Oxygen sensor- could be getting poor feedback from the sensor and the ECU is adding fuel

Poor spray pattern on injectors can cause a rich condition. If it's not atomizing the fuel correctly, you'll get an incomplete burn, the o2 will not be consumed and the o2 sensor will see that as a lean condition and add fuel.

Fuel filter/pump- loss off fuel pressure can can cause poor atomization and low fuel delivery. ECU will try to increase injector output but can really only increase the flow rate of the injector, not the flow of fuel.

Dirty IACV could be an issue with idle. I usually spray throttle body cleaner through the inlet before the throttle plate while manually keeping the car running. Make sure there is no debrs that can can be sucked in while doing this with the air box off.

Vacuum leak on intake can cause excess fuel to be added and make it difficult for the IAC to modulate airflow at idle.

I would think injectors last. They are not that hard to remove and soak the lower body In cleaner. Then blow spray through them. They can be tested for movement/function with a 9v battery and a couple pieces of wire. I'm not recommending that but, I have done it out in the desert before

Those are just some theories and possible explanations

 

Looks like ETC and I are on the same page


Interesting issues you've got going on here.
I wouldn't put the injectors as the first on the list to replace, pretty rare for them to fail.

The question is, are the misfires causing the negative fuel trims or are the negative fuel trims causing the misfires? I suspect that latter.

What is the MAP reading at idle when the trims are -25%?
Vacuum leaks can cause negative fuel trims on a MAP sensor engine.
Stuck open purge valve sucking in fuel vapors could cause negative trims.

All that being said I think the most likely culprit is the front A/F sensor. I recommend to take a good look at it's operation. What kind of readings are you getting from it when the trims are -25%? What happens when you induce a lean and rich condition?

I own the same car and had a failing A/F sensor cause +25% ST/LT trims with the only code being a P0171 Lean bank 1, I don't see why it couldn't fail in reverse and cause -25% trims. Unplug it, see what happens.

Thanks for the quick responses.

ECT- I'm seeing only negative trims, meaning the ecu sees too much fuel and is pulling it. I don't think a lack of fuel pressure is going to cause this issue,

cdc- I also think the fuel trim is causing the misfire.

The A/F sensor was actually replaced recently with a direct replacement Bosch unit, this problem existed before and after this replacement. I doubt the sensor is bad, the readings are consistent and much more normal during driving. Readings with a bad O2 sensor in my previous subaru were erratic and made the poor thing undrivable (no code either!). As fpr this case, this one was replaced for a p1130 code, which did not return after the replacement.

I can check for vacuum leaks and clean the IACV. I'll also pull the purge solenoid and check it over, make sure it is not stuck. I did these already but it doesn't hurt to do it again. I was going to move the injectors around and see if the code moved with it, I have not seen a misfire on cyl 3 yet. Might as well do all of this while I'm there. Most people have told me that even a MAP engine would lean out on a vacuum leak, but I can see it either way. It's worth a good once over again, this stuff can be a big pain to track down.

I'll take a look at what the map reading is. I don't remember offhand what the number was. Any other readings to watch?

EDIT: MAP reading on hot idle with no accessories was around 6.4 psi, though cold idle was 8.6-9psi. Misfire was happening during both but no codes until I sat in the driveway for a minute or two and played with the a/c to see what difference it made. With A/C on, the reading peaks at 7psi when the compressor engages and levels out at 6.6psi.

 

Looks like ETC and I are on the same page


Interesting issues you've got going on here.
I wouldn't put the injectors as the first on the list to replace, pretty rare for them to fail.

The question is, are the misfires causing the negative fuel trims or are the negative fuel trims causing the misfires? I suspect that latter.

What is the MAP reading at idle when the trims are -25%?
Vacuum leaks can cause negative fuel trims on a MAP sensor engine.
Stuck open purge valve sucking in fuel vapors could cause negative trims.

All that being said I think the most likely culprit is the front A/F sensor. I recommend to take a good look at it's operation. What kind of readings are you getting from it when the trims are -25%? What happens when you induce a lean and rich condition?

I own the same car and had a failing A/F sensor cause +25% ST/LT trims with the only code being a P0171 Lean bank 1, I don't see why it couldn't fail in reverse and cause -25% trims. Unplug it, see what happens.

