Something is pulling back ignition timing, not knock sensor

[Fairtax4me]

This one has me kinda baffled.

I made it two days with a "new" ECU with all I/M monitors set (green light) and everything working fine. Today, car runs like crap.

Drove the car across town to meet friends for lunch, car ran fine. Leaving the restaurant, the same old weird intermittent power issue (that I've dealt with for what seems like years now) once again reared its ugly head. Same MO as before, very light throttle produces no problems, medium throttle (about 15% or more) the car has no guts. Won't get out of its own way, lots of hills around here and it really had a hard time getting up some of them today. Put the gas pedal down hard and it bogs bad until it gets to about 3000 RPM then all of a sudden it kicks and the car takes off like nothing's wrong. Shift gears and the fun starts all over again.

 

Pulled the scan tool out of the door pocket and plugged it in to see how things look, and the big thing that stands out is the timing advance. Wih easy throttle it stay up around 30-35 degrees. A little more throttle and it drops back, way back, usually to around only 10 degrees. Calculated load is only about 25 - 30%. Throttle position about the same.

Knock sensor I'm thinking, so I get out the spare, hook it up, no change.

Thinking now that maybe it's actually pinging and the ECU is pulling timing for a good reason. So I hatch a plan, I build a jumper wire and bolt it on the knock sensor, bolt the other end to the block. Now the sensor is grounded, but its not actually touching the block, it's pulled off to the side away from everything so there's (theoretically at least) no way for it to pick up knock. My thinking here is, the ECU still has a clean signal from the sensor, so shouldn't go into fail safe, and with no chance of picking up the pinging from the block, the ECU won't know about it and will keep timing up where it's supposed to be. If its actually pinging, I should hear it pretty clearly. Right?

 

Drive up the road, no pinging, no CEL, ECU is still pulling timing back to sometimes as low as 6 degrees. No change at all in the symptoms.

 

Now I'm trying to figure out, what else does the ECU use to pull timing?

I know its dependent on load, throttle position, and knock. If its not picking up knock, why does the timing come and go as it pleases? And the whole problem comes and goes as it pleases? It's there today, it might be gone tomorrow.

Does anybody know for sure which other sensors play a part in pulling back ignition timing? Crank sensor? Cam sensor?

 

Things I've checked/verified while this problem is occurring:

ECT is good: usually around 190 degrees. Problem has occurred when engine is either cold or warm.

Map values are now good (thanks to the new ECU).

Fuel pressure does NOT waver during power loss problem. At idle pressure sits around 35 psi, and climbs to 40+ under throttle.

Fuel trims are within -5 to +5%. Usually -5 to 0. Drops no lower than -10% when coasting in gear. Tends to stay around 0 to +5% during acceleration.

O2 sensor voltages are good.

New spark plugs and wires about three weeks ago.

MAF sensor gets cleaned every 3 months. Have replaced MAF in the past during troubleshooting of this problem with known good spare and have no change.

Knock sensor clean, and replaced with known good spare, no change.

ECT, have no reason to suspect because it reports the correct temp, and doesn't change, has been replaced in the past with known good spare.

Fuel filter changed about 6 months ago, air filter 2-3 months ago.

No vacuum leaks or split hoses.

I know there are other things but I'm having a hard time remembering them all right now.

 

The last time this issue popped up in a noticeable way was about a month ago, which finally prompted me to replace the ECU. The problem has followed along to the replacement. Has to be something else on the car that I haven't replaced yet.

Igniter, cam and crank sensors, oil pressure sensor, fuel temp sensor. Those are about the only things I haven't swapped.

[1 Lucky Texan]

the driveability symptoms kinda fit with a TPS issue - but I admit i don't know what that would do to the spark timing.

[hankosolder2]

I'm wondering if it still IS indirectly knock sensor related. Say you had a bad engine ground or something... with the knock sensor physically floated but electrically grounded to the block, that bad ground noise could work its way into the ECU and cause it to pull timing. (This is purely theoretical speculation.) You could try the same experiment but use a different ground point for the sensor.

