Idling problem 96 ej22

[Inane Cathode]

First of all, hi!

 

Ok, that's out of the way. I've got a good one for you guys. I've got a fluctuating idle. Once the temp gauge hits just below 150 the engine starts slowing down until it almost dies then revs up to 1200rpm, then slows all the way down again, revs all the way back up.

 

I've done ALOT of testing on it. Ask away i'll provide any data you need. Heres what i think is important:

 

Only starts doing it once its over 140-150*f

It ONLY does it if the throttle is totally closed

The o2 sensor voltage reads .8v at all times only when the throttle is closed, slightly open and it starts fluctuating as it should

O2 sensor responds to a pulled vacuum line as it should, drops to .1volts and bounces around

Timing light reveals the timing retarding in notches (i think its retarding, marks move to the right) then as its about to die it jumps back to where it was

MAF sensor works fine

No intake leaks, or vacuum leaks (blew an entire cigarette through the IACV valve hose with the throttle and exhaust plugs up, no leaks

IACV has been cleaned, and two different valves substituted

Computer has been reset umpteen times

No check engine light, no codes

Runs absolutely perfectly otherwise, other than a very light surging when you're holding speed with the throttle

Power across the harness seems fine.

No grounding issues that my v-drop tests have revealed

TPS apparently seems to be working fine. Sweeps fine, reads closed throttle as it should

New ECT sensor

Fuel pressure is fine, pulling the vacuum line to the regulator makes the pressure jump as it should

 

I'm sure i'm forgetting things but i really really need outside help with this. It feels like i'm out here on my own with this problem.

 

Incidentally it's actually a 96 ej22 and engine management transplanted into a 1972 beetle.

 

I almost feel like offering a reward for the correct fix :/

 

Halp! :-\

[porcupine73]

Hello and just wanted to give a quick welcome to the board. It looks like you've tried a lot of things. The first thing I woudl've been thinking is IACV cleaning but looks like you had that covered. Does she have EGR (I think she does; my '96 2.2L Legacy is an AT and it does). I think I've heard people describe this happening sometimes if the EGR port is all carboned up.

 

 

Not sure if you mentioned it, are there any codes/is the CEL/MIL on?

[porcupine73]

 

Interesting, that is what the ECU will do if it thinks it is detecting knock. Have you tried a different knock sensor and/or looked at it for any cracks?

 

Here's a knock sensor with a nice crack in it. They are known to go bad on that era.

[subie/legacy]

Hello and just wanted to give a quick welcome to the board. It looks like you've tried a lot of things. The first thing I woudl've been thinking is IACV cleaning but looks like you had that covered. Does she have EGR (I think she does; my '96 2.2L Legacy is an AT and it does). I think I've heard people describe this happening sometimes if the EGR port is all carboned up.

 

Not sure if you mentioned it, are there any codes/is the CEL/MIL on?

 

Only the EJ22s that came with automatic transmission will have EGR..

 

TO OP: is it possible that you have a failing/clogged catalytic converter ?

[Inane Cathode]

Subie: I put a vacuum gauge on it, vacuum gauge reads normal. Doesnt reveal a plugged cat (vac would slowly come up to zero at held revs, plus driveability) and in fact it doesnt show a vacuum leak, funky valves, or wildly out of time ignition, so no, no plugged cat.

 

Porc: That's interesting you go to the knock sensor, for the first few weeks i was getting a persistent knock sensor code. I kinda just ignored it and figured it's got cracks and got some coolant in it and needed to dry out. After driving for quite a while i just decided to replace the sensor all together. The old one had cracks all over it, obviously bad. New one fixed the CEL, but didnt fix the idling. The timing thing though, i really dont know noone i have talked to has been able to tell me if its retarding the timing like that because it's in error or because the motor is simply spinning slower. Certainly too-far retarded timing would definitely show up as pig rich on the o2 sensor but not neccesarily cause a misfire (yet). But why would it just sit there retarding the timing only in idle loop?

