Positive LTFT at idle only when hot
 

First let me apologise for what will inevitably be a long post. I feel I should give all the details in the hope someone that likes to help does not have to berate me for a short and non descriptive cry for help!


I'm trying to chase a positive fuel trim at idle on my own car (231 S1), and am running out of things to check. I would be greatful for any suggestions.


On starting from cold and leaving to idle, as soon as the O2 sensor kicks in and closed loop begins the total of the trims equals bang on zero, up until the engine has run for a few minutes and ECT starts nudging 65 degrees C. From here, the trim starts building and I end up with around 3 to 4% positive trim total. Not bad, but I am used to seeing 1 to 2% on my engine.


If I leave the car idle longer however, as everything heats up the total trim goes higher, and after the fan has cycled once or twice it will hit up to +11%.
If I bring the revs up and let the trim stabilise after enrichment, anywhere above 1800RPM the trims are near zero, same when driving. The cut-off is very distinct.

Listening to the car, watching live data and inspecting the logs I found that it is not constant. Occasionally while the trim is high at idle, the revs will blip up to ~890 then settle, and the trim will come down for a while (as if something has gotten easier for the engine), then RPM will start dipping slowly as if struggling and the PCM will open the throttle momentarily and up the trim again.

If I stop the car and let it heat soak for a while, it will keep the positive trim on restart, unless it cools right down (overnight) and then it will be back to zero.
Something is causing the trim when it gets hot, but it is not directly related to ECT, more like something attached to the engine or in the bay soaking in the heat.

The engine:
I rebuilt the engine at 54K miles with a street port and D585 coils. It now has about 67K on it, and I compression tested it about two months ago with 7.1 across the board - very even results.

Fuel:
I recently fitted a low mileage R3 pump assembly, the high pressure filter is clean as a whistle and it looked to flow just fine on the bench. Having been through a few failed fuel pump issues before I know this one is good, and STFT when varying throttle while driving is nice and stable. I flow tested all my injectors after rebuild and they were very evenly matched and at the top end of the tolerance.

Air:
The MAF rate at idle seems happy at 5-5.2 until I get the high trim when it can drop to 4.9, but looking at the logs it seems the flow is being affected by the problem rather than causing it. Filter is clean, I have tried two different MAF sensors. I have had the UIM apart a few times, gassed all pipes/joints, drenched everything with spray lube and even blew the lot with smoke this weekend but can't find any leaks.
I did find the evap purge solenoid can be fun if there is pressure in the tank, but disconnected it and capped the intake port for testing.

I can see 229g/s at WOT on a cool day.


Spark:
I run the well used dwell table for the D585s. This weekend I caught one of the Magnecor leads arcing to a bolt on the coil bracket. None have any insulation damage, so I spiral wrapped them and rerouted to avoid any metal contact. No visible arcing now (but could they be internally damaged?). Tried different spark plugs including the brand new set I have on now. All connections seem snug and fully seated. The coils have some discolouration where the metal brackets contact the plastic casing.
The spark plugs are getting coked, and there is visible carbon on the rotors.


Exhaust:
Standard manifold (header?), free flow cat back, no leaks. The CAT is OEM and visibly fine, no rattles and running temp. is good. I have tried swapping the front O2 sensor.


As the plugs are getting coked I guess that the PCM is simply 'seeing' a lean condition under this RPM and adding fuel unnecessarily. Also as it only occurs under 1800RPM (doesn't the firing order swap somewhere around that point, ie. trailing first at idle?) and can cut in and out I am leaning towards ignition.
I am picking up a set of standard coils and leads tomorrow to attempt to rule out the D585s and Magnecors. If anyone has any further suggestions or observations regarding what I have done so far please let me know, as if the coils/leads aren't at fault I'm not sure what to try next.

Many thanks
Phil

Are you getting these numbers as an average from a idle data log or just from watching OBD data? When you rebuilt the engine did you replace any of the old sensors?

Both observing the live data and studying trends/averages in the logs. I do have some that I can upload tomorrow if it helps. I didn't replace any sensors during the rebuild, but have swapped out the MAF and front O2 sensor while troubleshooting this issue

Cheers
Phil

I should clarify that I had been used to a LTFT of no more than 2% since I rebuilt it. I am proud to say the engine is very strong (being my first rebuild I would have been happy with 'mediocre'), and without seeing this issue in the live data I would have been none the wiser as it runs splendidly.

Vacuum leak.


The engine is getting air at idle that the MAF doesn't see. 5-5.5 is normal for an engine to be consuming, so anything under that means air is getting to the engine that isn't passing the MAF. So the ECU is adding fuel to compensate. 4.9g/s (or lower) and positive fuel trim is classic vacuum leak.