 

You could float all the knock sensor wiring at the ECU and install a fakeout resistor between the knock sensor pin and ECU ground of a value equal to what the knock sensor measures with an ohmmeter. I somehow remember someone saying 180K ohms, but I could be wrong.

 

The ECU itself could have an internal fault. Unlikely, but.... It must have an A/D converter to process the knock sensor signal... that could be noisy, have a bad solder connection, etc.

 

 

Nathan

[Ricearu]

it should be 570 ohm, but don't do this, the ecu will run up the timing to find it's limit, and It will never find it, therefore pinging like hell. ask me how I know.

 

also, a partially plugged cat caused my forester to do similar things, but when it would bog down, it would lurch and bog, lurch and bog about 5 times before it would finally get the hint and take off. new cat bolted on and problem solved.

 

You could also put a vaccum guage on it and see if the vaccum drops a crap ton when it's bogging, which would indicate a crapped out cat.

Edited December 18, 2012 by Ricearu

[naru]

Seems to me that an inaccurate cam or crank sensor would have to result in inaccurate timing.

 

Have another motor to compare timing?

Might be normal.

[Fairtax4me]

Have another not entirely drive-able car. I could patch it together and go up the street a few times but it would be very loud.

 

also, a partially plugged cat caused my forester to do similar things, but when it would bog down, it would lurch and bog, lurch and bog about 5 times before it would finally get the hint and take off. new cat bolted on and problem solved.

This sounds pretty close to what it's doing. Thing is, it's there one day gone the next, and its been doing this for as long as I can remember. I think the cat would have plugged up entirely by now.

Have a vacuum gauge, it does drop considerably when the throttle is opened. Same with the MAP reading on the scanner. Drops to something like 5-6" when under heavy throttle. But it doesn't seem to jump around or waver in unison with the power problem. Seems to be pretty steady even when the car isn't acting up. Not to mention, engine still pulls pretty hard at higher RPM.

I'm wondering if it still IS indirectly knock sensor related. Say you had a bad engine ground or something... with the knock sensor physically floated but electrically grounded to the block, that bad ground noise could work its way into the ECU and cause it to pull timing. (This is purely theoretical speculation.) You could try the same experiment but use a different ground point for the sensor.

It's all theoretical at this point, but that's an idea worth trying.

Resistor I hadn't thought of, but I expected the engine to ping like hell after moving the sensor off of it, and it didn't. It didn't ping at all. I have a friend with muchos of spare resistors laying around, I'll get him to rig something up for me and try that too.

 

TPS I get a smooth increase when looking at values on the scanner, but I haven't checked it with an ohmmeter yet. Did replace it a while ago, like a year and a half, for gits and shiggles. I'll put that one on the list for testing.

The other things on the list are replace the igniter (have two spares, might as well, not like its hard to get to), check cam and crank sensor signals on my buddies o-scope, possibly swap in and compare spares if we have time.

Edited December 18, 2012 by Fairtax4me

[naru]

Does the high rpm power surge coincide w/a sudden timing increase?

Does the O2 sensor signal change significantly at the surge point?

 

RPM is important for the ECU timing calculation.

 

Intermittently plugged exhaust sounds promising.

Vacuum gauge won`t always catch it.Please don`t ask how I know.

Might be worth tapping in a pressure gauge ahead of the cat(or remove the O2) at this point.

Edited December 18, 2012 by naru

[idosubaru]

very light throttle produces no problems, medium throttle (about 15% or more) the car has no guts. Won't get out of its own way, lots of hills around here and it really had a hard time getting up some of them today. Put the gas pedal down hard and it bogs bad until it gets to about 3000 RPM then all of a sudden it kicks and the car takes off like nothing's wrong. Shift gears and the fun starts all over again.

i had similar symptoms on my daily driver 1997 OBS 4EAT years ago. could not make it up steep inclines either. light throttle it was fine, i could get to and front work except for one particularly steep switchback incline was really annoying to get up, like almost impossible limping it along.

 

i chased and swapped a few things. i read online that it could be EGR related. i replaced the valve and solenoid to no avail. i finally changed the small round filter inline with a vacuum hose that goes to the passengers side strut tower area - it's about the size of a quarter and costs a few bucks from Subaru. unbelievably (i'd be skeptical of this story myself, nor have i seen it since) the car drove like new. i'm not sure if it was the filter or a vacuum line i inadvertently "fixed" while replacing the filter and find it odd i've never seen anyone talk about it...i sold that car to a friend who is still driving it just fine many years later.