 

Does anyone have any magical xray glasses so i can see into the ecu case and figure out why it's acting the way it is? Uhg.

[Inane Cathode]

Waaaaait just a second. I have an idea.

 

The knock sensor is just a piezo element, right? Very very sensitive, very low voltages. How about this:

 

Damaged or otherwise disabled shielding on the knock sensor sensing wire leads to erroneous voltage signals to the ecu, telling the computer the engine is knocking when it isnt!

 

Hmmmmm, i think i'll probe the knock sensor wire and see if any voltage pops up.

[porcupine73]

Waaaaait just a second. I have an idea.

 

The knock sensor is just a piezo element, right? Very very sensitive, very low voltages. How about this:

 

Damaged or otherwise disabled shielding on the knock sensor sensing wire leads to erroneous voltage signals to the ecu, telling the computer the engine is knocking when it isnt!

 

Hmmmmm, i think i'll probe the knock sensor wire and see if any voltage pops up.

 

Yes I believe it is a piezo type element; I believe it is possibly an accelerometer but am not sure. I don't know that using a voltmeter would really tell you much since it is an ac type waveform, and there is still a waveform present with the engine running. The ECU I would imagine is employing some sort of DSP on the waveform to determine whether or not knock is present. Then if there is knock present it starts retarding timing until it sees it go away. Then it gets into this cycle of advancing and retarding.

[Inane Cathode]

I ran new shielded wire to the knock sensor, didnt do anything.

 

I put the dvom on the stupid o2 sensors, determined that there was a voltage drop from the first o2 sensor at the sensor and the wire at the computer plug, so i ran a new line there too.

 

Now it's all ************ed up, im not getting any sensible voltages from either of the o2 sensors. I have no idea what i'm supposed to be finding here. I'm so freaking frustrated at this stupid thing i'm ready to throw it of a FREAKING cliff. Three weeks on the same damn problem and i've gotten absolutely nowhere, not even a hint of whats going wrong.

[renob123]

I have no advice for you, but I'm subscribing in the hope that when you do find what it was, you'll let us know.

 

Jacob

[jp260z]

I had a Toyota 22re do that and it was an air pocket in the coolant. The coolant temp sensor must have been getting hit w/ waves of coolant followed by air pockets and it was causing a radically fluctuating idle just like you are describing. Problem disappearded when I got the air out.

 

I also ran across that condition on my brother in law's 92 Accord and found his coolant level to be 3/4 gallon low (probably same result, coolant temp sensor getting air then coolant then air), he had a bad headgasket as it turned out.

 

Have you checked your cooling system??

 

Jay

[Inane Cathode]

I had a Toyota 22re do that and it was an air pocket in the coolant. The coolant temp sensor must have been getting hit w/ waves of coolant followed by air pockets and it was causing a radically fluctuating idle just like you are describing. Problem disappearded when I got the air out.

 

I also ran across that condition on my brother in law's 92 Accord and found his coolant level to be 3/4 gallon low (probably same result, coolant temp sensor getting air then coolant then air), he had a bad headgasket as it turned out.

 

Have you checked your cooling system??

 

Jay

 

Yup, air free. Might be a huge cooling circuit but it's well bled

[GeneralDisorder]

Doesn't the OBD-II advanced code reader tell you when/if there is knock detected? Seems like a possible way to check the signal from it..... definitely sounds like a plausible cause of the idleing problem.

 

This sounds very much like what's going on with the frankenmotor swap renob123 and myself did. It's hunting at idle something fierce and it definitely could be knock related since it's a high compression build and is running the relatively primitive OBD-I ECU. Once you get away from idle it runs perfect and there are no codes present so we have been unable thus far to figure out what it's issue is. Our next step is going to be to run a diagnostic using the Subaru Select monitor interface and a laptop but we haven't got a chance yet.

 

GD

[renob123]

Here's a video of the ghost that keeps messing with my idle. Is this anything like your issue?

