1996 JDM Twin Turbo Legacy GT - Help Needed - Poor Idle/Stalling

[bwolterm]

Hey everyone! 

So I am experiencing an issue that I am having a hard time diagnosing. My 1996 Legacy (Ej20h TT engine - MT) is stalling after about 20 minutes once the engine has warmed up. The idle is rough and the only way I can prevent the engine from dying is by giving a bit of gas. Otherwise I get the chug chug, chug chug and dead. At some points the engine almost sounds like a cammed V8 due to the rough idle and vibration is pretty aggressive.

Some background:

I fixed an oil leak coming from the oil pump a couple weeks ago. Prior to this, it ran very well. Once I fixed the leak, re-timed the engine and put everything back together, I cannot get it to run right. I opened it back up and redid the timing to be safe. I was extremely thorough with my measurements this time around, but no change. The vehicle runs and drives when cold without issue, but immediately takes a turn once warmed up. I do not have a CEL and am not getting any codes after plugging in the diagnostic wires under the steering column. What is the deal!?

Not long before all this, I replace the following: spark plugs, coil packs, knock sensor, coolant sensor, cam position sensor, crank position sensor, TPS, O2 sensor (passenger side/left side). I cleaned out the MAF, so my next step is to clean the throttle body and look for vacuum/intake leaks. 

Any help is appreciate.

Thanks for your time!

[m sprank]

You missed something when you repaired the oil leak.  Something related to temp. Sure you got the valve lash set proper?  Valve gap tightening when hot? 

[silverton]

If all he did was reseal the oil pump, the heads didn't need to come off. did you fix the oil leak with the engine installed in the car, or did you remove the engine? a jdm 96.... I wonder what kind of live data you can get from that... were those obd2 yet?

[bwolterm]

No OBD2 yet and I did the fix with the engine in the vehicle. I am confused by this by how non-invasive I was. Timing belt removal, then a new seal was applied to the pump. It's definitely odd. I am doing all O2 sensors this week to be sure, but I will update soon.

[silverton]

Are you positive the timing was done correctly?  Take the covers off and double check.

[m sprank]

Probably off one tooth.  

[silverton]

maybe 2.  when I got my third gen it was a bit gutless and sounded off, did a compression test and one side was 15-20psi lower than the other so I built a new motor for it assuming it needed a rebuild. well... when I pulled the motor and was swapping the timing covers, sure enough... off one whole tooth.

[m sprank]

Cant recall if I tried to drive off two teeth.  I did an "experiment" and you can deff drive (like poo) off one tooth.  3 teeth was no bueno.  Went all out and made sure there was no interference but basically had the intake cams backwards.  No start.  This was probably in 2003 or so, lol.

[doublechaz]

My 2nd gen jumped to 3 teeth off.  It would run, but it was unsafe to drive on the road for lack of power.  I estimate it was something like 25 HP.  Test drives during diagnosis were only in the middle of the night when there was no traffic on the side streets to worry about.

From what I've seen one tooth off can make really strange symptoms and leave you about 25% down on power.

 

The timing mark alignment isn't an 'I guess that looks ok from over here' kind of thing.  If you do it right and use your phone or an inspection mirror to view from the right angle it is either absolutely lined up, or it is oh crap missed by a mile wide tooth again.  It took me three tries once to get it right as things move when you pull the tensioner pin.  You have to put it together estimating where the tensioner will pull it to, and then check it after a couple rotations to settle it.  It takes at least five hands on a dual cam to hold everything, then you use your sixth and seventh hands to settle the slack part and pull the pin.

Edited March 22, 2023 by doublechaz

[bwolterm]

Hey everyone! Sorry for the delay. I've been tirelessly busting my knuckles on this thing.

I appreciate all of the input. I wanted to provide a quick update and see what you all think...

I re-timed the engine with a new belt and tensioner. It's essentially perfect in that area now. There was maybe a 1/4 tooth discrepancy previously previously. No change. 

To reiterate, the engine reaches operating temperature then shortly after lopes then stalls. 

I have additionally replaced the crank position sensor and cam position sensor. I removed the Intercooler to inspect for cracks and that looked fine. 

I am planning on trying to do an ecu reset so the vehicle can relearn fuel mapping. I dunno! This is the first non-OBDII jdm I've owned, so it's tough to diagnose combined with how obscure the model is. 

Do you all think it could be a leak in the intake, or possibly a vacuum leak? I am still not seeing any codes, so I am leaning towards it not being a MAF issue. 

At this point, the issue appears to be temperature related without a doubt, so I feel safe ruling out spark and fuel. Would you all agree? 

The subaru world is so odd, frustrating and fun all at once.

Thank you for the support!

-A Subaru Novice

[doublechaz]

That is tough that you can't get any live engine data.  I'm surprised it doesn't have at least some of OBDII as my 95 USDM has it even though it wasn't required until 96.  I don't suppose you can find anyone with the Subaru Select Monitor equipment and adapters for that market and year.

When I had the massive intake leak I don't believe it set any codes as the MAF was electrically OK, it was just reading implausibly low air flow since all the air was coming in through the leak.  Seems like the early ECUs only set electrical fault codes and didn't really do the reading out of range stuff like more modern ones.  But mine only ran for like 8 seconds at a time, so that is a bit different.

Am I right that it runs smooth and correct until it shuts down? 

One thought, if there was a strange problem with your water temp sensor for the ECU, not the one for the dash guage, then the ECU might get real mad.  I'm thinking where the sensor shows the temp climbing from cold up util it hits a non-continuous response and suddenly the ECU thinks it is -44 and dumps a crap ton of fuel in.  You would need to multi meter the sensor output to see that without the engine data.  You could also partly test this by finding out if there is raw fuel in the exhaust right after it dies.

Another thought would be to get an oldschool timing light on it and see if spark breaks up and goes away at the point when it dies.  I've seen an ignition coil puke but only after it gets hot.  Drive aroundd the block kind of thing.  An analog fuel pressure gauge would also be nice.  Could be that the fuel pump activation relay pukes after a certain amount of run time heats it up.  Or even the pump itself, although I've only seen that kind of failing pump give it a fairly long stumbling die out.

Once it dies after warmup, does it die quick on each restart with whatever is hot, or does it take minutes after each restart?

Another thing the early ECUs don't set codes for is losing sync.  I had to use a cheap oscilloscope to figure that out.  But there shouldn't be anything heat related with your crank and cam sensors, they aren't in a place to get real hot.  They are typically a work completely or not at all thing.

 

[bwolterm]

I took a video of the issue today.

Additionally, I'll be throwing a new coolant temp sensor in as soon as it arrives!

Thanks folks!

[doublechaz]

That sounds and shakes like it is running on one bank of cylinders.  Have you confirmed compression is within spec?

[bwolterm]

Sorry for the delay. I have not had a chance to do a compression test yet. I just re-adjusted the tcp and it does idle without stalling for the most part now (stalled once since the adjustment). Running some seafoam cleaner appears to have possibly been somewhat beneficial as well. I just wish I could figure out why the idle is still inconsistent/so low. It does not feel underpowered when driven. I will check back in as soon as I assess spark plugs and run a comp test.

Thanks again!