Subaru Outback Forums banner

Lack of power, occasional misfire, then... normal

13K views 65 replies 8 participants last post by  CSFiend 
#1 ·
Hey folks,

Looking for some help diagnosing my sister's 01 Legacy. 244k, automatic, well maintained.

When accelerating, often times there's a noticeable drop in power. It's hard to explain, but it'll hesitate and bog and feel way down on power and then... poof, it runs perfect. It's a very definitive moment when the power returns to normal -- not a gradual increase, it'll be struggling one moment and all of a sudden it's running great. Cruising on the highway you can feel a very slightly surging.

I've checked for codes with FreeSSM on my laptop and there are none.

Things I've noticed:
- Coolant temp gauge never gets up to halfway -- sits around 175* F.
- Exhaust leak at flange between exhaust manifold and cat.
- TPS doesn't seem to report 100% -- when floored, it'd show 85%.

I haven't checked the PCV valve yet. Pulled the plug wires off the coils to check the condition of the posts and they were clean with some dielectric grease. I didn't pull plugs because it doesn't feel like a spark issue. Almost seems like a fuel delivery problem, but the fuel filter flows smoothly (took it out and blew through it -- no restriction). Haven't checked fuel pressure. The exhaust leak before the cat seems like a slam-dunk because it's just before the front o2 sensor, but the way it goes from running poorly to normal just doesn't seem like an exhaust leak.

Any thoughts? I realize this is a rambling post but I don't want to throw parts at it without some supporting diagnosis.
 
See less See more
#2 ·
Knock sensor.

They crack with age, and the computer can't tell the difference between real knocking and false positives, so there's no diagnostic code or check-engine light.

A cracked sensor will intermittently send lots of false positives, causing the ECU to radically retard timing, resulting in drastic power loss and drive ability problems.

They're under $20 on eBay. Probably counterfeit but they seem to work as well as the real deal.

Could also be the early stages of a failing fuel pump, which also won't generate any diagnostic codes. Fuel pressure test would prove it in or out quickly.

You should also look into calibrating that tps and sealing up the exhaust leak.
 
#3 ·
I was logging what the knock sensor was doing and I don't believe it was showing knock when it was running poorly but I may have to double check that. Same with timing -- what's the normal adv/ret degrees supposed to be?

I'm leaning towards a fuel pump. Is there a schrader valve to attach the gauge to or do you have to put it in-line coming off the fuel filter?
 
#5 ·
I pulled the wires from the coil pack to check the posts for rust, but I neglected to pull the wires from the plugs themselves because it didn't feel like a spark related issue. Easy enough to do; I'll check them later today or tomorrow.
 
#6 ·
While your at it closely inspect the ground wire from the battery to the ground we had lots of odd issues with our 2010 LEgacy due to the battery venting around the neg terminal and causing heavy corrosion under the cover on the neg ground wire which looked OK but was pretty much totally gone under the cover. New ground fixed all the issues a new battery fixed the corrosion issue.
 
#9 ·
Forgot to mention in my original post -- I cleaned all the grounds and checked for corrosion in the wire (under the insulation) and they all looked good.

in the end of this just before I actually had the front o2 senor swapped out, I had some major bucking (enough to jar me in the driver seat), WITHOUT a check engine code, ...but I was not standing on the gas trying to produce a code.

http://www.subaruoutback.org/forums...89-p0130-2002-ej251-climbing-grades-fast.html

in the last post, ....to get that 2001 OEM front o2 sensor out...."air fuel sensor", it did take a torch. (not ideal as it can leave crap on the threads).


but post up what you got in logs like I did, @plain OM @cdc @ntippet and others might see the exact problem in the data.
I'll try and get a drive in tonight and log all the requested info. Thanks for the suggestion :)

If you're not seeing knock activity or a big slew in timing coincident with the problem, I'd chase the fuel pump or ignition lines instead.