 

 

Jacob

[hankosolder2]

Well, if you think the problem might be due to the knock sensor pulling timing, you could do one of a couple of things. One is to just install a fake out resistor in lieu of the knock sensor for testing purposes. You can just measure the resistance of the knock sensor to ground and substitute a similar value resistor.

 

(of course, you'll have no knock sensing, but just to see if the idle stabilizes...)

 

If this solves it, you might be able to build a voltage divider circuit using two resistors to dial down the sensitivity of the knock sensor. (One in parallel with it and one in series.) Given that the sensor is near the bellhousing and you've got your block bolted to a VW transaxle (?) perhaps some other mechanical noise is being mis-read as knock. Just guessing.

 

Nathan

[hankosolder2]

This sounds very much like what's going on with the frankenmotor swap....

GD

 

GD- not to thread jack, but in a nutshell, what do you think of the frankenmotor?

[Inane Cathode]

Fixed!

 

A user on another forum told me to reset the base-line idle. You unplug the connectors to the IAC and the TPS, pull off the IAC hose, then plug the hose and the fitting on the intake tract, then you adjust the throttle stop screw until you get an idle speed of 750 or so, then you hook everything back up, key on to check the signal out of the tps (supposed to be .6 to .7 volts) then it should be good.

 

I did all that, started it up, and whaddayaknow it jumped up to 1800rpm, then came back down to 1k, then just below at 850 or so... no dipping, no revving back up, stayed right there. Drove it around for a while and the idle was hanging really high, around 2000-2500. So, i went under the seat and unhooked the neutral switch (some ecus need it grounded, apparently mine isnt one of them). That made it run -perfect-. Clutch it in out of high rpm, drops like a rock to 1200, then settles down to 850 and stays there, purring away. No light surging, no burbling at decell, actually alot smoother off idle (it was kinda snatchy before)

 

I like to figure out why problems happened after the fact, helps me learn so here goes:

 

The idle wasnt ever to be right even if i corrected it; leaky air intake tract, slight vacuum leaks at the iacv and elsewhere

 

The idle stop wasn't corrected for the lack of a PCV system, small 'leak' as it would be its still air that wasnt there when it should have been according to the computer

 

The neutral switch being hooked up along with the vss 'tricked' the idle into being correct, for some reason the ecu added a few hundred rpm onto the idle speed whenever it was hooked up. Set the idle so it 'idles' and drive it, the neutral switch forces the rpm up too high. Set it so the rpm does what it should at idle and it almost dies when you clutch it, as well as bouncy idle.

 

Of course, hind sight i should have started with resetting the idle, in fact i did actually at one point, but with the small vacuum leaks and not unplugging the tps when i did so lead to the computer compensating as if the throttle was opened so it wasnt actually in idle loop (along with the neutral sensor deal).

 

Lemons out of lemonade (or however its supposed to be) all the spurious testing and re-wiring has fixed or prevented alot of issues. Gave me a chance to check the fuel pressure, and give the entire car a very thorough checking it might have not gotten had things just worked out of the box.

 

Thanks everyone SO much for the help, i'd be a frustrated tangled up ball of gear head alone on this issue

 

 

Epilogue: If you've got idling issues, or you have a swapped motor, or some kind of custom setup, it'll behoove you to verify the baseline idle. If the throttle plate isnt open as much as it should be, or if the engine is getting more air than it should AT IDLE! You really should check and make sure it's correct. Unplug the tps, unplug the iac, unhook the iac hose, plug everything up, and make sure it's 'idling' at around 750 rpm with /no/ computer input besides what air its getting, and what fuel its getting by default. If you leave the tps sensor plugged in, when you wongleflute with the throttle stop it might go into open-throttle mode and it'll 'run away' and sound like its idling right.

[jp260z]

Glad you got it worked out. Sometimes the simplest solutions are the toughest to arrive at....

 

Jay

[porcupine73]

Glad it's working. Do you think maybe the throttle was misadjusted so it wasn't completely closed at idle?

[Inane Cathode]

I think actually the problem was it was too far closed, or some combination of messed up settings all around.

