I've had this problem for a while - car is running very rich on cold start, enough to make it splutter and stall until warmed up.

It's not the supercharger as i've run the engine with it detached with the same issue, and it's not the fueling in the tune, as it will startup better for a day or two after removing fuelling, then it will relearn this problem.

Basically at idle when hot, the fuel trim is adding about 10%+ fuel. It seems like a vacuum leak, but I couldn't find anything by spraying carb cleaner.

Is there anything that would make a vacuum leak / lean condition occur only as it warms up? EGR issue?

See the datalogs below:

Cold start, the fuel trim begins negative, then as it warms up, eventually settles trying add a whole lot of fuel...
http://i157.photobucket.com/albums/t80/jowet/EvoScan_Graph_coldstarup_zps18e65a23.png

Warm start:
http://i157.photobucket.com/albums/t80/jowet/EvoScan_Graph_warmstart_zpseee0c9ff.png

Driving (seems like a vacuum leak as trims go down as revs go up):
http://i157.photobucket.com/albums/t80/jowet/EvoScan_Graph_driving_zps55469128.png

What are your long term fuel trims (low, medium and high) like? The short term trim can bounce around a lot during normal driving, but the rolling average of the long term trims can give a better guide to certain problems. If all 3 long term trims are above about 6-7%, the oxygen sensor could be starting to die (they don't always die suddenly) or you might have a problem with substandard fuel delivery in some (but not all or even most) injectors.

As far as I can tell, the ECU uses its saved long term fuel trims when in open loop mode which would be why you might see excess cold start fuelling with a low long term fuel trim (the request 0C in Evoscan, which is records mostly at idle or just above) of ~+10%. This also fits your description of the problem disappearing after a battery disconnection, as the long term fuel trims get reset and take several averaging cycles to re-establish after the ECU enters closed loop.

You might also log the oxygen sensor voltage - at idle in closed loop it should exhibit a response somewhat like a sine wave with a fairly low frequency.

In terms of a vacuum leak, were you able to get the carby cleaner in around the injectors?

Long term fuel trims currently are:
Low +4.6%
Mid +4.6%
High -0.1%

I'll post up the oxygen sensor log, it did look like a consistent sine wave from memory.
I sprayed carb cleaner around the front injectors, and rear sides, but couldn't really get to the rear middle one.

Would you be able to post a cold start log of fuel trims etc. if possible? That way I could see what they're meant to look like. It's strange how the engine runs leaner and leaner as it gets warmer.
Good idea about fuel injectors, I hadn't thought of them, I'll check connections etc.

If it's a vacuum leak, it's almost as if the aluminium is expanding as it becomes hot and opens a leak somewhere...
Thanks for the reply

Hmmm... assuming they're reasonably stable at those values the long term fuel trims are within a range (+/-5%) that would be considered acceptable in many cars, though the discrepancy between the low and mid trims on one hand and the high trim on the other hand is of interest.

The low and mid trims being nearly +5% while the high trim is 0 (near enough) suggests that either your injectors have a bit more latency than the ECU is expecting (IOW they're slower to open) or that your suspicion of a vacuum leak has some merit. However I would have expected a 5% misfueling should be barely noticeable in terms of engine starting or normal running (there are many examples of people not really being aware of >10% over-fuelling from a dead O2 sensor). Some engines are more fussy than others though.

I don't have any logs showing all the fuel trims handy unfortunately, as I've never bothered logging the STFT, but I'll see if I can get one next weekend. When I first logged my car, which was fairly stock at the time as best I recall, all 3 long term trims were sitting on +5% running 91 RON ULP (no E10).

Your comment about heating causing movement in metal parts is in fact completely normal, however gaskets are supposed to fill in the gaps unless that movement is significant. This can happen as a result of poor heat treatment during manufacture (such as the problems with the alloy heads in EA Falcons which usually took some time to manfest themselves), though this doesn't seem to have been a problem for MMAL through the 3rd gen years... Hairline fractures or pinholes can however also show temperature sensitive behaviour :(

Some times the only way to find such leaks is to poke and prod at things - I remember an example from my youth, where a sporadic miss was finally traced (after several years of effort) to a brake booster vacuum hose when it was bumped by accident - and the engine nearly died...

The latency issue is interesting - I tried changing this a while ago but it didn't seem to help - the oxygen sensor trim would go down to 0% (when increasing or decreasing latency - I can't remember which), then it would slowly work it's way back up to 10%. I'm assuming it's correct as it's set as per the Raptor moristech original tune... I'll have another go to see if it helps.

Next step will be to start hitting things in the engine bay with a hammer to see what happens!!:happy:

The latency issue is interesting - I tried changing this a while ago but it didn't seem to help - the oxygen sensor trim would go down to 0% (when increasing or decreasing latency - I can't remember which), then it would slowly work it's way back up to 10%. I'm assuming it's correct as it's set as per the Raptor moristech original tune... I'll have another go to see if it helps.
Be aware that even though we have scalings in ECUFlash definition files for things like the injectory latency, they're all inferred or deduced from reverse engineering. Given the nature of production sensors etc and the assumptions made in production car ECU code, especially in ECUs that use integer arithmetic exclusively (which the Magna and similar vintage Mitsubishi ECUs do), the scalings have to be considered approximate. So having a good reference in an after market ECU doesn't mean that exactly the same values will produce exactly the same results in the factory ECU. After-market ECU code tends to pay more attention to fine details and precision, because the customers are paying for (and expecting) it...