Chances are the leak is in one of your hoses, and as the hose warms up and gets flexible it bends more under the vacuum in the line vs the higher ambient pressure, opening up the hole more and making the leak larger. Another possibility might be a hose that isn't secured properly, and it's gripping the fitting tighter when it's cold, loosening up when it's hot. Just a theory on the changing trims, but everything else fits a vacuum leak.

And vacuum leaks will always have the biggest impact in low RPM, because it's the greatest % change of the total airflow.


And i bet you have had this problem since the rebuilt, and it is just taking this long to get to LTFT. Pull up STFT and watch that at the same time, and i bet that as your LTFT comes up, your STFT goes down. The combined value is probably about the same starting as soon as the closed loop kicks in.

Clear the fuel trims and watch the STFT have all of the trim difference if you want to confirm for yourself. You will need to clear the trims anyway once you find the leak.

anything under 5% LTFT is not considered a BFD ...

edited: oops.

BFD?

Thanks for the input. It does have all the signs of a leak which is why I have investigated the intake so much. To re-cap, so far I have tried (with the engine hot and while the issue is present):

Disconnecting and capping off the evap
Disconnecting and capping the feed to the solenoids on the back of the UIM
Tested that the feed to the OMP injectors holds vacuum
Inspected the hose to the primary port jets
Thoroughly gassed/sprayed/wiggled the UIM manifold and its joints with the LIM

I haven't been able to find fault with any of the above, any further suggestions on what to check would be great. Has anyone had a leak past an injector seal?


Cheers
Phil

If you can't find the vacuum leak under any other method, take the car to a shop and have the intake system smoke tested. It should pinpoint the leak pretty quick.

You have to think about every single join point / seal / mating surface in the intake system. For example when i had a vacuum leak once, i ended up finding that it was from a tiny section of pinched gasket between the upper intake manifold rear rotor 3rd runner and the lower intake manifold, from when i re-installed the UIM without ensuring that the gaskets were seated correctly. My vacuum leak showed up in the data as 4.6g/s of airflow and +5% STFT/LTFT, and it never responded to the 'spray around the area and see if the idle changes' method.

@ team, please explain.

I ran about 5 with the old O2 (um, 130,000 miles) and now runs about 4 with the new one.

RIWWP - Yes I saw your posts about that when searching, and I had it in mind when I removed the UIM and inspected the gaskets - they all look perfect. I am finding it hard to locate somewhere that has a proper smoke machine - I attempted to do this at home by clamping a jar into the convoluted hose (to block the inlet), introducing smoke to the UIM and pressurising via the brake vacuum port. I couldn't get any to escape and the pressure would slowly and eventually leak past the rotors into the exhaust.

What sort of PSI would be applied with a proper setup, and would the exhaust be blocked as part of the process? I may try again with smoke pellets and a compressor charged to a low pressure.

One hypothetical question - if the PCM started dumping in fuel for whatever reason and the mixture was very rich, could this hinder air ingestion and cause a lower than normal MAF value, or does it behave very simply as a pump and the MAF always match the RPM (throttle allowing)?

Thanks

I've used a plug for the accordion tube and pressurized the intake to around 10-15psi. At that pressure you can hear most vacuum leaks. Don't expect it to hold pressure, it will flow out the exhaust and that's ok.

Yes, I use 15psi from my air compressor hooked up to a vacuum line and my turbo inlet capped off to test for leaks. You can definitely hear leaks.


Yes, the engine is essentially an air pump. The faster it spins the more air gets drawn through. More gets drawn through under load than when unloaded though.

At idle, the engine will pull approximately the same amount of air, regardless of the AFRs. If the MAF value is lower, that means air is being pulled from somewhere else or there is a MAF problem. If the MAF value is higher then either there is a MAF problem or something is placing the engine under load at idle, like a really thick oil would load up the engine more than a thin oil. You can spike the fuel at idle, but as long as the RPM remains stable the MAF value would be the same.

My thoughts are a bit scattered at the moment, but I think that answers your question.

For clarification the 5-5.5g/sec is at an 850rpm idle with a warmed up engine. If it's cold and idling faster then the air flow will be higher.

Correct, thanks for pointing that out.

The best solution, and what we did was make a smoke tester using a party type fog machine I got on Amazon for $30.00 or so. It identifies leaks quickly and easily.

I do have one of those but it didn't want to play last time I tried - the nozzle gets very hot and so it's hard to fit a hose to it, and it didn't like producing under any kind of pressure :dunno:

Thanks for the help so far guys it's really appreciated - I am happy that others have used a compressor so I will rig that up when I get home and see if I can find something. I will drop a smoke pellet in the jar that I use to seal the 'accordion' tube as well, might make things easier.