 

Sorry, was half asleep when I read that. I saw " rich fuel trims" as adding fuel.

Too much fuel should either be stuck open injector or a malfunctioning fuel pressure regulator. an extreme reduction in airflow could do it but, the car should barely run at idle much less at hard accel.

With then motor running ( not hot so you don't burn yourself) put your finger on each injector and see if they feel like they are clicking the same. I've had them get stuck open before ( the desert thing, sand rail) and have similar poor running situations.

I'm going with inj 3 is stuck wide open and the ECU is pulling fuel and causing misfires on the other cylinders. If so, it either won't be clicking or it will feel "muffled".

Let us know what you find.

 

No problem, semantics with fuel trims I have always been somewhat tricky because you can consider it both ways (e.g. the positive numbers could mean rich conditions as opposed to the addition of fuel to the map).

Hopefully I'll be able to take a look tomorrow. I found some helpful fuel trim information on NASIOC, most of which I was aware of. But I found an easy test to compare injector flow rates.

If you suspect a problem with the fuel injectors, use STFT or LTFT to check proper operation. Fig. 5 is a graph of LTFT from a 2000 Saturn with a cylinder misfire. The baseline is -10% LTFT. So the PCM is decreasing injector pulse width to compensate for a slightly rich condition.

Performing an injector balance test is simple, as long as you can gain access to the fuel injector connectors. Unplug one injector at a time and wait until the maximum LTFT change is reached. On some vehicles you'll use STFT for this test, or a combination of both STFT and LTFT. With one fuel injector unplugged, the oxygen sensor will see a lean condition and the PCM will compensate by,increasing the pulse width of the functioning injectors. to reach stoichiometry. The results of this particular test, with injector No. 1 unplugged the LTFT change is approximately + 14%, injector 2 +10%, injector 3 +17% and injector 4 +16%. Injectors 3 and 4 contribute a greater volume of fuel than injectors 1 and 2. We know this because the amount of fuel trim increase is greater with these injectors unplugged. Injector 2 is the cause for concern; with injector 2 unplugged, the remaining injectors need to supply only +10% total. This injector may have a slight leak that's causing the negative fuel trims. A new set of injectors fixed this vehicle. Unfortunately, I was unable to test the old injectors.

found here: Great article on Fuel Trim. - NASIOC

I think I can do that along with checking for vac leaks, checking for injector pulse feel, cleaning IACV and if nothing else swap some injectors around. I was going to start out with taking the #3 injector and putting it on cylinder #4. Cylinder #4 is the most common one to show a misfire code so I think that'd be the most telling if the CEL moved. I can check the air filter again but it wasn't terribly dirty a few months ago. As for fuel pressure, I'd have to rent a fuel pressure gauge setup to make sure but I'm fairly confident it's ok, I'd imagine a failing FPR would cause issues during many driving conditions.

Thanks again for the input, I'll keep this updated. Hopefully I'll have good news soon!

 

If your fuel pressure regulator has a leaking diaphragm the fuel will enter the engine through the vacuum line and cause negative fuel trims, more so at idle than when at cruise. Pull the vacuum line and see if it is wet inside or if any fuel drips out of the regulator.

Yes, that is a good fuel injector balance test thanks for reminding me about it. I could have used it a couple times at work lately but forgot all about it.

One thing that seems a bit concerning is your MAP sensor readings. Your engine vacuum cold is about 12inHg, that is very low. Once it warms up it is 16-17inHg which is better but this is still a bit lower than I like to see, here at sea level anyway. Might be worth checking your timing marks?

I also suggest you take a look at the rear O2 and make sure it responds to forced rich and lean conditions.

Cleaning the IACV shouldn't effect fuel trims, whether it's open, closed, dirty, or clean, everything it does is metered airflow into the engine, unless the gasket was causing a vacuum leak it's only symptom of being dirty should be poor idle quality.

 

Well, the vacuum can drop significantly when the engine is running poorly though I might have some issues with compression. I have noticable piston slap so I suspect my compression rings may be compromised. I have another engine from a wrecked forester so I can check it over and swap that in (I rebuilt that one a few years ago, as long as the wreck didn't break the heads, it's in good shape).