 

i doubt that's your issue but it's at least easy enough to look into.

[naru]

The rpm stuff does make it sound EGRsh.

Like it opens too far,but,closes properly.

[Cougar]

Like Naru, I was thinking of the crank sensor also. Perhaps the signal gets noisey and causes the problem to occur.

[presslab]

At 25% load if it's anything like my '92 it will not pull back timing regardless of how much it's knocking. There is no timing to pull in the max advance map at that load. Furthermore the only place on the map that will even go as low as 6 degrees BTDC is max load at < 1200 RPM, so I have doubts about that number as well.

 

Perhaps the ECU is getting a bad signal from a sensor which is causing it to pull back the timing. Like mentioned before, the cam and crank sensor come to mind. I don't think the ECU uses the TPS for timing, it uses engine load which it calculates from the MAF and RPM.

 

What is the IAM (ignition advance multiplier)? That right there will tell you if the ECU thinks it's knocking. When you say fuel trims do you mean short term fuel trim? What is the long term at? Have you hooked a WBo2 sensor up to verify the mixture?

 

If you dump your ROM I can probably find the timing maps for you. You can do this with any old serial SSM cable and JECScan.

[ivans imports]

had a crank senser get punched by a ball berring from a coged idler pully damaged the reluctor gear and the crank senser made the car not run and backfire. Towed in no spark but had injector pulse

[Speedwagon]

I just replaced the cam sensor on my '98, which was causing issues between 3-4k rpm that always felt like it was pulling timing to me. Swapping the cam sensor with a junkyard one has, so far, made the problem go away.

[Fairtax4me]

No little filter do-hickey, took it off the car a long time ago chasing MAP issues. Suppose I could put it back.

No EGR.

 

Presslab I was actually looking at the timing charts you posted in another thread, I think they're for a 2.5, but they can't be too different than what a 2.2 uses. I can't see IAM with my scanner, only the overall spark advance.

When it dropped to 6 degrees that was at about 2200 rpm, 50% load.

I can record live data with my scanner, but for some reason they made the data connector 9 pin serial, which my laptop doesn't have. I haven't felt like paying for a serial to USB adapter just so I can transfer data from the scanner to the laptop.

Anyway, I'm gonna take a closer look at load and timing values today and see if they seem any different than yesterday.

[presslab]

The stock timing maps I posted in that thread were out of a 1992 EJ22E ECU, "F9" label. As far as I know, the ECU will never pull back the timing below the "Base Timing" shown on the left. Under knock, it will pull back the "Max Advance" on the right.

 

http://www.ultimatesubaru.org/forum/showthread.php?t=134517

 

USB serial adapters are cheap online.

 

http://www.monoprice.com/products/product.asp?c_id=103&cp_id=10311&cs_id=1031104&p_id=3726&seq=1&format=2

 

I built my USB SSM1 cable for about $35. This works with EvoScan, JECScan, etc.

[Fairtax4me]

Crank and cam sensor signals looks good, but we couldn't take the car for a drive and monitor signals at the same time.

It acted stupid for about 5 minutes when I first left for my buddy's house this afternoon, then it cleared up and drove just fine after that.

Timing stayed up where it should be according to that chart, which should be pretty close I would think. 4 years difference but basically the same engine.

It actually seemed to run better today than it has in several weeks, but that could just be in my mind.

 

Hooked up the O-scope and we got nice clean signals from both sensors, but whatever the cause of the weirdness corrected itself so this test may be useless.

I have pics of the scope monitor that show the signals really well (as well as I could get a picture of at least).

 

I did dig up my spare sensors and ignitor, so if it starts acting strange again they only take a minute or so to swap, I can pull over and pop one in to see what it does.

[idosubaru]

you my friend are the man - you used an o-scope - wow!

 

have you swapped the coolant temperature sensor used by the ECU?

what sensors have you not replaced?