No valve. You need to make your own tee fitting at the filter.
Figures. Hope the kit comes with a T that fits... any idea what the fuel pressure specs are supposed to be? I don't have my service manual handy at the moment.
 
#7 ·
in the end of this just before I actually had the front o2 senor swapped out, I had some major bucking (enough to jar me in the driver seat), WITHOUT a check engine code, ...but I was not standing on the gas trying to produce a code.

http://www.subaruoutback.org/forums...89-p0130-2002-ej251-climbing-grades-fast.html

in the last post, ....to get that 2001 OEM front o2 sensor out...."air fuel sensor", it did take a torch. (not ideal as it can leave crap on the threads).


but post up what you got in logs like I did, @plain OM @cdc @ntippet and others might see the exact problem in the data.
 
#8 ·
If you're not seeing knock activity or a big slew in timing coincident with the problem, I'd chase the fuel pump or ignition lines instead.

No valve. You need to make your own tee fitting at the filter.

Edit: Though... As long as you're logging...

A fuel pump problem should show up in the form of massive positive fuel trims, disappearing once the pressure comes back.

This might be harder to spot due to that exhaust leak.
 
#11 ·
blind men have been known to make babies though. :cool:

1 plus (too many possibilities without logs or a code).
 
#13 ·
Had the gasket between the manifold and cat replaced over lunch, didn't seem to make much of a difference in performance.

There are three logs attached:
- 'accel_uphill_then_cruise_35mph' -- Stoplight, accelerate to and maintain 35mph on a slight incline, then put my foot into it before coming to the next stoplight.
- 'knock_correction_stoplight_and_parking_ramp' - Stoplight to stoplight, then through our parking ramp. Enabled knock correction logging for this one, just out of curiosity.
- 'stoplight_to_stoplight' -- Stoplight to stoplight.

The problem didn't appear as severe as it has in the past, but it seems to run better when my laptop is connected... I may just have to bring my laptop with more often and try to grab a log when it happens again.

Would someone more knowledgeable than me mind skimming my logs and see if anything looks out of whack? Much appreciated :)
 

Attachments

#14 ·
Not enough data, in my view. It's not apparent to me where the power loss is occurring.

The rear O2 sensor voltage seems to remain unusually low most of the time, especially in the first log, but also in the latter two. This could be a symptom of something else that's not right. We would need more data streams to see what's happening elsewhere to be able to follow this path.

Also notable in the first log, during the "cruise" portion, the Learned Ignition Timing is negative. That too would be a symptom of something else that's amiss.

For an example of a "well-stocked" log, look at the one eagleeye provided in this post
 
#15 ·
Yeah, I know... it's a pretty brief glimpse into a two mile stretch, just curious if anything jumped out because it certainly seems underpowered but didn't do the whole bogging thing while I was logging.

I agree on the rear o2 -- I noticed that too. It spikes in weird ways.

Same with the negative ignition timing. I'll take a look at eagleeye's log and report back. Thanks.
 
#16 ·
Log these values:

32bit ECU

Accelerator Pedal Angle (%)
A/F Correction #1 (%)
A/F Correction #2 (%)
A/F Learning #1 (%)
A/F Learning #2 (%)
A/F Sensor #1 (AFR)
A/F Sensor #2 (AFR)
Coolant Temperature (C)
Engine Load (4-Byte)* (g/rev)
Engine Speed (rpm)
Feedback Knock Correction (degrees)
Fine Learning Knock Correction (degrees)
Gear (Calculated)* (position)
IAM (4-byte)* (multiplier)
Ignition Total Timing (degrees)
Injector Duty Cycle (%)
Intake Air Temperature (C)
Intake VVT Advance Angle Left (degrees)
Intake VVT Advance Angle Right (degrees)
Knock Correction Advance (degrees)
Manifold Relative Pressure (psi relative)
Mass Airflow (g/s)
Mass Airflow Sensor Voltage (V)
Throttle Opening Angle (%)
Vehicle Speed (kph)

They should all be available via Romraider, as I used to log these in the past when I was tuning my BH wagon before I sold it.
 
