2008 Spec.B makes noticeably more power after battery disconnect.

[m sprank]

Yes, light weight flywheel.

knock sensor is under the intercooler.  Position and torque are very important, they are finicky.

I personally have had false knock in my own LGT from bad front endlinks.  Knock sensor is just a microphone.  If it picks up a noise it registers it. 

[SilentJ20]

Update:

Leak/smoke test successfully completed and showed nothing.  There is a chance there may have been something so small it was barely emitting anything, but something that small likely isn't creating the drastic changes in the ECU.

Unable to get to the knock sensor yet, but I'll see what I can find.

I still have the appt to get it into a shop on Tuesday.

[SilentJ20]

One interesting find, when I removed the intercooler, the hose clamp on the outlet side (non-turbo side) wasn't even tight.  I mean, it wasn't loose and flopping there, but it had no real load when I loosened it.  Looks like there has been a bit of oil mist coating the engine under there.  Not a lot, but if it's been loose for a while it could gradually seep out and coat things (I've read some oil in the intercooler and such lines is normal.  I have no puddles, just coating).  But, that being the pressurized side, air would be escaping after the MAF at boost.  Meaning the regular amount of fuel would make it run rich.

It must have sealed enough for the pressure of the smoke test to not not leak out there.  OR, that was a small leak I didn't see.  I know engine operating pressures are much different than the little smoke pump can put out.  Could that be my issue?

The knock sensor seemed tight, oriented correctly--or pretty close, I didn't bust out the protractor.

Edited February 24, 2023 by SilentJ20

[blackobxt]

What psi did you smoke test at?

[SilentJ20]

3 minutes ago, blackobxt said:

What psi did you smoke test at?

The kit I borrowed from a buddy is rated at a minimum of 7 psi.  However, as it has no gauge built in (lower end model), I am unsure of the actual pressure reached.  There is no aux way to add pressure in this setup, and at this time I do not have the necessary fittings to test beyond this tool.

[m sprank]

Throttle body tube being loose happens all the time.  OEM had/has a "condom" around the end of the hose that likes to malform.  The loose hose causes a lean condition at idle and a rich condition on boost.  At idle engine is operating at a vacuum so it inhales unmetered air through the leak.  On boost the hose is pressurized and leaks out so you are losing metered air. 

[SilentJ20]

50 minutes ago, m sprank said:

Throttle body tube being loose happens all the time.  OEM had/has a "condom" around the end of the hose that likes to malform.  The loose hose causes a lean condition at idle and a rich condition on boost.  At idle engine is operating at a vacuum so it inhales unmetered air through the leak.  On boost the hose is pressurized and leaks out so you are losing metered air. 

So if this hose were leaking, it would create the opposite condition that my ECU is trying to correct. Damn. Thought I might have been on to something.

[m sprank]

Could be a myriad and one of the issues. Like a ladder, step on every rung.

[SilentJ20]

Update:

The local Subaru experts at PIA identified the root of the problem.  Apparently, despite my COBB intake requiring the installation of a tune (per installation instructions), I am still on the stock tune.  This throws off the MAF/injector scaling significantly.  The ECU can just barely compensate, so no CEL (even though I feel there should be one).  They have a history and a good reputation and have seen it all.  I'm inclined to believe their expertise, haha.  Solutions are either install an OEM intake or get it tuned.  PIA is a COBB authorized shop and they only tune using the Accessport--which I do not have nor do I want to purchase at this time.  Nor do I want to tuner-shop and do logs and all that in the short term.  For the sake of simplicity and speed, I decided on the OEM intake.  It will arrive this weekend.

All other sensors tested fine, the MAF was permanently-dirty (despite my numerous cleaning attempts) so I have ordered a OEM Denso replacement.

The plan is to possibly do a tune later, opensource style, once I learn about and enjoy the car in stock form.  I understand the stock tune is a bit on the lean side and not ideal for long-term.  But I also know what I have going on now is worse.

NOW, once I install the intake, reset the ecu, install the new MAF and get a new LearningView, we'll have a better idea if I made the right choice, haha.  To be continued...

Oh, and PIA didn't charge me for the diagnostics.  So assuming they diagnosed it correctly, I'll recommend them.

[m sprank]

The "unknown tune" scenario has been popping up lately.  Unfortunately it goes with the age of the cars at this point. Passing hands without the needed info.  It was bad form to blame the tune in the past.  

The stock tune is not that bad. Subaru warrantied the cars with the tune.  Owners were able to drive over 200k miles on the stock tunes. No tune is perfect.  OEM engineers have FAR more to consider than aftermarket tuners.  

