1994 Legacy - good power, stumbles on idle

[Zootal]

1994 Subaru Legacy. FWD. 4dr. Non-turbo. 230,000 miles or so. Normally purrs like a kitten.

 

Symptoms: engine will idle, but miss intermittently, every few seconds or so. Sometimes it's better, sometimes it's worse, sometimes it will stall and sometimes it won't idle at all. When it's acting up, it is hard to start. Has good power otherwise, no noticeable problems at higher engine speed. It's been doing this for weeks, but today it got so bad that it won't idle half the time and now my wife refuses to drive it (I don't blame her).

 

Here is where it gets interesting. With the key on and engine *off*, the tachometer jumps every second or so. It will jump up to 500-1000 rpm point, then drop to zero. The frequency of this is ever second or three, and is the same frequency that it stumbles at idle. The check engine light is *not* on.

 

This eliminates a whole slew of possibilities, and narrows it down to some weird electrical problem. It doesn't get worse when I hit a bump or slam the door (my old chevy vega would stall if I slammed the door, but that is another story).

 

Anyone have experience with this model and recognize these symptoms? I'm not sure what to do other than check the various electrical connections and think about replacing the ecu.

 

The battery is fully charged and is fairly new, it cranks very well. Alternator is fairly new. It charges just fine when the engine runs. AFAIK there are no obvious mechanical problems, nothing makes noise, it normally just purrs and runs fine. Until this started happening. With the engine OFF and key ON, the tachometer jumps, but the lights do not flicker, and no other lights or instruments act up.

[choover1101]

Coil pack and or ignitor? I had an 82 Bronco with some issues from the coil pack back feeding into the tach and stumbling. It ended up burning out the tach and "fixing" itself.

 

Sent from my SAMSUNG-SM-N910A using Tapatalk

[Zootal]

That kinda sorta makes sense, as it would effect the tach and cause it to stumble at idle. I would guess at ignitor, as that would cause the tach to jump even with engine not running. Whatever this is, it acts up without the engine running, meaning there is no HV going through the coil - unless the ignitor is tickling the coils, which is also possible......hmm....

 

Coils are easy to find, I see after market "ignition control module" for about ouch $185 autozone, $40 amazon. I hate to throw parts at this, but that combo does sound like a good place to start...I wonder why autozone wants so much for the part?

 

I think I'm going to push it into my garage and take a closer look at effected parts, maybe something obvious will stand out that I missed before...

 

And hope it's not the ecu cause those are $300 plus **sobs**

Edited December 6, 2018 by Zootal

[Setnev]

Take a look at all your ground wires and the harness that goes to the coil pack. Make sure the grounds are clean and not corroded. Make sure your coil harness wires are in good condition.

[doublechaz]

I doubt it's the ECU.

 

I would also check cam and crank sensors. If they send a bad bit once in a while the ECU could think you have rpms when it's not turning.

 

You could probe the connections on the coil with a test light and see if it is dumping a signal sometimes. You might need near full dark to see it if the signal is too fast to light a bulb, or switch to an LED.

[choover1101]

Test thought. When it is ticking the tach unplug the coil and see if it stops. Just work backward unplugging and replugging to test where the ticking (jumping).

 

Sent from my SAMSUNG-SM-N910A using Tapatalk

[Zootal]

Test thought. When it is ticking the tach unplug the coil and see if it stops. Just work backward unplugging and replugging to test where the ticking (jumping).

 

Yah, I thought of that. I'm hoping it isn't cam/crank sensor, as IIRC you have to pull timing belt covers to get to it. I got finals next week so won't be able to do anything until next weekend, but I'll post my results here when I do get a chance to work on it.

[doublechaz]

Is a 94 that much different from a 95? Both cam and crank sensors can be checked/replaced without touching anything timing related.

 

For me, crank sensor is below the alternator but can be pulled without touching the alternator, and the cam sensor is on the drivers side top by where the dipstick passes behind the timing cover.

 

But I do like choover's method of checking the coil. That should be real quick to do.

[Zootal]

So the plot thickens .

 

It has got to the point where with the key on and engine NOT running, I here this clicking noise from under the hood and the tach jumps. Look around a bit and I realize the fuel injectors are firing along with the tack jumping. I unplug the cam sensor, no change. Unplug the crank sensor, it stops! Remove crank sensor, examine. Has buildup of crud on the bottom, but idk. Plug it back in, and the problem does not start up again. Now I gotta wonder, was it a bad connection at the plug, or is the sensor bad?