 

Have you considered doing the following two steps:

1. swap an entire intake manifold

2. get another body side wiring harness and stuff it around a strut tower for a test? buy one stripped and ready or get your own. you can likely resell it for swap material.

 

that would nearly rule everything out.

[Fairtax4me]

you my friend are the man - you used an o-scope - wow!

I didn't use it, he did. There were so many knobs and switches and dials I had no clue what all was going on. I plugged the paper clips into the harness connector and hooked the alligator clips he handed me. He turnedthis and that then flipped a switch and it slowly buzzed to life. After about 30 seconds for the tubes to warm up he said OK fire it up, I hit the key, and there we had a nice wavy line on the screen with a pattern that looked just like the crank sensor output. 1 2 1 2 1. Revved the engine to about 2500 rpm as he adjusted some knobs to keep the display stable. The only difference we saw was the peaks grew taller (voltage increased).

 

Speaking of the crank sensor. There was some oddness after I removed the probes from the connector. The engine didn't want to start. Got codes for the crank sensor. So I unplugged it, then plugged it back in, then it started fine. I might pull the sockets out of the connector and make sure they each fit tightly on the pins on the sensor just in case there's a loose connection there.

 

I did remove both sensors, and found a very small amount of either glue or corrosion on one pin on the cam sensor. Couldn't tell exactly but it was only a little dot right in the middle of the pin. Scraped it off and checked the socket side and didn't see anything on there.

 

He didnt have the right resistors to make up 570k ohm to try the knock sensor removal idea. At least not without stringing like 15 of them together. Gonna hit up the rat shack tomorrow and see what they have and come up with something.

[Fairtax4me]

Got cam sensor pics too. Was going to put them in the previous post but the uploader names them all image.jpg and just replaces the previously uploaded pic.

 

Anyway, was a bunch of fun seeing the signals display on the screen. I had hopes to see one flutter or have a strange signal, no dice.

[Cougar]

The signals displayed on the ol' Tektronix scope look good. Since the trouble wasn't happening at the time it still doesn't eliminate them as the source of the trouble. If you have some spares it may be best to just swap the original ones out and see if that helps.

[Fairtax4me]

I built my trick sensor today and put it on. Knock sensor measured at 580k Ohm, the false sensor ended up at 560k Ohm which is close enough that the computer doesn't care.

Car still isn't acting up, so I won't really know if it works until ??? Left the knock sensor plugged in for now. Really the only reason for this is to make sure I don't have two bum knock sensors. Don't feel like ordering a new one just to find out it isn't the cause. I'm still convinced its something else, but it never happens continuously for long enough to really tell if replacing this or that one thing at a time actually fixes it.

 

Oh well, it's running fine for the time being and no CEL. Can't ask for much more.

[cal_look_zero]

Been following this thread a bit. I pulled up my timing advance and am having it pulled back to as far as 6. I replaced my ignitor with a junkyard one this morning and never saw it go further back than 12. Seems like there may be a few things that all need to align in order for the cars to run right.

 

Frustrating...

[Fairtax4me]

A little more detail about the timing differences I'm seeing on the scanner.

When its being insolent timing is usually in the 12-18 degree range while trying to accelerate from about 2000 rpm. As rpms increase timing stays in about the same range, but it will sometimes jump around. Then when engine speed gets to about 3000-3500 rpm it will suddenly jump up to the 30-35 degree range and the car takes off. This happens under "normal" acceleration, and engine load is only in the 20-30% range, or only slightly more. If I put the pedal down hard load will increase into the 35-45% range, and this is when I saw the very low timing mentioned before, but it was at 2000 rpm.

 

When its acting normal timing doesn't usually go lower than 20 degrees unless RPMs are very low, like under 1500. Then it may be down around 12-15 degrees, but it rises steadily with the increase in RPM and doesn't jump around any. I've seen it as high as about 40 degrees when accelerating very lightly. With normal acceleration it'll be around 20-25 degrees and increases steadily as rpm increases.

[1 Lucky Texan]

pardon me if i haven't kept up with details, but, could your replacement ECU be from a later year model and is expecting variable valve lift to kick in?

[Fairtax4me]

ECU came from a 97 L. Same problem was occurring with the old ECU.