#17 ·
Thanks, guys. I didn't have all the parameters above available, but 90% of them.

I finally experienced the bogging my sister complained about. Taking off from a stoplight, put your foot down, and it just won't accelerate much. Put my foot down further and further until 3/4 or so, and then it started accelerating but was still bucking. As soon as it upshifted into second gear, the bogging picked right back up.

I logged the entire drive, and it's attached. The bogging starts around line 10,900.

Drove to the highway, drove 65mph for a bit, turned around, drove back, got off the highway, and at the next stoplight is where it started misbehaving.

Any suggestions are much appreciated.

Out of curiosity, what do you guys use to view these logs? Excel and plot the data on a line graph?
 

Attachments

#18 ·
rom-raider when you connect should give you a maximum list of parameters available for the specific engine / trans. when disconnected you might see that you have more available for more complex engines like turbos.

just make sure everything is checked off to log,....and to aid in the people reading and interpreting the logs, what units they want read the stuff in,...(just to save them a step when looking at the data so they don't have to convert units too much and maybe loose something)
_______

in my issue, I found it easiest to make the problem happen on steep grades where I could stop and then go full throttle, and then stop again and go full throttle again.


as time went on I could get problems on the flat, first just check engine lights, then as time went on real bucking.

@plain OM do you think it would be advantageous to yourself or others if I took some basic default logs on my EJ251 car?

maybe parked idle, parked reved sweep up to the red line, and some gentle one minute cruising, or a full launch.

(102,000 miles = 164,000 km, head gaskets done with turbo gaskets,...all the maint. up to date, including double platinum NGK plugs,...OEM sized wheels and tires).

my laptop is even running a lot better then it has in the past few months. (windows 10 fix).
 
#19 ·
Right, that's what I was saying -- when connected to the ECU I don't have a couple parameters listed -- IAM, calculated gear, A/F sensor #2 doesn't show up in the list but 'Rear o2 sensor' does.

I used the requested units from ntippet so coolant, vehicle speed, etc. are all metric values.

I'd love to be able to reproduce the issue but after 250 miles of driving it just now acted up for me. I've been driving around with my laptop hooked up and ready to log but it hasn't misbehaved until today.
 
#20 ·
Edit: Ntippets car was a Ej259 = the cal spec 2004 H4 car. (not exactly the same as a EJ251, lots of little fixes, including sensors)

the list may say "front o2 sensor" when you get it hooked up, instead of and A/F sensor for this particular car.

plenty of A/F sensors on the default list,...and something like a turbo car many have more.

do you know if they are the original sensors? ...If you look at the first post in my thread I tried to give a list of what was original and what was updated. (like front o2 being original, rear o2 set were swapped out for densos a few years ago due to P0420 code).

244,000 miles wow, that is like 16,000 miles per year without the car ever getting shelved for another ,better ride,

....but it is the creme de la creme of automobile transportation, what could be better.? :wink2:
 
#22 · (Edited)
ntippet or others will, hopefully, pipe in as well with views on the log, but here's my first impression.

Between lines 2502 and 3784, the throttle is at zero, engine speed is at idle from about line 2750. When the engine is idling, A/F correction is low (near zero), but A/F Correction #3 is up at +5%. In fact it stays there throughout the log. That seems a high number, especially because there's no change. However while I'm not familiar with the 2001 characteristics, I believe A/F Correction #3 is derived from the Rear O2 Sensor condition. Normally the rear O2 signal should change with when the throttle goes either very high, or is dropped to zero.

But more significant is the fact that during the idle period, the A/F Learning is holding at -15%, and remains quite negative, on average, throughout the log. The A/F Sensor is fairly close to the target 14.7:1 ratio, so the fuel system seems to be managing this. And again, the Rear O2 Voltage is sitting at the bottom.

