I have been having some strange performance issues with my 08 Subaru Legacy 5mt with the N/A EJ25 engine.
It's generally down on power (no jokes, please :-) and the power comes and goes i.e. while accellerating, if I hold constant throttle sometimes it pulls harder, then weaker, then harder again. It happens at WOT as well. I did an ECU reset, swapped out the air filter which was surprisingly nasty. That improved things slightly but it's still not right. No CEL issues/stored codes. Front O2 sensor is maybe 3 m/o and OEM. I sometimes will notice that the issue seems to be triggered by lifting to shift. It'll be pulling OK in 1st, I lift as I go into 2nd and when I begin feeding throttle in it's unhappy.
What do you folks make of these plugs? Cyl 3 and 2 seem to be running leaner than the rest. Unfortunately, I don't have a way of reading the fuel trims for the individual cylinders. (If someone can recommend a good value priced scanner that would let me look at some data that would be awesome.) It's always had a bit of a rough idle and with the color range on the plugs I'm wondering if I have some funky injectors?)
So I'm dealing with a very similar issue in my 07 Legacy 5mt 2.5i, I can feel the car pulling timing and with it goes the power. It wasn't as noticeable driving the car on my own (and I don't drive it that often anymore) but it was more obvious with the car loaded up for the weekend going to the cottage. I also have no codes which points me to the knock the sensor either picking up knock or thinking it is knocking even if it isn't. I've got a new knock sensor on the way just to cross that off the list as they are known to fail.
For software I don't think you can beat Romraider along with the included Romraider logger which is what you use to monitor the engine and check for codes. You'll need an OBDII to USB cable, a generic VAG-Com cable will do the trick and the software is free. Spend some time reading up on it on Nosiac etc. but it is pretty straightforward. With it I'm able to see when the car is detecting knock, ignition timing it is adding when there is no knock etc, it's a lot of information and you can datalog with it. If you spend the money on a tactrix openport 2.0 cable then you can really dig in and start to play with things.
I'll let you know what I find when I swap in the knock sensor and if it solves my problem.
That's great info, thx! I did wonder if it was pulling timing. My low tech plan (for diagnosis only) was to replace the knock sensor with a fakeout resistor, add some octane booster to the tank (to protect the engine by reducing the chance of knock at full advance) and see if the problem goes away. It's smarter to monitor things like you suggested and just see if the ECU is detecting knock.
How's the exhaust system on your Legacy? Mine's got some decent size holes in it after the cat. I was wondering if the lack of back pressure could be reducing EGR flow, leading to knock and the ECU pulling timing. I'd think it would throw an EGR flow code if it was too bad, so who knows. Wild ass guess. It did seem that the low power problem started around the same time as the exhaust system got worse.
My exhaust is in okay condition without any leaks. It still has the original system except for the part at the rear where it splits to the two mufflers which are also original. I've replaced one of the expansion joints ahead of the cat twice now along with some other repairs at the joints. The car was throwing a PO420 code for an inefficient catalytic converter but I disabled that using the fancy tactrix cable in the ECU so that problem was "solved" lol. Your theory might be right, maybe you can do a E36 M3ty muffle wrap type repair and see if that helps and justifies the cost to do it right?
These cars do run better with with higher grade gas so if you aren't running that try a tank of premium and see if it improves as well.
You can't make much out of plugs that have been run awhile. You need to put fresh plugs in, make a hard hit and kill the engine before decel. Pull then post pictures.
About all I can say is they look a hair sooty and old and put new ones in 
gap looks big but worth checking. If there is a cheap, copper plug you can use to diagnose, get a couple of boxes set the gap to 0.028-030" and do some pulls.