It doesn't hurt to check the timing marks, though I may wait to pull the crank pulley until I check these other things. I also need to put my mechanical vacuum gauge on this to verify, I suspect my scangauge is lagging. I've had problems with this in the past.

I also suspect the idle roughness and stalling is due to the mixture being incorrect and not the IACV not reacting properly to it's signal.

Out of curiosity, why would I be looking at the rear O2 sensor? It's readings would be skewed by the catalytic converter, which (assuming the catalyst is up to operating temperature and functioning) should be able to keep up, even with the potential raw fuel from the misfires. I'll have to see if I can get the readings from the rear O2 sensor on the scangauge, I don't have that one programmed.

 

The reason for looking at them is just to verify that they are working. It is widely believed that the rear O2 sensor is only for catalytic converter monitoring but that is not entirely true. I have seen cases where the rear O2 is bad and it skews the trims. This does vary by manufacturer and I haven't seen it on a subaru but it is worth taking a look at the rear O2 scan data to be sure it is not misbehaving.

If you are driving down the road and give the engine full throttle both the A/F sensor and rear O2 should go to their full rich positions, when you let you foot off the gas and decelerate from say 3000-4000 rpm down to idle the ECM will turn off the injectors and and both sensors should go to the full lean position.

 

I have done a few things today.

I put a stethoscope on the injectors and they all sound like they're running properly. I ran the injector balance test but my scangauge or my ecu is not able to provide accurate consistent readings (it'll stay blank or the last reading will just sit there). My father's scangauge had the same problem, and it is a newer firmware edition. So no results there. I need a better gauge I suspect.

I swapped injectors on #4 and #3 and the CEL did not move. So #3 isn't stuck open or anything like that. Still seeing the miss, still getting codes.

I ran a compression test and saw 170psi +/- 3psi on all 4 cylinders. Looks pretty solid. Didn't pull the timing cover but that suggests timing is close. I'll get to pulling that probably this weekend. Plugs all are still gapped and show normal color. There's no oil in the plug tubes.

Checked over for some vacuum leaks. Nothing that I could find but I've noticed there's very little crankcase vacuum. The PCV valve is functioning properly but I suspect there's a significant leak in the breather system. I can take the oil cap off and the idle does not change at all, nor can I feel much of any vacuum in the system. A small vac line off doesn't change it either. I could not find my mechanical vac gauge but I'll be finding one and getting manifold pressure readings. But if it is as low as the MAP sensor says, combined with the good compression readings- I think that may indicate leakage. I suspect I'll need to buy some new hoses.

cdc- To my knowledge, the subaru rear O2 sensors do not function as a tuning sensor. It does not make sense that the rear O2 sensor readings would be used for tuning. The catalyst is supposed to remove the vast majority of fuel out of the exhaust stream, which would use up the O2 in the exhaust stream, skewing any useful data. Some manufacturers could use the post catalyst sensor data to monitor not just catalyst health but also the temperature (if there's excessive heat/fuel in the catalyst), which could in turn trigger the car to pull fuel to try and reduce that temperature. Just a thought. And again, I'll try to find out how to grab the rear data with the scangauge.

 

Did you ever check for fuel leaking inside your fuel pressure regulator?

 

Yes, I did take a look at that. Vacuum line was dry, pressure was ~40 psi.

 

Update:

Problem has not been solved yet. I have done more work and replaced all the gaskets on the intake manifold and all the vacuum lines. This included the injector seals, throttle body gasket, both manifold gaskets on the heads, and IACV gasket. IACV is clean and new.

I have pulled the timing belt covers off and looked very closely at the timing belt. The timing is spot on.

Knock sensor was replaced.

While replacing the injector seals, I decided to rotate all of them around the engine this time (1 to 4, 4 to 1, 2 to 3, 3 to 2) hoping the problem would move. No change.

I also put a manual vacuum gauge on the manifold and verified that the MAP readings appeared accurate. I am at a complete loss still. This was only at hot idle however, so I am not totally satisfied.

The only theory I have right now is that the MAP sensor is bad but I have no way to verify this. I don't know if I should buy the 70$ sensor or not, as I would be very pissed if it didn't solve the problem. Too much has been spent to fix this issue already.