I suspect that if the situation at idle could be examined more closely, and these strange reports are sustained, the fix for will likely deal with the surge/hesitation when driving.

The short period of idle that is shown, however, does suggest the A/F Sensor might be faulty. Fuel is being pulled, and this is confirmed by the rear O2 sensor, yet the A/F Sensor signal appears to be fairly normal.

On other topics, I am using Excel, and running line graphs (sometime scatter graphs with lines).

The A/F sensor signal is the AFR, which is the apparent air/fuel ratio that the ECM targets. It does this by rapidly switching the actual fuel delivery from lean to rich and back. As a result of this, the Voltage put out by a traditional O2 sensor will switch between a voltage below 0.5 to something between 0.5 and 1.0 Volts. This voltage can be accessed in many cases by using a scanner that connects through the generic OBDII protocol. But Romraider Logger normally accesses through the SSM protocol, which has many more available parameters. Ideally, the A/F Sensor should be at 14.7, but it will move in response to external factors such as throttle position. Overall, however, it should remain around 14.7.

Incidentally, which engine are we looking at (I don't think it's mentioned above)?
 
#30 ·
ntippet or others will, hopefully, pipe in as well with views on the log, but here's my first impression.

Between lines 2502 and 3784, the throttle is at zero, engine speed is at idle from about line 2750. When the engine is idling, A/F correction is low (near zero), but A/F Correction #3 is up at +5%. In fact it stays there throughout the log. That seems a high number, especially because there's no change. However while I'm not familiar with the 2001 characteristics, I believe A/F Correction #3 is derived from the Rear O2 Sensor condition. Normally the rear O2 signal should change with when the throttle goes either very high, or is dropped to zero.

But more significant is the fact that during the idle period, the A/F Learning is holding at -15%, and remains quite negative, on average, throughout the log. The A/F Sensor is fairly close to the target 14.7:1 ratio, so the fuel system seems to be managing this. And again, the Rear O2 Voltage is sitting at the bottom.

I suspect that if the situation at idle could be examined more closely, and these strange reports are sustained, the fix for will likely deal with the surge/hesitation when driving.

The short period of idle that is shown, however, does suggest the A/F Sensor might be faulty. Fuel is being pulled, and this is confirmed by the rear O2 sensor, yet the A/F Sensor signal appears to be fairly normal.

On other topics, I am using Excel, and running line graphs (sometime scatter graphs with lines).

The A/F sensor signal is the AFR, which is the apparent air/fuel ratio that the ECM targets. It does this by rapidly switching the actual fuel delivery from lean to rich and back. As a result of this, the Voltage put out by a traditional O2 sensor will switch between a voltage below 0.5 to something between 0.5 and 1.0 Volts. This voltage can be accessed in many cases by using a scanner that connects through the generic OBDII protocol. But Romraider Logger normally accesses through the SSM protocol, which has many more available parameters. Ideally, the A/F Sensor should be at 14.7, but it will move in response to external factors such as throttle position. Overall, however, it should remain around 14.7.

Incidentally, which engine are we looking at (I don't think it's mentioned above)?
Thank you for the explanation of what you’re seeing.

It’s a 2.5 SOHC.


There is still some decent data immediately after the lost section.

Have you noticed knock count PID counting up on any of your logs or viewing data while driving? I don’t show a knock count, but I have watched in SSM and the knock sensor is confirmed to have picked up knock by a max input of ‘on’.
Anything rattling on or around the engine? – Negative.
Exhaust heat shields loose? Does the engine have piston slap while cold? – No loose heat shields (welded all of them to the exhaust). Motor does have piston slap when cold but goes away when warm.
Have you replaced the knock sensor? – Negative.
Are the knock sensor wires damaged? Is the shielding around the signal wire still intact? No rodents chewing on any of the harnesses? – No damage to the sensor or harness that I can notice.