So I swapped out the knock sensor tonight after it landed on my doorstep. Couple of tips, the sensor is located down on the block on the drivers side of the throttle body just in front of the clutch slave if you haven't already found it, it's a 12 mm bolt and you should be able to reach it with a 10"ish extension, BE VERY CAREFUL DISCONNECTING THE PLUG! I pulled one of the wires out the back so that turned a 1/2 hour job into a 1.5 hour job figuring out how the terminals are held in the connector and re-soldering the wires in, don't be me! lol
So what do I think? I'm not sure to be honest. The car feels better pulling to redline and I don't feel the timing coming in and out as I modulate the throttle like I was before. Overall it feels a lot better...but conditions are also quite different. We were in a heat wave last week that ended today around noon with thunderstorms and showers so the temps have dropped almost 10 C. The car also just had me in it, not packed to the gunnels with 4 people, a baby, two bikes on the roof and the A/C on full blast, that all makes a big difference. I plugged in and the car is still picking up some knocks but I'm not sure that is unusual. The car also gave pretty decent fuel economy on our trip so I don't really think there is a major problem.
Tomorrow I'll be in the car for 4 plus hours so I'll update again.
Thanks for the update, glad to hear you're seeing some improvement, it'll be interesting to see what you find after more driving. I was thinking about how the knock sensor, being a single wire device, is referenced to engine ground. The knock sensor is essentially a piezoelectric microphone and outputs a fairly weak signal, so any noise picked up by the knock sensor wiring has the potential to be misinterpreted by the computer as knock. Given that ground is the return circuit path for that sensor....
My car had a clutch installed by some hacks who cut and spliced a ground wire that (on my car) runs from the trans to a bolt near the upper pitch stopper/dogbone mount on the firewall behind the plastic intake plenum. I decided to peel off the electrical tape covered splice in the middle of the wire and found two oxidized ends of wire just twisted together. They couldn't even be bothered to install a crimp connector, much less solder the thing. I did a soldered/heat shrink splice, cleaned up the ring terminals on the ends. Not sure if this is going to do anything, it's a bit of a long shot, but Subaru put a ground wire there for a reason and it's one possible cause crossed off the list. We'll see.
There was a lot of discussion with similar symptoms on Australian forums for the 2.0R Legacy (EJ204), the eventual conclusions seemed to be the knock sensor, and that the curves were set with too little tolerance. In some cases as big/little ends got worn, it was enough to start tripping the knock the sensor to pull timing. In other cases the knock sensor itself having tiny fractures. Some folk unbolted the sensor and zip tied it out of the way where it couldn't move, and never reported any more problems. Others got a chap in Russia to crack the ECU and reprogram the knock curves..
The knock sensors on these are two wire so they are not relying on the engine to complete the circuit. It wouldn't surprise me if they are a bit on the sensitive side of things (and the stock tune on the crappy side of things). I don't think that is my main issue but something I might play with in the future (I have the cable and definitions to change that stuff in the ECU).
On to the update, during my drive today which had lots of highway and traffic it was obvious things weren't any better. The car generally felt a bit down on power again and surging slightly. My car finally threw a code on the way home - PO171 Too Lean Bank 1 (there's only one O2 sensor so it's not specific to bank 1 in our case). Doing some research and this code is usually triggered by vacuum leaks or air filter issues throwing the MAF off and therefore the amount of fuel it thinks it needs or exhaust leaks ahead of the O2 sensors causing a false lean condition. I went for a drive with the windows down and the computer plugged in and I could see it doing all sorts of corrections on the fuel side of thing, with full throttle pulls when it is operating open loop (so no corrections happening) below 2000 rpm or so I could see it holding 14.7 and going a bit rich but as rpm climbed up beyond 3000 the AFR were going into the 15-16's so lean and staying there. The expansion joint that I have replaced twice now is failing again, I think at low rpm exhaust is flowing out but as engine speed and exhaust speed increase it is pulling air in giving me a false lean condition.
If you have a basic OBDII reader you can plug in and read the AFR, when my car had a stock tune at full throttle it would hold 14.7 up to about 4000 rpm and then start to slowly richen it up to redline. Do some full throttle pulls and see if that is what you are seeing or if you're also seeing it go lean.
