Detonation in naturally-aspirated Phase II 2.5

[southsidescooby]

I own a 2002 Outback Legacy which I purchased used with 25K on the clock. What bothers me obout this vehicle is the almost constant spark knock under light/moderate acceleration. I have taken it to the dealer twice and they have reportedly done extensive testing on the car including real-time data logging. This information has also been extended to Subaru of America. They report nothing wrong with the car. At the advice of SOA, the dealer reflashed the timing control map in the ECU. This had little if no effect. I have noticed through experimentation over many months that premium fuel will supress the pinging almost completely and I have also reset the ECU by leaving the battery disconnected overnight.

I am concerned that this will lead to early head gasket failure and other problems. Other things I have noticed are off-idle hesitation and inconsistent idle(intermittent). This car also accelerates like I am towing a boat or something. I just thought they all did this since it weighs in at over 3500 lbs. and has such a small displacement engine. I was really surprised when I drove an '03 Outback loaner from the dealer with 12K miles. That car responded like a completely different vehicle!!! Has anyone else experienced these problems (what I perceive as problems but the dealer tells me otherwise).

[RallyKeith]

It sounds like there is something wrong somewhere that just isn't obvious.

 

Something important to know is that Subaru ECUs are learning. When you reset the ECU you are actually going to make the knock worse. The ECU is supposed to sense knock and retard the timing, and also adjust a Timing advance setting in the ECU. This was a big problem when the STi came out. It had too much timing advance at one spot in the curve and it seemed that the ECU was learning too slow. The subaru dealership should have the ability to "globally" retard your ignition by a few degrees, but as soon as you reset the ECU it is lost. By Globally I mean that it steps the entire timing curve, and not just timing at a given engine RPM. ECU Tech makes a $400 software package called Delta Dash that allows you to look at all the parameters in the ECU and Globally adjust timing and idle speed, but the same still goes, it forgets as soon as you reset the ECU.

 

I would say keep on the dealership until they can resolve it, and try to find some other 2002s on car lots to drive and compare. No matter what a car should not ping unless something is wrong, or it's an extremely high performance car like the STi.

 

Keith

[southsidescooby]

Thanks for the reply RallyKeith! The ECU was reset when I completed the 30K service and was adjusting the valve lash. The dealer flashed the ECU @ approximately 38K which was some 500 miles ago. When I drove the '03 Outback I noticed spark knock as well but it did not seem to be as noticeable as mine. I figure it must just be the way these cars are programmed regarding fuel management as they obviously live on the stoichiometric "edge" from what the plug insulators looked like after the first 30K miles or so.

 

My car also used coolant when I first purchased it but the Bar's Leaks (The OEM variety, not the heavy duty type) I administered seems to have stopped that. I have a feeling this is the "cooling system conditioner" that Subaru is now recommending for these vehicles. I appluad Fuji Heavy Industries for extending the powertrain warranty on these cars...many of us will need it!

 

I am still baffled by the apparent lack of off-idle power in this car. I don't see a difference in the two years unless it is a software issue. I actually thought of driving another '02 Outback and think this is an excellent idea.

 

If it is of any consequence I am a mechanic(aircraft) by trade and have maintained my own cars for some 20 years now. These include Volvo, VW, Mazda, SAAB, Chevrolet, and Honda. This is my first Suby and hopefully not my last!

[99obw]

Have you tried a carbon removing cleaner such as Techron or Seafoam?

[cookie]

gas at he pump here in CA. It does not ping at all and has good power for its size.

It does the psiton slap till warm and then is smooth and quiet. Actually smoother than both the Mercedes and the Accura I have had.

Like kieth I think there is something wrong that may be hard to find.

If it was my car I would ask them to change the knock sensor as I have heard so much about them that was not good.

[RallyKeith]

If it was my car I would ask them to change the knock sensor as I have heard so much about them that was not good.

 

Good Point. I read something about that at some point. If the sensor isn't detecting it than it will never retard the timing.

 

Keith

[blitz]

I had the same off-idle hestitation and weak-in-the-knees throttle response problem with my '02 from the day I picked it up.

 

Lean mixture city.

 

Part-throttle mixture is controlled by the front 02 sensor "shooting for" (repeatedly crossing over) the lambda target. The default lambda target is stored as a part of the base (non-volitile) map. Resetting the ECU restores the lambda target back to this default value meaning that after a reset, the part-throttle fuel mixture will default back to a base starting value.