Also there is a wide section during accel in either 2nd or 3rd gear where A/F correction completely bottoms out the available range -30%.
@ time 1162688 it then immediately recovers A/F correction to 0%, my guess is it switches to open loop, disregarding the A/F sensor values.
This is probably when you feel the car recovering during acceleration, it is then ignoring the A/F sensor completely and only using stored values for timing and fueling. This would make sense; there's a definite switch between accelerating poorly and normal acceleration. A/F senor value drops to 10.22, otherwise in closed loop the ECM is always targeting 14.7 A/F
You can log ol/cl PID to see when this switch occurs.

Have you replaced the A/F sensor?
Are the A/F wires damaged? No rodents chewing on any of the harnesses? – Negative. The front sensor in my 03 Outback is fairly new; I need to check if it’s the same part and I could probably swap it to the 01 Legacy to prove the fault.

1st problem I see is the A/F sensor almost never hits 14.7.
The ECM is continuously trying to pull fuel to try and hit this value, the A/F sensor never goes up.
Process repeats.

Change your A/F sensor and report back.
Get an OEM quality one.
Also fix the exhaust manifold leaks, this can also be contributing to it. – Exhaust leak between manifold at cat was fixed last Thursday or Friday, no change in behavior.
 
#23 ·
Please note that there's a time gap between lines 10963 and 10964, covering a span of about 10 minutes.

In addition, at line 10965, vehicle speed drops to zero, stays there until line 11030 (about 4 seconds), and then slowly increases until at line 11352 the speed is close to where it should be.

More or less in the same time frame as the vehicle speed anomaly, a number of other data series drop to zero, or to a level that seems to be more consistent with an idling engine. I could imagine some of these appearing if the throttle pedal was released, but the vehicle speed isn't likely to drop from 100 kph to zero in less than a second.

I raise this only to caution about the reliability of the data in those areas (which, unfortunately, coincide with where CSFiend identified the bogging down).

CSFiend: Can you make a short (1-2 minutes) log with the engine idling (after full warm up and letting the engine idle for a period to stabilize before logging). There should be about 29 parameters available in Romraider Logger under the "Parameters" tab. Use pretty well all; but select only Manifold Relative Pressure (out of 3 related manifold parameters that are listed).
 
#24 ·
Please note that there's a time gap between lines 10963 and 10964, covering a span of about 10 minutes.

In addition, at line 10965, vehicle speed drops to zero, stays there until line 11030 (about 4 seconds), and then slowly increases until at line 11352 the speed is close to where it should be.

More or less in the same time frame as the vehicle speed anomaly, a number of other data series drop to zero, or to a level that seems to be more consistent with an idling engine. I could imagine some of these appearing if the throttle pedal was released, but the vehicle speed isn't likely to drop from 100 kph to zero in less than a second.

I raise this only to caution about the reliability of the data in those areas (which, unfortunately, coincide with where CSFiend identified the bogging down).
if on a cord or a dongle such things do wiggle off the access port and then reconnect. I guess it could possibly wiggle loose on one side and reconnect.

alternatively:

I wonder if it is possible there is a power problem causing this,...like something is shorting the whole car out. = 100kmh to zero in one second as the whole car is electrically shutting off.

(not that I looked at the data,...just making a comment based on plain OM's post)
 
#25 ·
The display on the laptop turned off and when I hit the spacebar to wake it up I was back in romraider and it was still recording. I must have interpreted it as a constant recording but it probably went to sleep.

Argh. I'm so sorry guys. I'm fairly frustrated. I'll tweak the power settings on my laptop and try to reproduce the issue this weekend.

Thank you so much for your help.
 
#27 ·
stuff I learned from my log attempts:

I slave my laptop on the power cord into my portable jump pack, so it gets continuous power. ....driving around with the log ready to start with the touch of the cars rear defroster button would eat through a laptop battery sadly fast.

considering putting the laptop into the cars own power supply is something I would only consider doing on a car with zero problems. as one is giving the car a extra power drain,...and to cancel any check engine lights you have to stop and start the car.
 