I still watch the STFTs and LTFTs and it appears that I run abnormally rich even on the highway, albeit to less a degree than when at idle. I usually see about -10% combined correction on the highway and about -30% at idle.

Apologies for leaving this thread to die, I was in the process of moving and getting college and work in order, this had to take a back burner for a while.

 

Go read through this thread


http://www.subaruoutback.org/forums...2002-outback-ltd-wants-stall-out-after-high-rev-banging-up-steep-hill-help.html


Can you post a video using the same phone based scan tool (torque app)?

 

Go read through this thread


http://www.subaruoutback.org/forums...002-outback-ltd-wants-stall-out-after-high-rev-banging-up-steep-hill-help.html\


Can you post a video using the same phone based scan tool (torque app)?

I read the thread. Seems like a different issue than mine as I have a 2.5l, 5mt vs the 3.0l, 4eat. Many different things to consider between our two cars. Then there's the issue itself. My issue only really occurs at idle and dying at stops.

I can post a video of the scangauge readings but as far as something like was discussed in the thread, I do not have that sort of capability at the moment. I was going to order a VAG-COM and work with FreeSSM (already downloaded) to see if the logs will be better suited to weeding out the issue but I need my next paycheck first.

The other thing I can record is the exhaust sound; the misfiring is very apparent at the tailpipe. When I get a little extra money, I may grab a bluetooth adapter and record a video if the problem still exists.

 

Yeah, I wasn't suggesting to compare the problem so much as the way we diagnosed it using the scan data. The video of him driving wasn't even important, just the data stream was. So yes, a log of the freeSSM would work similarly. Any way that you can record multiple PIDs over time and they can be analyzed.


Listening to the exhaust sound won't help diagnose why it is misfiring.




In the mean time, when you are at idle and the fuel trims are reading -30%, what does your rear O2 sensor voltage read?

 

Unfortunately, my scangauge only reads O2 sensor data in %. I had to search around and it seems that it is a percentage of oxygen in some way shape or form. It reads 0% or close when engine braking (lean extreme) and reads 90-99% when idling and WOT (in rich conditions).

I also was able to finally get my A/F ratio function to work and I can see that any time I go into open loop, the ratios go really rich. Above idle, ratios in open loop were 12-13:1 but at idle it dipped almost to 10:1. I suspect that is what is choking the engine. The ecu is able to bring it back to ~14.5:1 but with the trims at -30% that is probably about the limit of its ability.

Side question: I know that at WOT the ecu goes into open loop, but I didn't think it was supposed to deliberately dump excess fuel to the tune of 12.5-13.0:1, is this normal? If not, it may be another clue to the problem.

I will be getting a cable ordered probably this week for FreeSSM, hopefully if what I am finding out with new gauges doesn't solve it, FreeSSM should.

It seems this problem is more of at an individual cylinder or cylinders rather than the whole system (e.g MAP, TPS, Knock, O2s). And the misfire codes I get do agree, as I have not once seen a cylinder 3 misfire, 4 and 2 are very common and 1 happens occasionally. I suspect cylinder 3 is over fueled, the question is how/why.

I've moved the injectors with no change, twice. Cylinder 3 is nearest to the fuel pressure regulator however there has been no leakage through the vacuum lines. If the ecu operates how my 97 OBS did, the injectors are powered all the time and the ECU grounds each to fire them. I've seen it happen before that someone's harness was shorted and it caused the injectors to leak, but the fuel would get into the cylinder and into the oil fairly quick. I have not seen nor smelled fuel in my oil, I've changed it twice now at ~6000k mile intervals. I'd think such a problem would be very noticeable with both engine operation throughout the rev range and in fuel economy (I see mid 20s on the highway).

 

The O2 % reading you are seeing sounds like it is reading in Lambda, it behaves similar to fuel trim readings.


Wide open throttle is fuel enrichment mode and it is normal to see 12:1. At idle it should not dip down to 10:1.


Just because the fuel trims are saying it is making a correction for a rich running engine does not mean that the engine is actually running rich. This is why I keep asking about the rear O2 sensor data, you can use that sensor to test the system. When you see the fuel trims at -30% at idle look at the rear O2 voltage and if it is at .8 or .9 volts then suspect a rich running engine, but if it is at 0 volts then you will know the engine is not actually running rich and suspect that the front A/F sensor is lying.