 

Average readings taken from the REAR sensor are used to shift this base lambda target (trim the part-throttle fuel mixture) over the course of several "driving events".

 

On my vehicle, part-throttle mixture would be fine after an ECU reset, but would learn itself to be leaner after several driving events, causing the off-idle-stumble, and the big hole in the throttle response.

 

The fix: I tied a 27k resistor across the output of the rear 02 sensor. This loads down the sensor's output, dropping the voltage being fed back to the ECU by about 10%. ECU thinks that the cruising mixture is a little bit lean, so then predictably compensates by learning itself to be just a little richer.

 

Voila, another 4 cent fix.

 

You may have to run 89 octane anyway as your mileage accumulates, especially in hot weather.

[cookie]

Does my 99 have two knock sensors also?

All I have had trouble with have been mechanical items and the electricals have given no grief at all.

I do have a light out on my AC switch and the CD player sucked but I can forgive those.

The only Subie I had to do with years ago had a carb and gave me no trouble either.

I was under the impression that I have one knock sensor, one cam sensor, and a temp sensor or two.

[RallyKeith]

Does my 99 have two knock sensors also?

All I have had trouble with have been mechanical items and the electricals have given no grief at all.

I do have a light out on my AC switch and the CD player sucked but I can forgive those.

The only Subie I had to do with years ago had a carb and gave me no trouble either.

I was under the impression that I have one knock sensor, one cam sensor, and a temp sensor or two.

 

I believe every car that is OBD-II and on has two oxygen sensors. One is in front of the Catalytic convertor and the second is behind it. They actually use the difference in readings between the two to determine if the Catalytic convertor is working or not.

 

Keith

[cookie]

is one more indication of my senilty. I was jammed on knock sensor and read knock sensor into that message.

I do know that it has two o2 sensors and now that makes sense. I guess I better go back to rocking on the porch.

[southsidescooby]

I am certainly impressed with all these great responses!

 

Blitz, if I might ask how did you arrive at that particular resistance value? This really makes me wonder if a software change did not occur between '02 and '03 Subaru model years. I would imagine the oxygen sensors are probably the same P/N for both years (front and rear respectively). I am definitely going to discuss this with my dealer (not the resister mod) and see if a software upgrade might be available. If not I will certainly try the sensor compensation trick.

 

99obw, I also considered a fuel/induction cleaning and/or additive but found it hard to grasp how those systems could be that contaminated considering the good fuel we generally get here in Georgia. I regularly use the Redline fuel additive which seems to keep things quite happy in my other cars. I have not tried those products you mentioned but am sure they perform well also.

 

Cookie, I mentioned the knock sensor, MAP sensor, and 02 sensors to the head of the service department as possible culprits. He assured me that all these are functioning within tolerance. I have a mind to break out my maintenance manual and do some checking for myself. I just would like the repair to be on their nickel not mine!

 

Again, I appreciate all the help and suggestions. I will certainly keep you all posted.

[99obw]

I don't have nearly the knowledge about these systems that Blitz and some others have, so I have to start with the simple stuff. Using Techron is dirt cheap. If I were you I would run Techron in two consecutive tanks of gas just for giggles. I carpool to work with a guy that has a newer dodge truck with a 3.8. I know, apples and oranges, BUT, this truck periodically develops similar symptoms and Techron cleans it up every time.

 

You said that the engine still detonates on 92 octane, in my limited experience this makes no sense to me. To me either you have mega carbon that is causing preignition, or your knock sensor is totally hosed. If this happened to me it would probably be the excuse I would need to buy a scan tool for my palm pilot and start monitoring things like the knock sensor and ignition timing while driving the car.

 

Blitz, may I ask where you learn this detailed information about the operation of the ECU? I would love to learn more.

[myles]

The signal from the rear 02 sensor is not used as input when computing the air/fuel ratio. It's used to keep tabs on the health of the cat.

[cookie]

I still would be very tempted to go old school here. I have seen knock sensors that seemed to test perfectly to me. (this is an old fart in his garage with a radio shack meter).

I saw absoloutly nothing wrong with my neighbor's sensor by testing but the bloody thing lept pinging. According to the book the knock sensor was fine and I checked the connections and to see if it was the correct sensor.