OK I tried applying 15 PSI but could not hear any leaks. So I ignited a smoke pellet (the kind used for testing draw on a flue/chimney), dropped it in a glass jar and stuffed this into the convoluted hose and sealed it. These pellets unleash a ton of smoke, but none came out. I then applied pressure again while the intake was full of smoke but could not get any to escape.

In frustration I removed the UIM (upper and lower) again and checked all the OMP injectors and SSV/VDI valves, pipework, UIM seals etc. but could not find anything. I did not remove the injectors as this would require more time, however they all feel snug.

I put it all back together and reset the trims (battery disconnected for a few minutes), and started it up. From cold-ish, it trimmed positively for a while as it settled down, and then hit around zero at idle (810-840 RPM) while flowing 5 - 5.2 g/s

I listened to the engine while watching the live data, and sure enough it began - the revs would stumble and drop (almost a cough) and the load would increase (I presume it opens the throttle slightly to recover), and it started applying positive trim. It would eventually reduce this back to zero, and then cough again and the process would repeat.

The intake noise did not vary between positive and zero trim.

If anyone would be so kind as to take a look, I have attached the log from this short idle session - taken with Torque Pro (I had to zip it, if this is not acceptable let me know and I will convert to text). If you scroll down to row 2027 where the revs have settled you can see the trim is around zero. As you scroll further down you will see the revs drop at row 2134 (this is where it stumbled/coughed) and the AFR starts going lean so the trim is increased. Later on it then recovers for a while but the process begins again at 2459, ad nauseum.
Toward the end, the issue has grown in magnitude and the trims cease returning to zero.

I thank you for your intake leak suggestions and welcome any further ideas on the matter, but can't ignore a suspicion that the ignition is breaking down. Unfortunately I didn't collect the test coils today as I knew I would need as much time to have at the intake, but will pick them up tomorrow and try swapping them.

Cheers
Phil

If you repeat the cycle a few times, does it sort itself out?

I ask because that is how mine behaves after are fuel trim reset. the first cycle gets it idling pretty good, the second one, after the engine cools does. does better. After that life is pretty good. I colored this as normal. then again, I have a 100,000 miles on this engine with about 8,000 of them on the track.

No, the issue disappears when the car is fully cold but reappears as it warms and so it always rides a positive LTFT in the lowest MAF range. After a run the LTFT at idle will be anywhere around 4-7% positive, but will climb again up to 10/11 if left to idle for some time.

sorry, it was supposed to say "not a BFD"

BFD is slang for "big f***ing deal" i.e. American sarcasm

under 5% can be normal ...


.

Ah I see.

Update time - it's not the ignition. The hunt for 'intake suction' resumes - tonight I noticed if I stand well back from the car I can hear a muted hiss. Up close I can hear it from within the UIM if I use a funnel and hose pressed against the intake, but can't locate it anywhere outside of it.

It's not the brake servo supply, rear vac tank or solenoids, vfad supply (capped off) or the three breathers to the convoluted hose. I might pick up a second UIM and omp injectors next week and swap them over. I could also remove the fuel injectors and check the seals at the same time, and check the LIM bolts are torqued. Short of that I am all out of ideas - any further suggestions gratefully received :)

Phil

I was fighting the exact same issue recently. Low MAF, in the 4.5 g/s range, but high positive fuel trims. STFT would start pegged at 25% and LTFT would gradually increase as high as 11%. Engine speed at idle would rev up and then stumble as AFR gauge would climb up to 16-17 then plunge to 12-13. It turned out to be vacuum leaks. Yea, leaks plural. I found this cap that I have on my jet air intake split open (yes Team, I know I need to order silicon caps :lol:). The other leak I chased for quite a while. I finally took off the UIM and reseated all the seals. Put it all back together and g/s is now in the mid 5's, LTFT is 0% and STFT are under 10% once the car is warmed up. It's cold here, in the 30's, so I my current STFT is likely IAT related. But MAF is looking much better, increased about 1 g/s average back into the normal range for my idle RPM.

Yep, I went crazy finding a leak once on an RX-8 and found this. The way it was turned it was facing downwards so you could not see it. I have no clue how it could have happened.

https://farm9.staticflickr.com/8062/...ed136475_c.jpg

I'm in the same boat. I thought it was a bad AFM but after replacing it my LTFT's are still creeping up. CEL just came back on. I'll be trying the 15psi pressure test this weekend. Thanks for the tip.

9krpmrx8: Where is that plug from? Brake assist vacuum line?

Yes that is the check valve at the brake booster.