Another thing you can do is when you see the high trims and rough running at idle, add an alternative fuel source (propane or intake/brake parts cleaner in the air box) and see if the engine smooths out or stalls out.


And yet another thing you can do is just unplug the front A/F sensor and see if it starts to run better. Until you can upload a data stream over time I'm just guessing, but my money is on a failed front A/F sensor as that is the most common reason for this problem.

 

I have done the fuel test (starter fluid) and it has a stalling effect. Interestingly, the idle speed raises and smooths out upon opening a vacuum line. And the bigger the line, the better it runs. The brake booster line (which often causes properly running engines to die) makes it run almost perfectly and the engine dies upon resealing the vacuum leak. This seems like a rich condition to me.

If the rear o2 readings were showing lambda, they would be getting larger as things went leaner. I don't think that's what I'm seeing here.

I did unplug the front O2 sensor (a while back, looks like I forgot to mention, whoops) and open loop running made everything that much worse. I am confident the front O2 sensor is functioning properly.

My other concern is that if the cause is a system-wide sensor failure, why do I see distinct problems on only one side of the engine most of the time? Timing is bang on, so misfires should theoretically at least show on all cylinders if all are suffering from the same bad mixture.

I guess we'll know more when I get some logs. Maybe I'll bump the idle up 100 rpm or so and use that as a band-aid until I can uncover the culprit, raising the revs smooths everything out at least.

Any other tests I could do while I'm waiting on diagnostic equipment?

 

I have done the fuel test (starter fluid) and it has a stalling effect. Interestingly, the idle speed raises and smooths out upon opening a vacuum line. And the bigger the line, the better it runs. The brake booster line (which often causes properly running engines to die) makes it run almost perfectly and the engine dies upon resealing the vacuum leak. This seems like a rich condition to me.

On a properly running speed density system it is normal for the RPM's to raise when a vacuum line is pulled and for the engine to stall when resealing the vacuum line. The vacuum leak raises intake manifold pressure which the MAP sensor sees and adjusts fuel accordingly which is why the RPM's raise, just like opening the throttle. Often creating a large vacuum leak like pulling the booster hose will cause an idle surge. That said your engine is smoothing out with the large vacuum leak and might suggest a rich condition.


Does raising the RPM's by opening the throttle smooth out the engine the same as pulling a vacuum line?

Any other tests I could do while I'm waiting on diagnostic equipment?

Are there any exhaust leaks near the O2 sensors?


Do you have a vacuum gauge? I would be curious to know the intake vacuum and see if the needle is bouncing back and forth.


Have you adjusted the valves ever? (How many miles are on the engine?)


How did you check the timing?
I have read twice in the past week of Subarus driver side cam sprocket malfunctions. The outer ring of the driver side cam sprocket is a stamped piece pressed on to the inner portion which is bolted to the cam. The outer stamped piece which has the timing mark on it can rotate on the inner piece causing the mark to not be in the right spot.


Have you done a compression test comparing the left and right sides of the engine might prove or disprove a timing issue.

 

The engine does smooth out significantly when raising the rpms, the same as removing the vacuum lines. I also had the cruise cable too tight, which started pulling on the throttle, which essentially fixed the stalling, but made the idle wonky (IACV was probably having problems compensating).

The timing was checked by pulling the timing belt covers, rotating the engine (by hand) until the crank sprocket timing mark was aligned with the mark on the case, then looked at both cam sprockets and verified the marks were facing their respective marks (straight up). I also inspected all bearings and tensioner for any obvious play/damage. I did not notice any slop/play on the cam sprockets while I was there, though it's not hard to check again if necessary.

I have not found any exhaust leaks around the O2 sensors, nor around the header.

I did do a compression test and all cylinders read 170psi +/- 3 psi. Factory spec standard is 176 psi, limit of 137 psi. Variation limit of 7 psi. All of which checks out.

I used a mechanical vacuum gauge which showed a fairly steady reading of 18 inhg while the MAP sensor readings were around 5.8 apsi. There was a little bit of noise but the port I used was for the evap system, so it was on a single runner, a little fluctuation is expected. Idling for extended periods seems to lower the MAP readings a bit; by maybe 1 psi.