The low long term fuel trim is the one to monitor when adjusting latencies - as you increase the latency it should come down. Ignore what happens with the short term trim. The procedure for dialing in injectors, both size scaling (which I doubt is a problem with yours) and latency, is spelt out in Merlin's Evo Tuning Guide which shouldn't take much effort to find if you're not familiar with it.

Thinking of other ways that your symptoms might manifest makes me wonder if one sticky injector might also lead to this sort of behaviour - lagging on open at low revs and tending not to completely close at higher revs perhaps?


Next step will be to start hitting things in the engine bay with a hammer to see what happens!!:happy:
Your intake runners play like a xylophone? :eek2:

More seriously... I'd hang off using the hammer until you've tried moving flexible stuff around ;)

I've now got a log including the STFT from a cold-ish start (coolant temp <20C), but I can't conveniently post it or an Evoscan graph - if you contact me by email (see profile) I can send you more detail. In this log the STFT starts at 0% but begins changing once the coolant temp rises above 36C. It then stays within +/-2%, changing mostly in sync with the oxygen sensor changes. At the moment my low LTFT is sitting on -5% and didn't change at all over the 12 min 45 sec logging period (all at idle) so it is quite stable.

That sounds good I'll send you a PM.

Your STFT sounds like what it should be.. When I first installed the larger injectors, the engine was missing severly at idle, pulled the spark plugs and one was white - so assumed a dud injector... took them in for testing, they all tested fine on the flow bench, so re-installed and the problem magically disappeared --- I wonder if it's that same injector going whacky?

If it is one injector playing up, that cylinder will still be running lean but all the other cylinders will be running a bit richer so that the oxygen sensor balances it out. Whether that cylinder will be running lean enough to show on a spark plug I don't know.

I finally got around to boost leak testing the car and I'm glad I did - I found major issues!
I'd recommend anyone to do this even if they're not supercharged, to fix all sorts of driveability issues.
This is after trying carby cleaner/propane to find vacuum leaks without success.

The setup:
- homemade plug with pressure guage and bike tyre stem (you can also buy these made up).
- 24L air compressor from bunnings.
- Spray bottle with soapy water (you can also hear the air squealing out of the leaks).

http://i157.photobucket.com/albums/t80/jowet/boostleaksetup_zpsef295377.jpg


Leaks in order of severity:

1. Intercooler.
This is silly, the unused bolt hole that I assumed was a threaded insert, actually is a hole into the intercooler air plenum. It's been like this since supercharging - years :nuts: gauge wouldn't even hold 2psi. Fixed with a bolt and o-ring.

http://i157.photobucket.com/albums/t80/jowet/01_zps55dee57d.jpg

2. S/C outlet
Timmed piping to sit properly in silicon piping, fixed. Pressure builds to 5psi then stops after fixing these two things.

http://i157.photobucket.com/albums/t80/jowet/03_zps336e3cd9.jpg

3. MAF
Huge leak here with air blowing out of the seals of the top part of MAF. Don't know if boost caused this or what... sealed around edges as a temporary fix, waiting for it to dry. Probably need a new MAF.

http://i157.photobucket.com/albums/t80/jowet/02_zps30d72f97.jpg

4. PCV valve
Air sqealing out of PCV valve (removed from rocker cover while still attached to manifold). Cleaned but still won't hold pressure, need a new one.

5. Lower inlet manifold
Drivers side, losing a lot of air here. Needs new gasket etc.

http://i157.photobucket.com/albums/t80/jowet/04_zpsb3a0b678.jpg

6. Throttle Body
Leaking out idle set screw.. Does this have a seal or o-ring?
See second photo for another smaller leak - does anyone know what part this is?

http://i157.photobucket.com/albums/t80/jowet/05_zpse97340e9.jpg

http://i157.photobucket.com/albums/t80/jowet/06_zpse543b888.jpg

Also leaking at base of throttle body piping - the TB metal lip has four cutouts around it, which doesn't make sense because a good seal can't be achieved.

Hopefully the post-MAF leaks are the cause of the idle problems.

I can't believe how much boost was being lost!!! Looking forward to driving it tomorrow with just the MAF and intercooler fixed :)

Glad to hear that you got to the bottom of this, even though it was more like opening a can of worms....:eek2: and glad that it looks like your injectors aren't part of the problem.

I would think that the MAF leak is more likely to only be a serious problem under boost due the design of the MAF casing - under vacuum that cover/panel would be pulled in tighter against the gasket improving the seal, whereas boost would have the opposite effect. For this reason I've always been a little surprised at people putting the MAF downstream of a boost source (SC or turbo) as I've not heard of any of the Mitsubishi MAFs being run in that manner in factory cars.

Yes looks like an issue with these MAF's, I didn't realise the boost could get up into that upper casing part. I might look into a blow through type MAF with translater to replace this one if possible, or make a clamp to clamp the top down.... or may have to go back to draw through.
I like the blow through as the temp sensor will adjust timing etc. to any temperature changes affected by the intercooler and water meth.

An update:

Replaced all gaskets/o-rings etc. that were leaking, made a boost-proof MAF by removing and re-sealing the top cover with a plastic epoxy. Engine runs much smoother having fixed these leaks... however, the cold start stalling/ hesitation was still happening!

Last fuel top up I added a bottle of wynns fuel system cleaner to see what would happen - within 5km, the engine totally smoothed out, idle trims went back to normal, idle vibration mostly gone, no hesitation, and much more power.... So something is gumming up in the fuel system, next week going to change fuel filter, check pump, and injectors :) very happy to have isolated the problem!

Good to hear!