Timing retards at 2000 RPM

[amsponr]

(First time poster) Hi everybody, I'm usually a reader and not a poster, but i have this baffling problem with my 99 Forester.

I bought the car last year with this problem and figured i could fix it. Well, i have done what i can and now i need help.

The 2.5 SOHC starts and runs fine. But around 2000 RPM it acts like it loses power and you have to push the pedal down further until it  takes off.

I have hooked it up to a monitor and watched the timing retard every time it hits 2000 rpm. 

I have replaced the front O2 sensor because there was a code. That took care of the light.

I have replaced the knock sensor and still no go. I also replaced the PVC valve.

If i unhook the battery for 10 min, to reset the ECM, I will run fine for 20 -30 miles, but then back to old tricks.

There is no CEL and the trans have no codes either.

Anyone seen this before? Any help is appreciated.

Thanks, AMSPOON

[Fairtax4me]

Either its leaking out and knocking, or the knock sensor is sending a false signal to the ECU.

 

I had a similar problem and replaced the knock sensor with a 580k ohm resistor and the timing pull went away. But then the thing was knocking like crazy. Turned out to be because of the crap fuel they're making us buy around here.

 

During that 20-30 miles after a reset the ECU is playing around with fuel trims trying to find its most efficient trim setting. It's also running emmissions monitors to make sure all of the emmissions control systems are working.

Once it finishes running the monitors and all that it jumps over to the settings that it determined would be the best, and you get the lean mixture and timing being pulled back again.

 

One thing you want to check is the mounting surface for the knock sensor. If the block is all corroded the sensor will not ground properly and it can cause false signals.

It's also possible to deform the sensor by over tightening the bolt. This affects the sensors response to knocks and can cause false signals.

 

Another thing to try is higher octane fuel. See if the problem goes away by running a tank of premium, or non-ethanol fuel. This would indicate the knock sensor is working properly and the problem is either fuel related or something causing the air/fuel ratio to be too lean (vacuum leak, low fuel pressure, clogged injectors, etc.).

[1 Lucky Texan]

how many miles on the car? wondering if it's worth the trouble to check the cam/crank timing. (though, seems unlikely a battery reset to the ECU would restore fully normal perf.)
 
If you could get the FreezeFrame data that some code readers produce, or get an elm327 device to use with Torque app on a smartphone - that data might be helpful.

Edited March 14, 2014 by 1 Lucky Texan

[amsponr]

Thanks for the response, 

 

I've tried high octane, seafoam, a different location for the knock.

I even tried taking the knock sensor, unbolting it, hanging in midair, hooking a jumper cable from the sensor to the ground on the battery. No good.

 

What do you mean by (leaking out?)

Is there anything else that would retard the timing? 

I have heard that the trans can cause a timing change. 

Why no codes?

I may try the 580K resister.

Thanks, AMSPONR

[1 Lucky Texan]

typo - LEANing out, not leak.

[amsponr]

OK, 

Any idea why 2000 rpm?

Car has just over 200 k miles. 

Would a timing belt 1 tooth off do this?

[1 Lucky Texan]

if a crank or cam pulley is one, maybe 2 teeth off - they will run, just not well.

 

you are either WAAAAYYYY overdue for a belt, or geting close to needing one (105K is your schedule I think)

 

 

 

However, the fact that it seems to run well with the factory settings after a reset - would seem to support some air/fuel sensor problem. Though i don't understand why there's no code for an O2 sensor or something related.

 

Does the monitor you have hooked up report fuel trims or other data?

 

if this car has a MAF, you might try cleaning it with the correct solvent - but maybe you have MAP?

Edited March 14, 2014 by 1 Lucky Texan

[amsponr]

The engine runs good at start, and up till and after 2000 rpm.

[1 Lucky Texan]

wait - it ONLY stumbles around 2000? it's OK at 2500, 3000, 3500, etc.?

 

that seems like a Throttle Position Sensor issue, re-reading you original post, I'd say it's quite likely the TPS has a bad/dirty spot on it. Try replacing it. I THINK I've read of some folks using contact cleaner in them but I have no experience with that. I have no idea why that would show as retarded timing other than the ECU assuming you were at idle when it loses the correct throttle opnening percentage.

 

sorry for my confusion/lack of reading skills.

Edited March 14, 2014 by 1 Lucky Texan

[amsponr]

This sounds like a very good possibility!

I will try this.

Is there a trick to replacing and positioning?

[1 Lucky Texan]

probably have to rely on fairtax or other gurus here - i have no experience with the TPS.

[amsponr]

Thanks for your help Tex, I will wait for fairtax or others.

Thanks, AMSPONR

[Fairtax4me]

If its were the TPS sensor the problem would happen only at certain throttle position. If your foot stays in the same place, the problem will happen at all engine speeds.

Since it only happens at a certain engine speed, you can rule out the TPS by using more throttle. If you use 70% throttle to accelerate, and get the same hiccup as you do using only 20% throttle, then the TPS is not the problem.

 

You may also be able to watch the TPS % reading change on the scanner when the problem happens. If TPS reading is unchanged when the problem happens, TPS is not the problem.

[amsponr]

Thanks fairtax, I agree.

 

Today i installed a known good knock sensor, reset the ECU, and drove the car for 20 miles.