I have not yet adjusted the valves. I had planned on doing this some lazy weekend as they are a bit ticky for stock. The engine has roughly 95k on it now since it was rebuilt- as per what the previous owner has told me. (The body has ~225k) I plan was to do the timing belt and the valves on the same go since they're both due.

 

Well snap, I'm all out of guesses.

Lets see what we can see in the scan data.

When your cable arrives warm up the engine and log a test drive or separate test drives that includes an extended idle period, a full throttle acceleration up to freeway speeds (revving through 5000rpm), and a steady cruise for 20-30 seconds.

Include RPMs, TPS, both short and long term fuel trims, front O2 mA reading or voltage if it doesn't display mA, rear O2 voltage, MAP sensor, and calculated load.

Be sure to point out where the problem occurs.

 

Sounds good guys, will update when I have more.

By the way, the problem on the link was an extra backslash ( \ ) at the end of the url, probably fatfingered when you pressed enter. I've done it tons of times.

 

Ok, cable is on the way. I'm hoping it will be here late next week. I've downloaded romraider and I'll hopefully be datalogging soon.

 

Cable arrived today. I am having trouble getting the drivers ect installed to my computer. I'm running windows 7 64bit, and the disc fails to load and all the other drivers I can find online also fail to work. I succeeded in crashing my computer and almost didn't get it back so I feel like I should ask someone how they got it to work. The last thing I am going to try is download the XP mode and run the drivers there, but if that fails I am stuck.

This is a VAG-COM 409.1 cable by the way.

UPDATE:

I was able to get the driver working by letting windows update search for it. I was amazed, but it is now working. For those curious, both FreeSSM and RomRaider are working. Logging will start probably tomorrow if I have time.

 

Made logs this morning. Apologies for how long and hard to read they are, I didn't pay attention to the order of data. MAP readings are absolute, I'll be sure to use relative readings next time.

Colddrive- Car was still warming up, usually this is the worst time for the car stalling. It died around line 615.

Cruisecoast- Car is about at operating temps. I was cruising around a bit, went down a few hills and coasted. I was hoping to find something as I transitioned back to throttle.

Rev- Sitting at a stop light, I blipped the throttle a few times. Engine seems to stumble every time it returns to idle.

Idleslow- Slow speed, stop and go traffic.


Let me know what parameters and what order to put them in works best.

I'll be doing a bit more and try to make it easier to read as I go.

Thanks in advance!

 

The parameters I would like to see (in this order preferably) are throttle sensor voltage, engine speed (rpm), A/F correction, A/F learning, Rear O2 sensor voltage*, Front A/F sensor mA or voltage or both, MAP sensor, calculated load*, and finally open/closed loop status (maybe called fuel system status?).


*Note that you included 'Rear O2 Heater Voltage' but we need to see the sensor output voltage, also you included 'engine load (relative)' but we need to see calculated load.

Get your engine hot then I need you to log an idle period, a full throttle acceleration up to freeway speeds (pedal to the floor through 5000+rpms in 3rd gear), a steady cruise for 30 seconds or so (steady pedal, cruise control helps), and a deceleration back down to a stop (foot off the pedal). Doesn't have to be in that order. Point out when it is running the worst, or is that just any time at idle?

 

I think I goofed on the Rear heater voltage, I thought I got the sensor voltage, my mistake.

I do have some of this done, though with only part of the parameters. I did a couple pulls, an idle period and I may have a part where I did some decelleration as well. I can put that up now and log the rest tomorrow if you like. One of the pulls was essentially full throttle (though I had to wait so there is a lull under 2k rpm before 100%), another was a steady foot 28% throttle pull, which I did to get an idea what was going on without going into open loop as long as possible. Not sure if it is useful but that's what I did. I'll try to get the full setup tomorrow regardless.

And the worst running that I notice as a driver (not looking at data) is definitely at idle. Looking at the logs some more it looks like the lack of power I am seeing comes from the knock correction.

 

It doesn't surprise me that you feel the engine running its worst while at idle, this is where the largest fuel correction is happening.

Yes, the test drives you did are good they just need to include rear O2 voltage, calculated load instead of relative, and front A/F sensor mA or voltage instead of AFR.