Intermittent rapid rise in a/f learning

I have had this issue a few times with my car and am at a loss to explain or fix it. Car is a 5th gen LGT Stage I bone stock including newish subaru air filter and clean MAF. A/F trims A-C generally are +2 to +4 , D is usually +0.5 to +0.9. Only time they went negative is when the TMIC to TB coupler popped a leak, which has since been removed.

 

I drive the car for a while, park for an hr or 2, then on restart the A/F learning rapidly shoots up from 3 to 7ish. It stays there in both A and B trims with C and D less affected. The upper trims aren't going negative so its not a boost leak. AF sensor readings are also showing intermittently lean conditions. After driving for 40 minutes or so, the trims started to improve a little but still over 5% in A and B. Next day I start the car and within a few miles this returns to normal +3%. What could possibly cause this? An intermittent pre turbo leak? A fueling issue? The fact this has happened more than once is a bit strange. I have logs of this but as they are really long I'm just including the initial rise in a/f learning here.

 

Car is seemingly running OK otherwise, not eating oil, but has always had a problem with increasing timing advance , particularly in the mid loads. This was an issue with the Cobb OTS 93 map as well as an etune where knock was prominent in the 3-4.5k rpm range in 1.3-1.7 load until timing was dialed back.

I dont know if there is any relation but I thought I'd add that info anyway.

 

Any thoughts here?

datalog7 af rapid change.csv

Sounds like a minor vac leak to me that varies in severity depending on temperature... Just a thought

I thought the same thing, but have been unsuccessful in tracking it down. Since it never goes rich it would presumably have to be post MAF pre turbo.

 

Other thoughts I had were bad MAF, but the numbers look good elsewhere and it's relatively new (2011 car) or some sort of intermittent fueling issue, but I would have thought that would be more apparent during WOT, which seems to be OK.

Also I'm not sure if you got tuned before you cleaned the maf, but if so maybe your maf scaling is just a bit off. Not sure if that would explain +5% and above on idle fuel trims though...

It was cleaned before the tune (wasn't dirty but couldnt hurt) but since the intake and filter are bone stock the MAF scaling wasn't touched anyway.

I would say nothing is wrong.

 

In the many hours I've spent fine tuning my maf calibration I find that there are large positive corrections at low airflow levels when intake temps are extremely high.

 

There is an intake temp maf compensation table in the ecu. It is unused in the factory tune.

 

Keep in mind A and B trims cover from 0 to 10 grams per second of airflow. Datalog airflow and see exactly what type of driving conditions you are in at those airflow levels. A minor change in correction at idle and very low throttle driving isn't much to be concerned about. A very large, consistent change is something that would require some attention.

I would say nothing is wrong.

 

In the many hours I've spent fine tuning my maf calibration I find that there are large positive corrections at low airflow levels when intake temps are extremely high.

 

There is an intake temp maf compensation table in the ecu. It is unused in the factory tune.

 

Keep in mind A and B trims cover from 0 to 10 grams per second of airflow. Datalog airflow and see exactly what type of driving conditions you are in at those airflow levels. A minor change in correction at idle and very low throttle driving isn't much to be concerned about. A very large, consistent change is something that would require some attention.

 

I hear you and while there is definitely some degree of heatsoak of the MAF here, in the past that has caused levels to go up to+5,this is only recent that it is shooting past that, so I am trying to see what else it could be

 

A little more background here, this car, even with good gas cannot run a Cobb OTS 93 tune due to mid load mid rpm knock causing DAM drops. I had an etune which does work, but requires really really low timing in some areas. A smoke test was done a few months ago and didn't show any leaks, but I am still at a loss as to why I cant run normal timing without knocking, hence my search for mechanical issues.

 

As an aside, is it normal for the A/F Correction to shoot up to +25 if the brakes are pumped rapidly at idle? I have no idea if this is just normal behavior from the brake booster line or if it is a sign of a problem or leak with it.