 

Shop for sensors. Most after market sensors have the wire connector on the sensor itself. However, this sensor does not - it has about a foot of wiring coming out of it, with a plug on the end.

 

Sensors with plug on sensor, which don't look right - $20-$40. Sensors with the wire attached and a connector at end of wire - $200+. Ouch. Did they put two different sensors on this?

 

And the sensor is under the alternator, no need to pull timing belt. Not sure why book says to pull timing belt...

[Zootal]

Is a 94 that much different from a 95? Both cam and crank sensors can be checked/replaced without touching anything timing related.

 

For me, crank sensor is below the alternator but can be pulled without touching the alternator, and the cam sensor is on the drivers side top by where the dipstick passes behind the timing cover.

 

But I do like choover's method of checking the coil. That should be real quick to do.

 

PS You are absolutely right. The crank sensor is indeed under the alternator. I pulled the alternator cause it takes 3 minutes to pull it and gets it out of the way. And the cam sensor is right were you said, also.

 

My guess is that the crank sensor is sending spurious signals to ecu, which causes it to fire ignition and fuel injectors, which is what was happening, even with engine not running. I took it out, cleaned it, plugged it back in and the problem went away. I don't trust it, however, so I'm replacing it. Weird. Found one for $180, O'Riellys. I'll install it this afternoon and post results.

 

Thanks everyone for your advice and for being a sounding board. I was a mechanic many years ago, but I retired my tools in 1984. Things have changed since then....no more carbs and points to adjust, this car doesn't even have a distributor.

[doublechaz]

I figured with the tach jumping it had to be something confusing the computer. You really cinched it with the injector firing thing.

 

I thought it took steel moving near the sensor to make it send a signal so I don't think crud on it would make it send spurious signals, but that would widen and flatten the signal leading to the computer being less accurate with timing info.

 

I think you are on the right track with not trusting it. That really sucks about the price difference. Even if the sensor guts are the same, I bet the wires wouldn't reach without that pigtail.

[Zootal]

So I replace the crank sensor. Sadly, nothing changes. It sits there intermittently cycling the injectors, the tach jumps, and now the check engine light flashes on and off irregularly - there is no pattern to it, it's not flashing a code, it's just flashing on and off as the injectors rattle and the tach jumps...<sigh>....

 

I know what the problem is not - it's not most of the sensors under the hood because I unplugged them all, and the problem persisted. When I unplug the crank sensor, it stops hence my decision to replace it. But with the new one, it still does the same thing. And I like doublechaz said, it's really unlikely it would send spurious signals and confuse the computer like this, but I had to replace it to be sure. So something else in the crank sensor circuit is confusing the computer, assuming the ecu itself is not bad. I'm beginning to wonder.

 

My next step is find the computer, I think it's under the seat. Unplug it. Examine it. Check the wiring between the computer and the various sensors to make sure there is not a bad connection somewhere (that is gonna take a while). Then I think about replacing the ecu.

 

As always, advice/suggestions welcome....

Edited December 9, 2018 by Zootal

[Zootal]

I'm looking at the manual, and all of these things - ignitor, crank sensor, fuel injectors - go directly into what it calls "MPFI control unit", 1990-1994 2.2L engine. I'm looking at a bad connection to this unit, a bad ground, something wrong with the power feed, something wrong with the wiring between the unit and the crank sensor - or the unit itself, in that order.

 

MPFI - multi port fuel injector - is this thing a multi port system? It's the only diagram for the 1994 2.2L engine module.

[doublechaz]

Are we talking about an OBD II car, or OBD I? Mine is a 95 and OBD II(ish, long story). I'm not sure when the cutoff happened. The OBD II car is definitely multiport. In 1982 they were still carb. I don't know if they jumped from carb straight to multiport or not. Sounds like you are OBD I. My compter is at the passengers feet when stretched out, but I'm not sure on yours.

 

What level of service manual do you have?

