Misfire at idle only, worse in the cold

OK, I'm getting to the end of my chain.

2002 WRX. Put a new engine in after overheating the old one. Stupid me. Idled a bit rough from the start, but calmed down a bunch after chasing down stupid vacuum leaks. Still idled a bit rough, but nothing big. It threw one cylinder 3 misfire within the first 20 or so miles, while idling in a car wash. Reset it, nothing came back. Until recently.

As the weather has gotten colder, it's gotten worse. The colder it gets, the more likely it is to throw a misfire code, and the misfires have progressed to all four cylinders. The codes only show up when the car is cold, but the rough idle remains even when warm.

Since this is the car I want to "just work" and there's been a family emergency where I need to jet back and forth between the city and my folks place, I sent it to a reputable shop last week.

They did a "smoke test" and did not find any vacuum leaks. They told me they could see on their live OBD-II reading that there were misfires on all four, but said they could see no reason why. Fuel good, spark good. They (and another shop that looked at it superficially as a favor) said my "fuel trims" were spot-on. Whatever that means. After checking that the timing was set properly (it was), they threw up their hands and said it must be a computer glitch, take it to a dealer.

Above idle, the engine runs just great. I've put about 1300 miles on it now, no codes ever when I'm not idling it cold. It got 28mpg on one light-footed all-highway trip back when it was warmer, but the last driven-like-a-GRM-reader tank was 19mpg.

The coil packs are original to the car (113k) AFAIK. The plugs are new, OEM, properly gapped. I did not check the valve lash (stupid) when I put the reman'd heads back in, but the Subaru shop that sold it to me assured me they were ready to just drop on (and the car makes good power). The battery is five years old, I think. The O2 sensor is new with the engine, the non-generic one, top-of-the-line.

I have my hunches, but what does the collective wisdom think? What are my next steps in diagnosing this?

Does it still have the stock air box and filter or is it an aftermarket C.A.I.?

My Brighton, after the rebuild, had a K&N open element "cone" type filter attached directly to the mass air meter. When I recently swapped out the EJ-18 intake manifold and injectors for the proper EJ-22 bits I also swapped in the Legacy intake tube and filter box with a stock type filter. It starts much easier and the idle is now quite a bit smoother. If it has a C.A.I. maybe try going back to stock?

My $.02.

Stock air box.

Stock everything except a catless up-pipe and a axle-back exhaust. And a Cobb 93 "Stage 1" tune.

Also worth mentioning - I sprayed the snot out of the AFM with some AFM cleaner this weekend. No change. And I cleaned all the engine grounds (save the one at the starter), no change (all were in good condition).

Does it stumble at all if you mash the throttle?

I'm talking from idle straight to W.O.T.

Mine started doing that when the weather turned cold, hence the recent intake swap. It behaved like a carburetor with no choke; it was EXTREMELY "cold natured" unless it was up to full operating temperature before trying to drive.

Does the engine temp make a difference or does it have a whingey idle regardless of whether or not it's warmed up?

Tight valve somewhere. If you didn't adjust when it went together...Cold misfire will most likely be a tight valve or a shrinking intake gasket allowing a leak when its dead cold. Those two things always show up worse at idle, and cold.

It won't be ignition. That almost always shows up under load.

In reply to The_Jed:

Nope, perfect behavior except for this stupid idle. Mashing the throttle just makes the car leap forward. :)

Idle is "whingey" regardless of whether it's warmed up.

In reply to Streetwiseguy:

Sheeeeeeeet. Maybe I should re-torque the intake manifold.

"tight valve" means the valve lash was set too small? Not questioning your diagnosis, just trying to understand the situation. What's going on here? My thought would be that too-tight valve lash would just result in extra wear on the cams... but I know far too little about engines (really wish I had measured that valve lash now...)

OK, more questions (since I'm obsessing over this tonight):

If it is a tight valve (I looked it up - essentially you're saying the valve isn't closing 100%), how much damage am I doing by driving the car like it was meant to be driven?

Compression/leakdown test, or actual clearance measurement would reveal this?

Finally - how realistic is my suspicion that a faulty coolant temperature sensor could cause this?