 

It's still got some funkiness, but nothing i cant live with. Starting it, it hangs up at 2k a little too long for my comfort. It does come down but a bit slowly. It'll also hang up around 1500-2000 if you put it in neutral while the car is moving, or clutch it, unless you come to a stop twice and let it come back down. Then its right as rain. Odd, but i can deal with it. I might wongleflute with the vss and neutral switch settings.

[porcupine73]

Too far closed? You mean so closed it started opening in the other direction? Because at idle there should be zero -no- air going through the throttle. The ECU modules the idle air control valve for air at idle. If there is air coming through the throttle (or other vacuum leak) that might mess up the ECU's attempts to control the idle.

 

Some of those era engines can develop odd running/idling conditions if the battery is unhooked for too short a period of time. Like it ends up with random values so to speak in some memory locations. So something easy to try would be unhooking it overnight.

 

They will idle high initially until the emissions stuff gets warmed up, it wants to get to closed loop as soon as possible. The other idle issues sound like possible neutral position switch issues, like it thinking it's in neutral and isn't or the other way around.

[Inane Cathode]

Disreguard, its still broken

 

It might 'idle' alright but the idle hangs WAY too high. You have to stop twice, sometimes, other times not, it'll stick up above 2000, it'll bounce around up there, it isnt right.

 

Damn it i thought i had it. The throttle plate is obviously open too far. When it is set where it should be it isnt running correctly.

 

When the throttle plate is totally closed it does the bouncy idle thing. I dunno, maybe i should just get used to it. Noone seems to have the right answer :/

[GeneralDisorder]

You need to inspect all the data streams as you are trying to do in your other thread about cables/software, etc. It's obviously something that the ECU isn't detecting so you are going to have to check the data from the sensors yourself and determine the problem.

 

GD

[Inane Cathode]

FIXED!

 

Actually fixed now. Dig this, this is a good one.

 

So in another forum, someone mentioned something about there being a 'starter' input to the computer. Battery voltage while the engine is cranking, and mentioned that if its unhooked the engine is harder to start. Interesting, because mine starts right up and i dont remember doing anything with that wire.

 

Well, in a fit of mechanically inclined boredom i ventured out to find that wire, find the VW solenoid wire, and reunite the two. Maybe it'll start a little easier (sometimes it cranks a little long, i think theres something wacky with how the vw key switch works with how the subaru ecu wants it to work).

 

So i traced the wire out of the computer to find it ALREADY CONNECTED! Not only already connected, but already connected to 12V ALWAYS ON! Hmm, not right! So, i unhooked it there, stripped back the solenoid wire from the car, connected the two, started the car and VOILA! Idling perfectly.

 

I was a bit suspicious, so i went around back and dicked with stuff to try to get the idle to shoot up like it normally does. I opened and closed the deck lid (the lid pushes on part of the intake hose) normally this makes the idle drop a bit and the motor races up to 1500-2000 rpm then settles back down. this time, however, it dropped for a split second then continued to idle where it should. Then i put the index on the IACV back where it should be, again it dropped for a second then climbed back up to where it should be evenly, and idled just like it should. I put it in gear, let the clutch out as if to stall it, and it just goes back to idling like it should.

 

Woohoooooo!

 

So, still some issues though. I'll continue with this thread, it should be pretty easy to track down compared to the last problem. Ahem.

 

So, when i start the car, it'll start then immediately race up to 2000rpm, then sloooooowly come back down to idle. If i take off before it settles back down and i clutch it or put it in neutral the idle will fall to 1800 or 1500 and just stay there. Obviously, this makes going down hills kind of a pain in the rump roast.

 

What i've done temporarily is disconnect the VSS so the computer thinks the car is sitting still. It hadnt thrown a code yet (it probably will throw a p0505 if it does) but at least right now it goes back to an idle so it doesnt stick when i'm driving.

 

So, is that normal? I've got the PCV system hooked up, everything seems to be ok (god knows all those checks/replacements i've done) so why do you think it's doing this?