#26 ·
I subsequently recalled that I've seen these types of breaks in logs before -- not sure when (I've been looking at logs, my own and others posted here, for almost four years . . . ), but as I recall, they were, as in this case, long logs. I appreciate that the objective was to continue logging until the power loss appeared and, indeed, the data at that exact time might be useful to see. However, at least at this point, I'd be interested in examining the conditions at idle more closely, because that large negative fuel correction at idle could well be indicating a problem.
 
#28 ·
There is still some decent data immediately after the lost section.

Have you noticed knock count PID counting up on any of your logs or viewing data while driving?
Anything rattling on or around the engine?
Exhaust heat shields loose? Does the engine have piston slap while cold?
Have you replaced the knock sensor?
Are the knock sensor wires damaged? Is the shielding around the signal wire still intact? No rodents chewing on any of the harnesses?

Also there is a wide section during accel in either 2nd or 3rd gear where A/F correction completely bottoms out the available range -30%.
@ time 1162688 it then immediately recovers A/F correction to 0%, my guess is it switches to open loop, disregarding the A/F sensor values.
The throttle & engine load was up long enough to count out the "ol/cl delay" value to switch to open loop fueling.
This is probably when you feel the car recovering during acceleration, it is then ignoring the A/F sensor completely and only using stored values for timing and fueling.
A/F senor value drops to 10.22, otherwise in closed loop the ECM is always targeting 14.7 A/F
You can log ol/cl PID to see when this switch occurs.

Have you replaced the A/F sensor?
Are the A/F wires damaged? No rodents chewing on any of the harnesses?

1st problem I see is the A/F sensor almost never hits 14.7.
The ECM is continuously trying to pull fuel to try and hit this value, the A/F sensor never goes up.
Process repeats.

Change your A/F sensor and report back.
Get an OEM quality one.
Also fix the exhaust manifold leaks, this can also be contributing to it.
 
#29 ·
I grabbed two logs today -- one sitting in park after driving for 15 minutes, and one driving with the cruise set at 65mph.

I'll reply to the above posts shortly, just wanted to get these logs posted. Thank you for taking the time to review them.
 

Attachments

#32 ·
I'll throw the car in the air and double check, but I'm fairly certain that the only exhaust leak in front of the o2 sensor was at the manifold-to-cat flange and that was fixed last week.

As long as that's the case, does that pretty much condemn the front A/F sensor?

Aside of knock, what else would cause timing to be pulled like that?

Is there any data that gives you a clue as to what the fuel pressure may be doing? Are we looking at a symptom of an intermittently failing fuel pump (although I don't know that I've ever seen a fuel pump work intermittently versus either working 100%, working at less than 100% all the time, or complete failure)?
 
#33 ·
How did the car drive on that last log? Normal power delivery?

At this point I wouldn't point fingers at a fuel pump, the ECU is trying to pull fuel, not add it.

If you do test swapping out the A/F sensor (should be the same), pull the ECU fuse or battery cable.
Reset the ECU to stock values and see if it continues the trend of negative learned fuel correction.

Might be worth logging A/F ratio on the donor car, to verify that sensor is good before taking the time to swap them.

Also verify all exhaust leaks have been fixed before front A/F sensor.
Sometimes they hide underneath the heat shields....can be hard to spot..
Manifolds can crack in addition to possible leaking sealing surfaces.
 
#34 ·
Normal delivery only in the sense that it didn't do the extreme bogging that it did during last week's partial log. It's definitely lacking power and surges under constant throttle input.

Good point on pulling fuel. Won't worry about the fuel pump for now.

Rockauto shows different part numbers between my 03 manual and the 01 automatic. I can't imagine they're that different though.
 
#35 ·
Yeah strange, they do look different.
That year 2001 subaru spec'd 2 different A/F sensors a 6 pin and a 4 pin based on trans type.

Later they only used 4 pin.

If the 01 was a standard trans you might be able to swap them over.
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top