[Zootal]

Are we talking about an OBD II car, or OBD I? Mine is a 95 and OBD II(ish, long story). I'm not sure when the cutoff happened. The OBD II car is definitely multiport. In 1982 they were still carb. I don't know if they jumped from carb straight to multiport or not. Sounds like you are OBD I. My compter is at the passengers feet when stretched out, but I'm not sure on yours.

 

What level of service manual do you have?

 

1994, definitely OBD I. They changed something in 1995, not sure what, and in 1996 I think everything went to OBD II. Did Subaru go to OBD II in 1995?

 

I have a Haynes and a Chiltons. Both more-or-less the same AFAIC. The wiring diagrams are fairly generic but for the most part close enough for what I need.

[doublechaz]

95 was OBD II for the protocol and connector, but it doesn't implement all the sensors. That's why I say OBD II - ish.

[Zootal]

Interesting. I guess that is what they changed in 1995. I see a lot of references to this and that for up to 1994 model only.

 

Is the ECU for these things, the one under the driver seat, the same for this model year (1994), or did they use different ones? I'm looking at junk yard units, and they all look the same.

[doublechaz]

I don't know about 94. I have learned about mine because I've had such bad luck with hired mechanics. I put a "parked to rot" 82 GL back on the road, so I know a little there. Otherwise I'm all the way down to generalities.

 

I think the 94 ECU is different from 95, but I'm not completely sure.

[Zootal]

I hear ya. I don't hire mechanics. I was one in the early 1980s (carburetor specialist and dang good at it), and while things have changed since them, they haven't changed all that much. If I can't fix this car, it goes to the scrap yard.

[doublechaz]

Anything can be fixed if it was built in the first place. But can it be fixed for a reasonable amount of work/parts/time. If you had a bunch of spare parts lying around you could get through this easy and cheap. But as it stands, I'm not sure it makes sense.

[choover1101]

Are you using the FSM (Field Service Manual)? I believe one of the stickies at yhe top of the forum has links to them.

 

Sent from my SAMSUNG-SM-N910A using Tapatalk

[Zootal]

I pulled the ECU. Took apart. Nothing obvious wrong. Found a junkyard with an identical model, with same part number. $50. I'm betting at this point 50/50 or better this fixes it. If not, then I get to start checking wiring to/from ECU, but that is really really time consuming. I'd rather risk fifty bucks on an ECU before I start tearing into the wiring. it will be a week or so before I receive it, and I will come back here and post results after I get it and install it. Cross fingers for me!

[doublechaz]

Good luck indeed. This is certainly quicker than chasing wiring. I used a spare computer in testing my big problem a couple years ago. Made for a very quick confirmation that I needed to learn more about how the software actually works, which is pretty hard to do since we can't look at the software directly.

[Zootal]

The SAGA continues :-)

 

Received ECU from junkyard today. Installed. Turn key on, did not start engine. Nice and quiet like it should be (before it would trigger the injectors and the tack would jump, with key on and engine not running). Now it sits there quietly, tach does not jump, injectors do not cycle, cel is steady. This is very promising!

 

Tried to start engine. WHAM. Engine jumps, horrible pounding noise comes out of engine. I only cranked it for a second, it did not start. I stopped cranking immediately when I heard the noise.

 

omg wtf now what.... So, come to find out that the ECU failure caused the injectors to cycle when the engine was not running, and this poured gasoline into the cylinders. I drained the oil pan, and I drained almost 5 1/2 quarts of fluid from it (this engine takes only 4 quarts of oil). Very strong gasoline smell in the oil. It pumped so much gas into a non-running engine, that it caused the engine to hydrolock when I turned it over, and over a quart of gasoline drained into the oilpan.

 

I have not had time to do this, but my plan this weekend is:

 

Pull spark plugs. Turn the engine over by hand and make sure there is no gas in the cylinders. I'm thinking bring each cylinder to the top, and stick a rag into the spark plug hole to soak up any excess gasoline. Then disable ignition and fuel injection, and crank the engine a bit to clear out remaining gasoline from cylinders.

 

Install plugs, put clean oil in engine, replace oil filter while I'm at it, try to start engine.

 

I just hope it didn't bend a rod or otherwise damage the engine.

[doublechaz]

Sounds like the right plan from here. Hard to say if it damaged anything. Partly it's how quickly you got off the starter, and partly luck. Cranking with the injectors electrically disconnected and the plugs out is good to blow out any remaining fuel.