 

It ran great until it hit around 15 miles. Then up to is old tricks.

 

If it is actually sensing a knock, why doesn't ECU adjust it back so it doesn't?

 

I am starting to think it is the ECU. 

 

Any thoughts?

[Fairtax4me]

Put a bolt through the knock sensor, get a length of 18awg wire with an eyelet crimped on the end. Bolt that to the sensor with a nut on the end of the bolt. Zip tie the sensor up off the block against one of the hoses in that area. Hook the other end of the wire to chassis ground or battery -.

Drive the car and see if you have the same issue, or if it starts pinging like crazy.

 

If it is actually sensing a knock, why doesn't ECU adjust it back so it doesn't?

That is what it's doing. Once the knock goes away the ECU ramps the timing back up. The ECU doesn't learn to retard timing all the time because its not as efficient to run with less timing advance. So the ECU tries to run as much advance as it can all the time, until it senses knock, then it will pull the timing back a little until the knock goes away. How far it pulls the timing back is determined by how much it knocks, and the severity. If it continues to knock after the ECU has pulled the timing a little, it pulls it back more and more until the knock stops. Edited March 15, 2014 by Fairtax4me

[amsponr]

I did something like this by unbolting the sensor and using a jumper cable,

 

I clamped one end to the sensor and the other to the battery negative.

   

I suspended the sensor and drove, same results.

 

Do you think the results would change if i do it your way?

 

Also, what about the 580K ohm resistor? Would that be better? (is 580k a magic number, or could it be close high or low?)

[Fairtax4me]

580k is approximately the resistance of the knock sensor. If you get within 20k +- tha will be fine.

You should try the resistor next to see if you get pinging.

[amsponr]

I will try this today!

[amsponr]

Well I installed the resistor and went for a drive with the OBD monitor hooked up.

 

It worked fine for my test drive.

 

No ping or knock heard. 

 

No hesitation or timing retard at any RPM.

 

​So if it stays this way for a few days, What is the permanent fix? 

 

 I will let time be the judge. 

[Fairtax4me]

The permanent fix? Leave the resistor in place!

 

Was the "known good" knock sensor brand new?

 

It's possible the shielding on the sensor signal wire was not grounded properly and it was picking up interference from somewhere.

Could also have just been a failed sensor.

 

I have an old dead sensor that I've been meaning to cut apart to see how its constructed.

Id also like to get a scope trace of a sensor signal and compare a good sensor vs a known bad sensor.

[amsponr]

I can't explain it, but with the sensor hooked and the monitor hooked, I took a bar and tapped on the top of the sensor and i didn't get a reading!

 

Still somehow it would sense a knock from the engine.

 

The known good sensor was from my 98 legacy with a 2.2.

 

I had to take the plug apart to make it connect.

 

Same ohm reading.  

 

I am planning on Seafoaming the engine and fuel the next chance i get. 

 

I am thinking that there is a build up on the heads causing the knock.

 

Also i will make sure that the plugs are the correct ones.

 

Anything else i should check?

[Fairtax4me]

The knock sensor is designed to only respond to a certain frequency. Anytime detonation happens it is always the same frequency, just with a different volume depending on severity. So smacking the knock sensor with a screwdriver handle probably wont reproduce the same frequency sound wave that the sensor is tuned to listen for. The type of knock its looking for is more like a hammer striking the block. Though it still may not respond even to that.

 

If you entirely remove the sensor from the block, hang it up where it isnt toucjing anything metal, then it should be "deaf". It shouldn't respond to knock even if it occurs. If it still pulls timing while the sensor is deaf, that (to me at least) implies a problem with the sensor. Especially when the problem is NOT present with a "fake" sensor, a resistor, installed in place of the sensor.

 

 

Seafoam is a good thing to try. Lots of people use it with the engine warm. I think it works better with the engine cold, since you can let it soak in longer without it evaporating.

 

Other than Seafoam and other fuel/induction cleaners, make sure all the hoses for the breather/PCV system are intact and connected properly.

 

Check any vacuum hoses for splits and loose fit.

 

Make sure the air and fuel filters are clean.

[presslab]

Hmm, could be a knock sensor issue, but sounds like it's leaning out.  What are your fuel trims at after driving for a while?  I'm going to guess PCV valve or a cracked hose.  If you know someone with a smoke machine it makes finding leaks easy.

[amsponr]

I did replace the PVC valve and i thought that would make a difference with the oil consumption.

 

Didn't help. (Maybe oil rings stuck?) (Don't know.)

 

The next chance i get i will check all hoses because is keeps coming up.

 

I don,t know anyone with a smoke machine but have looked into making one.

 

My monitor gives fuel info, but honestly i don't understand what its doing.

 

Can you inform me? What trims am i looking for?

 

My wife drove it today with the resistor in place, (Its Her car) and she said its still running great.  

[presslab]

Oil consumption you say? Hmm, that can effectively lower the octane of the fuel. Did you try premium fuel to see if it helped?

 

There are a few different fuel trims. Short Term Fuel Trim, which is what the car learns quickly since you started the car, and Long Term which it learns over many minutes and remembers. Ideally they should be zero percent. Short term will move around somewhat, that's okay. +/-20% is bad. Also there are separate trims for idle.