15-20% i think.

Short term fuel trim + long term fuel trim should be -5% to +5%. Yours is, but if everything was working correctly, it would be 0% long term, and 2% short term. Something was seriously out of spec showing rich long enough to give you a -22% long trim, and the ECU is now trying desperately to correct that and is adding 24% back.

Clear your fuel trims (20-brake-pedal-stomp or battery disconnect), then let it idle for 10 minutes or so, then check the fuel trims again and report back.

 

Do what pearl said. The load might drop if you get the trims settled out. If it doesnt than that engine is probably done.

Clean the maf also.

 

All that is good. Especially the MAF. If still wonky, look closely at the accordion tube for cracks, and every one of the lines that come off it.

Those trims say that the ECU thinks it is running crazy rich, and is doing all it can to sort things out.

 

Im not thinking its a vac leak. It seams like its not getting enough air.

If it was getting too much unmeasured air than it would lower the load and develop high fuel trims.

 

Thank you guys for the input. I'm eating humble pie over here, again. I want to say in my defense that I did that twenty-pedal-stomp thing a half-dozen times in the past few weeks as I tried different things.

This time around, I did the stomping and started it up, coaxed it into an idle, and started watching the odb values. As with the times before, nothing had changed (LTFT at -22, STFT around 25). This time, however, I knew I was supposed to let it idle for "10 minutes or so" per RIWWP's suggestion. So I just played with my odb app for a while -- sort of got lost playing around with that thing, it's pretty neat. Anyway, after what felt like 10 minutes or so, I noticed that for the first time my LTFT and STFT values had actually changed. Astonished, I start tapping the gas (which, before, would cause it to hunt and stall) and nothing bad happened. Turn on the AC, tap the gas, all good. Wow.

I've driven about 30 miles or so since then and the LTFT at idle is settling around 7-8 and STFT likes to be 0, but fluctuates a little. Calculated engine load at idle is still in the high 30s - low 40s.

Every other time I did the reset, I immediately started driving it. I thought that was what I was supposed to do. I guess letting it sit and idle for a while is some kind of extra little magic? I HATE MAGICALNESS IN MY CARS!!

Anyway, just to add slightly more depth to this story, for the record: This started a few weeks ago when the car was dying at idle if the AC was on. I did my version of the 20-pedal reset with no change. Then I cleaned the MAF (it was filthy!). And then it wouldn't idle at all. Replaced the MAF with a new one gave me no change whatsoever. Replaced the air filter; nothing. Search for vacuum leaks, can't find one. Clean the throttle body (carbon built-up); nothing. Clean the ESS (also filthy!); nothing. I'm doing the reset after each of these.

Moral of the story: If having idle problems and doing the 20-brake-pedal-stomp, be sure to let it sit there and idle for a while.

I realize I went about writing this thread completely backwards (complete list of symptoms and procedures at the end, literally after the problem is solved), but I thought it might be worth having it all out there. I use this forum A LOT these days, and I always appreciate it when I'm reading something from ten years ago that is a complete story.

Thanks again, guys!

 

Actually, moral of the story is:
Any time you touch anything that could affect fuel trims, you need to reset them, and the fact that your long term fuel trim did NOT go immediately to zero, means that you never actually reset the fuel trims. If you were getting the oil pressure needle sweep on the reset, then i might be wrong about the fuel trim reset method. Yank the battery cable as the alternate option.

The dirty MAF was likely the cause from the beginning, and cleaning it was the appropriate thing, but that suddenly changed how much air the ECU thought it was seeing, despite being the same airflow as before, and without a fuel trim reset things stayed wacked.

Your LTFT is still too high, I recommend pulling the battery cable for a few minutes to force them clear, then let it idle for 10 minutes again to build up idle trims correctly.

 

I was, in fact, getting the oil pressure needle sweep each time I did the stomp reset. I also had coaxed it into idle, and let it sit there for long periods.

I just hadn't done a stomp reset, and then immediately coax an idle and let it sit there for a long time. I always did the stomp reset, then take it for a drive. By my reckoning, the stomp-reset didn't work (at least for values at idle) until I did it and then let it sit there for a while. This is the magicalness I was talking about.

 

Well, a fuel trim reset immedaitely puts the long term fuel trim to 0%. By letting it idle, it corrected down from the big numbers it was at, which isn't the same thing as rebuilding trims from 0%.

The difference is going to be mainly evident everywhere else in the rev / load range. You almost certainly have really wacked out fuel trims all over the place in other locations on the grid (think rpm x load excel sheet), not just at idle. You can get them corrected over time, but it is really better to just clear the entire table back to zero and build from there. A fuel trim of -22% in a high load high RPM cell would scare the crap out of me. That is engine destroying numbers. The default map with no trim is going to be inherently safer.

I'm puzzled that your 20-stomp wasn't reseting the fuel trims, since i thought sure it did. I guess I was remembering incorrectly. Pulling the negative battery cable definitely does (since idle issues due to fuel trims resetting is common after a battery disconnect).

I'd still recommend doing that to clear the entire table, just to be on the safe side.

 

I thought after the stomp you should start it and rev to 4500 for 30 sec or so , the 3500, 2500, 1500, and idle. I remember reading a paper about it somewhere

 

Yeah, there are various things you can do after the reset to accelerate the learning pattern, but none of them are really necessary. The learning patterns looked for are specifically designed so that an average driver will encounter each pattern relatively quickly in their normal driving cycle. The only result from not hitting each pattern immediately is that you are operating on what has been learned already until you do hit the pattern. Either way, the first time you hit the pattern, you are on the default settings until the learned data is stored, which happens immediately. No penalty or problem with that.

 

The biggie from drivability though is idle. You cannot cheat that one. I used to try to chase it, but read one of the articles here and switched to just letting it idle, and life improved dramatically.