RomRaider

Hi RR community,

I'm new to RR and I've recently been using it to diagnose a very strange misfire issue with my subie. I've searched tons of forums including this one for someone who has encountered the same issue and I've come up empty, so I decided to create this post.

Car: 05 Subaru Baja Turbo AT 79K miles, bone stock (EJ255 motor).

The issue started when pushing it near WOT around 5,500 RPM, I got a flashing and solid CEL. The codes were P0302 (cylinder 2 misfire), P0304 (cylinder 4 misfire), P0171 (lean on bank 1).

Reset the ECU and tried another pull, same CEL codes.
Went under the hood, found some loose and bad hoses including one leading to the intercooler. Fixed those leaks and gave it another shot. This time, same misfires, but no lean code.

At this point, I knew the motor ran fine at idle and it ran well until getting into boost where it would hesitate and misfire at high RPM & high load.

I then tried the following to diagnose the issue, but still got the same misfire codes on cyl 2 & 4:

- Swapped 1 & 3 coil packs with 2 & 4, replaced spark plugs with NGK platinums (local parts store only had these in stock), and checked compression - got 120 PSI across all 4 cylinders on a rented O'Rielly Auto Parts gauge.
- Swapped 1 & 3 injectors with 2 & 4
- Replaced the front O2 sensor with OEM Denso
- Boost leak tested (although the shop decided to smoke test via the oil dip stick hole?? )
- They only found a leak at the PCV valve breather (which I believe would not create a boost leak unless the PCV valve is bad, but it is functioning)
- Threw in a bottle of fuel system cleaner and full tank of Shell 91 gas
- Checked the timing belt to make sure it isn't loose - it is on tight
- Cleaned the MAF
- Checked grounds to the motor with a multi meter, all grounds read good
- Fixed an exhaust leak after the front O2 sensor, but before the rear O2 sensor

Here is where it gets weird -

After none of these efforts yielded anything close to a fix (except for eliminating the P0171), I decided to set myself up with RR. I started logging parameters and found that the motor is actually misfiring in lower revs.

Here is what I found while the car was running in park:

- At idle - no misfires, but very high AF learning #1 (maxed out at 15% or close to it)
- At around 2,500 RPM - multiple misfires on cyl 1 and one or two misfires on cly 3
- At 5,500 RPM - multiple misfires on cyl 2 and 4

So, it is misfiring only on bank 1 at low RPM and misfiring only on the other bank at high RPM. To this point, I have not seen it misfire on all 4 at the same time...

Does anyone know what could cause this condition? My only hunches are the OCV angles not following each other or maybe timing being a little bit off? I tried logging OCV angles and the two sides stay pretty close to each other with 1-2 degrees variance every once in a while.

Could this still be a vacuum/boost leak? My logs show high AF Learning #1 at idle, but AF Correction #1 goes down significantly when I rev the motor.
My Manifold Relative Pressure at idle reads in the -9 PSI range which I believe would signify good vacuum. Is this a valid assumption?

Should I get a leak down test to rule out burnt/tight valves?

At this point any help pointing me in the right direction is welcome.

My various logs are attached. I can always log other parameters under other conditions if suggested.

Thank you!

Update: I logged the TGV sensors today. At warm idle the left side reads 3.4 volts and the right side reads 0.4 volts.

According to the FSM, both sides should read between 3.8 - 4.9v fully closed and 0.2 - 0.9v fully opened. If that is true, then it looks like the left side could be stuck almost fully closed.

I've been searching for TGV logs for the last hour and can't find any posts. Would anyone be so kind as to log their TGV sensors to confirm my voltage numbers are no good?

Thanks for the link. I ran the same exact procedure and the right TGV dropped from 4.2 to 0.4, but the left stayed pinned at 3.4 the whole time.

Pulled the sensor and used a screw diver to turn it and watched voltage move within range on the laptop so it's good.

Pulled the motor and gave it 12 volts and watched the gear spin so it's good.

Tried to turn the TGV by hand and it was very hard to move. The motor side turns but the nub on the sensor side never budged. Does this mean the valve is broken? Going to look at getting a replacement. Don't really want to do a delete but I'm starting to see why people do them!

Update: After finally getting the stuck TGV replaced, the misfires seem to be gone. The old TGV seemed to be mechanically broken. Not sure why this did not throw any TGV code, but glad I checked those values on RR