In reply to peter: I'm no subie expert, but I can answer these questions. First, driving with tight valves will over time result in damage to the valve or valve seat, but, for most engines, it takes a long time. Don't panic, & don't stress over needing a compression on leak down test, just check and adjust as needed.

Second, a faulty coolant temp sensor could definitely cause an idle quality problem.

Bad coolant sensor should also set its own code, always assuming it gets out of spec. I lean toward tight valves as well.

A tight exhaust valve will not transfer heat out of the face of the valve as efficiently as one that is closing properly, and will burn given enough time. Tight intake doesn't get as hot, so less of a worry.

Tight valve clearance is actually a very common problem on Honda V6 engines, since its a fairly big job, nobody does the valve adjustment when its due, so up around 100k they often start to set misfire codes.

Apparently you can replace the shims on this car without removing the motor, given the proper special service tool. Doesn't look like a cheap job either way. Should have checked those clearances...

My thought on the coolant sensor is that it's got an offset in it (reading warmer/colder than it should) or it's got flutter at low temps. Hopefully a quick look at an OBD-II scanner with the engine cold will let me see.

Anything else this could be, that's cheaper than a valve adjustment?

If the heads were fresh rebuild the lash could have been in spec when new. Sometimes things tend to...well pound in after a few miles. Always recheck lash after a short drive.

Canoe

Do let us know if you come to a solution with this. My Impreza has a similar issue I haven't gotten to the bottom of yet.

Agree with 44dwarf. I learned the hard way over the years (particularly with bucket/shim type valves) to add about .002 when putting things together, the clearance will close up quickly when the engine is first run. The valves will seat in the head and as they do they 'grow' outwards, closing the clearances.

from searching more thoroughly online, this sounds like a semi-common high-mileage WRX problem. I really don't have time or space to do the work myself, I hope I can find someone who can do the adjustment with the engine in the car. This car is turning out to be an expensive albatross...

Thanks for the help folks. Off to find a reputable local scooby shop...

I'm going to ask at work if they've seen this before. I haven't been there long enough to know what common issues these things have other than head gaskets

I'd appreciate any input you've got corytate, and I'll remember to provide an update ValuePack... thanks!

I read through the previous posts, and unless I missed it, have you cleaned/checked the IAC? Its right on top and the gasket is about $6. I would do that before the coolant temp sensor which is about $30. And I would definitely do that before checking the valves. I say that just because of the time and the valve cover gaskets are about $30 and the Honda Bond is $10 (personal favorite).

And yes you can check and replace the shims with the engine in the car and also without the nifty $300 tool. If it comes to that point (and you want to do it yourself) let us know and I am sure that I or others can walk you through how to do it.

In reply to Ojala:

I replaced the gasket on the IAC when I reassembled the motor. Cleaned it up pretty well too.

How are the plug wires ?

I am thinking a sensor or fuel injector issue. It would help to know what the MAF and IAT voltage was at idle. I dont trust using fuel trims for diagnosis at low load. Fuel trims can be in an acceptable range and yet the spray pattern and delivery will be all over the place for one or more cylinders. The toughest thing with misfires is that so many things can be the cause.

I had an aggravating misfire under load so I repeatedly checked all the plugs and moved them and the packs around. It wasnt till I realized that I had never actually removed the plug from the spark plug socket that I found a cracked insulator.

In reply to iceracer:

Coil on Plug. No wires. The harness connectors are on tight.

In reply to Ojala:

I guess I should bite the bullet and order that OBD-II scanner. Seems like it'd be useful here.

What exactly are "fuel trims"? Everything I learned when I played with a MegaSquirt a while ago has leaked out of my brain...

You're right, lots of things to go wrong. Very frustrated that I don't really have the time or space to deal with this right now.

If you have a laptop (I have an old one with Win98) and a vag-com cable you can download Rom Raider, Learning View, and all kinds of other utilities.

OBD readers can often tell you your long term and short term fuel trims. Learning View is customizable and can tell you in percentage how much fuel your ecu is either adding or subtracting at different flow ranges in addition to telling you a lot of other information. Rom raider can even show you your engine data in real time (well at a delay of a couple ms but close enough).

I have not read all the posts, but if this were a Mazda MPV I'd say its the coil. I've had something similar happen to me twice. Maybe worth a check.