[SilentJ20]

Update:

I think the major issue has been resolved.  Thanks to some quick ebay shipping I got my new-to-me OEM intake in a few days, installed it, reset the ECU and then went for a fairly thorough drive.  Here's the new LearningView:

IAM is back to 1.0, all knock-correction reads zero.  I've only went for a 30 min drive and tried to cover the basic driving conditions, but this seems to be a good sign.  I'll check it again after a week of commuting.

Only question I have now is the minor adjustments needed to fuel at idle and WOT.  Is that couple percent anything to be concerned about?  What I've learned so far tells me no.  Just nice to have the power back and a happy motor.

Thanks for all the wisdom imparted on this journey.  I'm sure it won't be the last.

[Infosecdad]

No, that’s a great LV. If you’re within 5% you’re generally considered good. 

[m sprank]

That LV is CLEAN.  

[KZJonny]

That’s wild dude. My car also came with a Cobb ‘CAI’. It idled like garbage, and the old owner said he ‘couldn’t figure it out’.

Thankfully he had kept the OE intake. Slapped that on, and like your results the thing immediately behaved better. I hate to think what the fuel correction would have been on my drive home with it!!

Glad you got that all sorted. I wonder what the kid I sold that thing to is doing with it. Probably driving it untuned, like the catless DP he had installed, and hadn’t tuned for….

Edited March 6, 2023 by KZJonny

[Max Capacity]

So many years, we have been telling people not to get a CAI for these cars. But like Mike said, cars passing hands to a new group of owners. These cars came with a great CAI.

[SilentJ20]

I've always heard with my Saab, too, is that with a turbo it's not that big a deal. Light restriction isn't a big deal for a turbo.

[SilentJ20]

Update:

IAM is still at 1.0 and I have no learned knock correction, so that's awesome.  Downside is the fuel correction for the lower airflow ranges is creeping back up over 10%.  Also, I've noticed some significant hesitation/stuttering/pulsing of the power in certain situations.  If I do a 3rd gear pull and just plant the throttle at 2500rpm and hold it, it's smooth power delivery.  But if I run through the gears with some good throttle into it (over 50%), getting back on the gas after shifting into 2nd or 3rd I really get some strange uneven power delivery for a second or two.  No codes.  Sounds like the 2800-3000 rpm stumble stuff I've been reading about, but it's not limited to that particular rpm area.

Captured a log to see some of the data.  I was concerned that maybe the pedal or throttle body position wasn't being read right, but it seems to be tracking.  One thing that caught my eye was how the fuel correction is jumping all over after a shift.  If any of you gets a chance to check the attached file and have any insight I'd appreciate it.

excel log file:

light blue is areas of full throttle acceleration (except first gear, that was more like 75% and varies a bit)

darker blue in the box shows jumping A/F correction in an area of otherwise steady throttle, steady airflow.

romraiderlog_20230314_104452_stumbling.xlsx

Edited March 21, 2023 by SilentJ20

[m sprank]

I see a small bounce (MAF/MRP/IDC/Timing). I also see your clutch on the verge of slipping.  Have an LV?

[SilentJ20]

Alright, now that the forum is working again, here's the LV:

Two items of note.  I've been running a new MAF for a couple weeks now, and I replaced the AF sensor (upstream O2 sensor) a few days ago, but have not reset the ecu since the O2 sensor.  IAM and Knock Learning are still perfect--which I'm happy for after the intake issues.

Why do you say that about my clutch?  I don't doubt you, it's still the original I assume, so I know it's time is limited at the 135k miles I have.  I just want to know what you saw that made you notice.

[m sprank]

Rather than send you on a snipe hunt, first reset the ECM.  Any time you change sensors like that, reset the ecm.  You are adding fuel in a/b/c and then removing fuel in d.  So you are lean at less than full load/throttle and rich at full load/throttle (75-100%). 

This could be open/loop closed loop transition related.  Meaning the stumble you feel is the ecm attempting to correct fueling when it can as the O2 sensor defaults during closed loop.  But if the ecm is reacting to false data or un-metered air (causing false data) then we need to find the source.  Start fresh and then pull an LV again after some driving.  The ecm will need to see boost to set the IAM back to 1.0.

As for the slip, you can see the rpm stall and then jump under WOT and boost.  The stall will get worse or you will catch the rpm fall and rise again as the clutch grabs, if it grabs. 

[SilentJ20]

4 minutes ago, m sprank said:

Rather than send you on a snipe hunt, first reset the ECM.  Any time you change sensors like that, reset the ecm.

 

Ok, will do.  I know the ECM will eventually re-learn, but I don't know to what degree or how long it would take.  If this actually fixes the issue, but I didn't notice because I didn't reset the ECM, I'll sure feel silly...