I gotcha. Issues like that are frustrating to track down. How's the power, acceleration and drive ability otherwise?

I gotcha. Issues like that are frustrating to track down. How's the power, acceleration and drive ability otherwise?

 

Other than the goofy AF learning and the knock problems, the car drives fine. No problems at idle that I would expect with a vac leak, no taipipe smoking or stuttering.

 

I did a basic low pressure intake leak test by blocking off the inlet pipe at the airbox connection and then blowing into the BPV actuator vacuum line, and there were no audible leaks and it held pressure. However today even when not heatsoaked, my A/F learning A and B numbers went up to 7 and 7.8 respectively. While idling at the end of the drive, A dropped pretty quickly to 5.

 

Last pull I did a few days ago car hit full boost easily without and noticeable increase in wastegate duty cycle. Other than an intermittent pre-turbo leak which I cant seem to find, what other mechanical issues could cause this?

Has the warm idle gotten a little rough lately? If so, the misfires can cause the ecu to think its running lean and start making the AFR richer.

No, idle is totally smooth. No symptoms of a misfire at all. No trouble codes either.

 

One thing I noticed is my tuner has the a/f learning ranges changed from stock so the A range is idle but the B range goes up to about 40 gm/s, so I'm reading relatively lean through the whole cruise range with the B cruise trim being worse than the idle trim. I would have thought a pre turbo leak would be leaner at idle. The plot thickens.

Understood.

 

Reading back in your posts you mentioned something about the brake booster. I suppose you could run a test by pulling the vac hose to it and plugging the hose. I wouldn't drive it anywhere this way, but you could monitor the idle AFR anyway to see if the learning values go down (which would indicate a rare case of a leaking brake booster).

What I noticed is that if I keep pumping the brakes at idle a few times, the a/f correction value gets progressively higher all the way up to the max at 25, then quickly settles back to normal. This may be totally standard behavior from the manifold vac being tapped, but I have no idea if that is the case or not. I know that the vac line for this itself under the intercooler is attached and doesn't have any apparent leaks.

 

Sent from my DROID RAZR using Tapatalk 2

Just fired mine up, and pumped the brakes multiple times. So it DOES go lean when that happens then the A range starts climbing to add fuel. I only did this enough to see it climb a couple of points, but enough to think maybe this is "normal".

Excellent, thanks for checking that. I kind of figured that was normal but when grasping at straws gotta consider anything

Update here, I reseated the OEM paper air filter again since that was a source of problems before, reset the ECU, and saw normal A/F learning numbers again, so I thought that could be it. But on a long easy drive this weekend, the B range a/f learning (which my tuner seems to have set from approx 10-50g/sec MAF) suddenly went from 4.6 where it was for over an hour of steady driving to 7.03. No heasoak (75 deg ambient), no stop and go driving, it just randomly went up while cruising and stayed there for a few days. It even stayed high 6s to low 7s for a 2 hour drive home 2 days later. The A idle range stayed normal which doesn't suggest a vac leak.

 

My thoughts are intermittently bad MAF or Front o2, but these seem less likely in a sub 3 year old car with 40k miles, or some sort of fueling issue, be it injector or fuel pump. Since at WOT everything is fine I have a hard time thinking its the second issue either, so again I'm at a loss

af learning B range rise dl5.csv

Problem persists:

A/F Learning trims

A (0-10 gm/s) 1.8

B (10-50 gm/s) 7.03

C (50-90 gm/sec) 3.12

D (90+) 0.78

 

What fueling/mechanical issues could cause these trims?

Your A and B trims are going to really be the only relevant numbers. Your trim C will be an indication of how your car is running (a/f correction) beyond what the front o2 sensor can handle, realistically. Once you get beyond the 60 g/s range, you would need a wideband installed to read beyond that. Not sure why it's scaled to 90+, sort of irrelevant.

