RomRaider

I've been fighting this 2005 FXT for awhile now. It's a fairly basic setup that I'll try to summarize here:

Type-RA block with 5k miles.
B25 heads rebuilt maybe 15k miles ago.
VF48 & STI TMIC
2011 STI fuel pump, 2007 FXT injectors, -6AN fuel lines with IAG fuel rails and Aeromotive FPR. 2007 FXS injector connectors spliced in.
TGV + evap fully deleted.
OEM exhaust manifold, catless aftermarket uppipe & downpipe, OEM catback.

It's my father-in-law's (FIL) car. He has complained about surging/bucking, worst in 2nd and 3rd gear, mid-2000 RPMs, moderate throttle like 25%. It's been doing this ever since the top-feed conversion I believe.

I borrowed the car from FIL recently to do the fuel system upgrade and VF48 swap. It runs great except for the persisting stumble. I logged data and see that there is a big dip briefly in the timing when it stumbles. Since it's a manual 5MT, I think the timing dip causes the bucking/surging to reverb so to speak, especially in 2nd gear.

I have my own 2004 and 2007 FXT with various degrees of modifications, so I swapped MAP sensors just in case it was spiking in voltage and throwing the base timing to the right-most column. This nor anything else I've tried has helped.

There is no FKC or FLKC occurring when the timing gets hit with a large reduction. I do have the load minimum set to like 0.95 for feedback and fine knock controls. If I have it lower than that, it tends to randomly detect knock and the corrections don't help it, nor does big reductions to the timing table. My other Subarus have been like this too, where you have to find the floor of the knock control and it's not always the same, but usually is around 0.8 or 0.9 load.

Here are some examples. I've seen it with higher loads like when lugging it a bit more, but these should suffice:





I'm running out of ideas. My next plan is to swap coils, but I don't think that will help.

Thanks for any help!

for me its hard to process a screenshot of a log - much easier to just zip the log/tune(assuming its your own work)/LTV info and then play the log against the tune file to see how it looks.

from what I can see there it seems like you're also pretty rich considering the RPM is still quite low and your very much in vacuum - but really hard to say much else without seeing the proper info.

Sorry, I don't know exactly what you mean as far as "play the log against the tune".
AFR was at or near stoic whenever it happens since it's in closed loop.

Basically I can feel when it happens and see it in the logs where very briefly it will be 15 to 25 degrees retarded. It's almost like the knock sensor still does something even when it's not in range of FKC, FLKC and roughness.

I recently tried different coils and a speed-density tune with no change.
I think I'll add a ground wire since it only has the pair that go to the intake manifold. Wiring or ECU seem like the only thing that COULD be wrong at this point.
I might try an older version of the ROM but highly doubt that will help and would prefer to use the latest.

Attached the tune, need to log some more on this revision but am sure it will show random timing drops too.

I started logging MAP instead of boost and have seen it drop like an electrical issue with the sensor:



I replaced the sensor a couple days ago with the one from my '04, so I'm confident the sensor is good. I know the wiring to the sensor is stressed at all times due to the weird orientation of the connector on top of the throttle-body. Maybe it's not a good connection. Maybe I can fix that. I'll go forth with adding at least 1 ground wire like from the bottom of the heads to the chassis.

what i mean here is if you zip and upload your LOG in its CSV format it's easier to look at what's going on vs you posting a screenshot of what a few lines and values of your log where showing at the time you happen to have open on the screen.

looking at your tune tho its a bit weird, why aren't you using the knock correction advance table? with it zero'd out i'd guess your base timing is a little on the low side on the higher load / higher RPM regions.

your MAP signal could be dropping as a result of the timing change as well don't forget, if the car basically goes full retard and reduces its torque output significantly you'll see the MAP reading drop as a result.

Thanks for the feedback.

Yeah, I was thinking the MAP drop could be from the brief drop in timing. It seems like such a quick dip in timing that the MAP couldn't possibly drop to a value that's even lower than when engine-braking for less than 100ms.

Oh yes, I zeroed the KCA table at one point when I was getting desperate and forgot about it. I will put it back to stock values from a 2004 STI JDM rom that I took some per-injector comps from. My IAM starts at 1 and hasn't dropped at all so the KCA table wasn't getting used anyway. If I understand it correctly, with KCA zeroed out, the IAM would never return to 1 if it dropped.

On a positive not, I believe I have mostly fixed the stumble. I need more testing but it seems a lot better, like 90% less stumbling. What I did was: zeroed the AVCS in the stumble zone. Since it felt better like that, I then put AVCS at more static numbers instead of smooth wide-varying values. It seemed like as AVCS altered quickly during light acceleration, that was making it stumble. If AVCS sits at, say 15 degrees, it doesn't stumble or barely at all. The stock values are like that anyway, where it's 15 or 20 across the board, not smoothed interpolated type of transitions.

My other cars that I've tuned didn't behave this way, but this '05 has been the bane of my existence so I am not surprised that it still fights me.

Another issue I'm working out is that it feels like 1 or 2 cylinders would misfire at very light loads at like 2600 rpm in 2nd and 3rd gear, like the smallest throttle input possible when slowing down or maintaining speed getting ready to accelerate (IPW @ 2.0ms). This is a zone the car had misfire counts at a long time ago when I had the 2005 per-injector compensations in place, which add a lot of fuel 10 to 16% in the low load RPM zones that misfires would count up. I reduced those values by like 67% and it helped, but I had zeroed them out after reading about those tables and stuff. Now, I put the values from the 2004 STI JDM rom and need to test it more but seemed good on the 30 minute test drive I've done so far.

This freaking car is such a pain in the arse man...I have like 200 tune revisions over 3 years and it's not even my car. Meanwhile, the last 4 Subarus of my own only took like 8 or 15 tunes to really hone them in, and they never had the stumbling, surging crap that this one has.

It's pretty easy to avoid the stumble if you just shift at like 4500 rpm when accelerating moderately, but my father-in-law will never do that it seems. I have lectured him several times about lugging it but he's stubborn. Plus you can feel it stumble with like 24% throttle at 2200rpm which isn't really lugging it, so I want to smooth that out. If the bearings fail on this type-ra block, I'm sure will be from the constant lugging. The last log he sent me showed 80% throttle at 2400rpm in 5th gear, probably going up a hill...

Looked up some info about KCA and see that I'm wrong. After restoring my KCA table to that of the '04 STI, I noticed my car trying to hit 14° advance at WOT where the base timing advance table says to hit 10° and it quickly learned -3.85°. I'll change some values to allow KCA to find the best timing now that I realized that I was wrong.

Cheers!

You should really use mega logviewerHD

have you figured out a resolution to this? On my 05 FXT I recently did IAG TGV delete housings and top feed conversion and then got a etune from Migtuned to get it running. Ever since I have had misfires on all cylinders and some stumbling in that same RPM range and on very light throttle like maintaining speed. Tuner swore it was mechanical but theres only a couple things that were touched that could have any effect and ive ruled them out. I firmly believe its tune/fueling related but I cant figure it out. Its been almost 2 years now and I havnt driven the car much because of it.

you might be better to start your own thread with some logs of the issue including what things you believe you've ruled out as possible causes and how you ruled them out. I've lost track the amount of times a car 100000% triple checked mega super sure has no boost leaks that turns out to have a boost leak once its actually tested for example.