I'll hopefully get that expansions joint replaced this weekend before I'm gone for the week and see how the car does after that.
Final update for me. Fixing the exhaust did not solve the issue, the expansion joint wasn't great but replacing it had little to no affect in my case (a worse leak could be a good place to start). I was still seeing roughly ok AFR's at idle and lower RPM and then going lean above 2000 RPM or so (and very lean under full throttle, open loop).
I had cleaned the MAF with no change and spent some time looking for vacuum leaks. Everything looked good there and the fact the car was a hair rich at low RPM to me didn't point to a vacuum leak as I'd expect it to be lean at idle and get better as vacuum went down and engine load up which was the opposite for me. I decided to try a new MAF sensor and that did the trick, AFR's were pretty much where I expected them to be and full throttle power is back. The engine is a bit rough at the moment but it needs to adjust out what it learned with the bad MAF so I expect that will improve as the car is driven some more.
Hopefully you figure out what is going on with yours, post an update if you do!
Did not read, just saw the pic lol.
Those plugs look fine, even though they are not all the same shade, they burn nice.
The gap looks huge though. Measure that even though some come pregapped.
They do look large, but believe it or not, the gap on the old plugs is just a hair out of spec. The plugs in the photo have about 30K on them. I now have fresh, properly gapped plugs installed with no difference.
I ordered exhaust system parts. Someone welded the rear cat to the rear resonator/ Y pipe assembly which is rusted through at the Y, so I'm going to replace both of those, at least my car will sound better, and it's _possible_it will help the performance issue. If not, further troubleshooting will be needed. Mine might also have a bad MAF. It still isn't setting any codes though...
An exhaust leak will definitely affect performance.
It's solved. I replaced the rear Y pipe and rear cat due to some small holes. As step one for replacing these parts, I removed both mufflers from the Y pipe, and when I started the car to drive it off the ramps, and noticed that the car was unusually quiet for a car with no mufflers. I then installed a new rear cat and resonator /y-pipe assembly. (Someone had welded the flexible ball and socket type joint between the rear cat and the resonator together, so I changed them both.) After starting the car with the new cat and resonator/y pipe assy (but still no mufflers) it was much louder- about what you would expect from a car with no mufflers. I then installed the new mufflers and everything sounds normal. The car is definitely up on power compared to before, particularly on the low end, and all driveability issues are gone. In fact, I think it's got more power than it ever had during my 4 yrs of ownership.
I'm thinking there must have been an obstruction in either the rear cat or the resonator. I cut the pipe at the outlet of the old cat and the core of the cat doesn't appear melted or plugged from the rear view but I still want to cut the pipe in front of the cat so I can inspect there.
Puzzling things- 1.) the car has never had a misfire or rich mixture code in the time I've owned it. It's not a big oil burner either. 2.) You'd think that in any case if there was an incident in the car's past which lead to a cat-melt, the front cat would have been the one to be damaged. 3.) the biggest driveability issues were happening at low RPM, maybe 1/2 to 3/4 throttle. I'd think a restricted exhaust would lead to the car falling on its face at WOT, high RPM. Who knows? Maybe back pressure was causing excessive EGR flow?
In reply to flat4_5spd :
Maybe there was a mouse in your cat?
Glad you found your issue even if it is still puzzling.
To follow up on my similar issues the MAF sensor solved my problem for a while but I still had an exhaust leak somewhere at the front of the car even though all the joints were leak free. It would be quiet for around a minute after startup and then start, that should've been my clue to look for cracks. Anyways, the leak eventually got bad enough and triggered another lean code as the fuel trims were way out. The car drove fine at this point as it was getting lots of fuel. Anyways, pulled the front header/cat again and finally found a good sized crack that opened up so I tig'd that up and everything is nice and quiet again.