I changed it and it works fine. Why? I don't know.

[blitz]

SNIP

Blitz, if I might ask how did you arrive at that particular resistance value?

SNIP

 

It took about 8 months of driving and observing.

 

First I extended a length of shielded cable from the rear 02 sensor into the cockpit so I could hook up a Fluke Multimeter and monitor the sensor's voltage output during real-time driving. The meter has a data-logging feature that records minimum, maximum and average.

 

I noticed that the rear sensor doesnt "switch" in an active closed-loop control circuit like the front sensor does, but rather reads the oxygen content as it appears at that point in time.

 

At the end of a drive I would read the recorded average, and after about 20 driving events I realized that the average of every drive would be close to .555 volts. Short, punchy, around-town type drives (more hydrocarbons) tended to move the average number higher initially, but ultimately it would learn itself back down to .555 Volts. Long, steady-state highway cruising (lesser hydrocarbons) would move the average number lower initially, but ultimately it would learn itself back up to .555 Volts. Thus the nature of adaptive fuel trim.

 

The ECU averages the output of the rear 02 sensor just like the Fluke meter does, and when the drive is complete and the key is shut of, the ECU logs this average, and uses it to slightly shift the lambda target up or down in anticipation of the next driving event. Remember that the lambda target is the number which the front 02 sensor actively shoots for during all driving (except for full-throttle and cold startup). In other words the average voltage taken from the rear sensor ultimately becomes the arbiter of that

point (lambda target) to which the front sensor is shackled to, as it attempts to control fuel mixture.

 

I decided that using a resistor to shift the number by 3% would be a good, practical starting point. 100k gave me that 3% number shift. After a week of driving, I lowered the value to 82k for a week, then 68k, then 56k... and stopped at 27k to where the car feels strong now. I guess you'd know if you've gone too far by a drop in fuel economy and soot at the tailpipe. Just make slow, steady changes until it feels right.

[blitz]

SNIP

Blitz, may I ask where you learn this detailed information about the operation of the ECU? I would love to learn more.

SNIP

 

Until about two years ago I knew avsolutely nothing about OBDII, but a buddy of mine who formerly was a service tech for Chrysler and did a lot of troubleshooting regarding "driveability issues" on individual customer vehicles explained a lot of things to me in a way that made sense.

 

Myles is correct about the "cat-efficiency" function of the rear 02 sensor, but it functions on a different principle than the adaptive fuel trim. Instead of the ECU reading voltage, it counts and compares the number of "crossings" of the rear 02 sensor against the number of "crossings" of the front 02 sensor.

 

The front sensor is supposed to switch like a mutha, and if the number of crossings slows down appreciably, then the sensor has become contaminated or "Lazy". If the cat is performing properly, the rear sensor shouldn't be switching much at all. However should the cat became completely non-functional (test-pipe?) then the rear sensor would be switching at the same rate as the front sensor thereby triggering a cat-efficiency (inefficiency) code.

[cookie]

When I started doing this we turned a screw or changed a jet to accomplish the same thing.

What is frustrating is that you seem to have had to re engineer a system that should work right from the factory.

Regardless it seems to me that this car should be repaired on warranty to actually run the way it should.

We are probably all running slightly different gas in different conditions but this system should compensate if all components are in specs and properly engineered.

[blitz]

Cookie, I started driving during the 70's "emissions era" so the throttle feel of an overly-lean motor with retarded timing is etched in my brain. Lousy driveability as the end result of trying to get 60's engines to meet 70's emissions requirements. Who would've guessed that all these years later it would be essentially the same thing all over again?: Driveability issues related to getting 90's engines to meet 2000's emissions requirements?

 

I get the feeling that certain '02 vehicles were beta test cases for trying to get the EJ-25 to conform to ULEV in advance of the Federal mandate. This way Subaru could monitor customer (guinnea pig) complaints in a more relaxed fashion as a measure of success or failure and make changes accordingly. :-\

[cookie]

stretching an old design a bit too far. VW did this in the seventies and buying Audi got them out of it. Audi had some fine designs but was a failure in the market.

Subaru put a lot of its development into the rally cars and the WRX line when they were tight on money.

Makes me wonder what effect all this had on the decision to stretch to 2.5 and why it did not seem to be up to the old standards.