If it's not a post maf leak, your fueling map parameter could cause an issue, if it was scaled in an odd way.

it's nothing major at this point, but it would be nice if the trims were scaled differently such as:

A (0-10 g/s)

B (10-20g/s)

C (20-40g/s)

D (40+ g/s)

 

Talk to your tuner. see what he says.

I agree that the trim ranges are a little different than what I'm used to, however with the 5th gen o2 sensors post turbo location it should still remain relatively accurate at those ranges and has been shown to be so down to 11.5 afr or so. Regardless, I've had this tune for a few months now and never had this high of a positive trim until the past weeks, so something has changed.

OK, flashed back to COBB OTS 91 map just to get different A/F learning ranges and to r/o the tune as a source of the issue.

 

A/F learning ranges are Cobb map settings: A=0-10 B=10-50 C=50-80 D=80+

 

Drove around for a bit logging fueling parameters and noticed a few crazy things.

The B range A/F learning again worked its way up to +7 range, but the A/F correction variance is really all over the place. One distinct pattern that I noticed is that in cruising conditions in 3-5th gear, when the RPMs are below approx 3000 to 3200, the A/F correction is consistently pulling fuel, generally a mild amount, but occasionally having maxed out corrections to -25. This is not under boost so does not fit a boost leak profile. However, when the RPM gets over this into the mid 3000s, even under the same load, the A/F correction starts adding tons of fuel, even on top of the already lean looking A/F learning value of 7.

 

I tried logging this behavior a bunch on a long drive and was able to reproduce it pretty consistently, to the point where it almost looked like a bad MAF scaling, but this is with the stock intake and filter and stock injectors!!! The massive short term rich dips were more unpredictable. WOT pull still looks perfect, although it is a conservative map with 91 OTS and 93 octane gas.

 

Clearly I have some fueling issue, but these numbers don't fit the pattern for any specific issues I am aware of (ie, not lean under vac, rich under boost like a boost leak, not lean everywhere worse at lower flow like a post maf pre turbo leak) Any other issues this could be? I was thinking fuel pump or injectors but wouldnt that be worse under WOT when fuel demands were higher? Fuel pressure problems under low pump loads? I am at a loss. I would love some guidance as to what direction to look at here.

 

Logs attached below to illustrate some of the problem areas.

 

Rich Dip Cruising

 

Rich to Lean Transition

 

WOT pull then rich a/f correction

 

(the slight timing pull in the WOT log was learned correction from before)

That is some wacky stuff. Can you check your fuel pressure? Possible FPR issue? Possible issue with a fuel injector?

I was thinking a fuel pressure issue too, although I don't think it's the pump since a failing pump would probably have more problems at WOT and high rpm while I'm fine there. And im not getting any obvious misfire so i am less suspicious about injectors. wouldn't a bad injector causing a misfire cause a dramatic lean shift since unburnt air would make it to the front o2 sensor? Any advice on testing the FPR or injectors

fpr can be tested by connecting a fuel pressure gauge, even temporarily right by the fpr, where the injectors rail goes back to the fpr.

 

With the small vacuum reference line disconnected, it should be around 43-ish psi. the reference hose should affect fuel pressure proportionally - for example -10 psi of air at the ref line (like u would see idling) should reduce fuel pressure by 10 psi. and +10 psi of air on the ref line should increase the fuel pressure by that much.

Swapped the MAF with a known good one from an Outback 2011 and got the exact same numbers, so at least thats been ruled out as the problem. Now narrowed the potential problem list to fuel pressure (pump or regulator), Injector, Front o2 sensor, or intermittent pre turbo post MAF leak that I and a smoke test have been unable to find. Fuel pressure test up next. Any recommendations on where to T in? I was thinking on the supply line right after the damper.

 

Also noticing that I am still getting pretty significant knock at mid load mid RPM even with the Stock map flashed. Since some of this is in open loop, I'm less inclined to think that the O2 sensor is to blame.

the stock map knocks a surprising amount, from my personal experience and logging.