6 ignition coils in 9 months- 2000 ob 2.5 mt!!! Please help

[SubieDrew]

Hi CNY_Dave... that is my biggest concern... the dealer throwing parts at it and not pinpointing the answer either...

plain OM Checked with my mechanic- he totally understands the ignition analyzer thing. Thinks it's a good idea, but he doesn't have one.

The multimeter I have has a dwell setting for conventional coil- not sure how I would be able to use it for the distrubutor-less ignition since I can't get the probes to contact those wires with the engine running.

It also has instructions for testing coil switching action.. by reading RPM. but again, without access to the wires I'm not sure how to test- unless i strip back the insulation or figure out a way to connect a small jumper wire. instructions say "connect red test lead to the TACH signal wire going from the DIS module to the vehicle engine computer. then connect black lead to any engine ground. view reading while engine is cranking" I'm not sure that this will tell us what we need to know. I think this test is more for a no start situation.

Funny you mention the cold spray- I was tempted to do that to the plug connector with some dust-off I have...

Oh well. Monday I'll bring it back to my guy. Checking the plugs- installing new ones with the correct plug number, checking the connection at the 4-pin connector and putting in the new coil. fingers crossed. Then well see what happens.

Maybe I should go to the scrap yards and pull as many ignition coils as I can. And keep them in a box in the back of the car

Thanks again for all your thoughts, and insight guys.

[SubieDrew]

Last week I re-checked all the continuity back to the ECU. All was good. Can't rule out a short to another wire... but everything looked ok.

Took the car to my mechanics on Monday to have him put in new plugs, wires and ignition coil- just incase something was amiss on that side of the ignition coil. As logic should have ruled... seems like everything was ok

Just got back from my mechanic. Not good news. Cleared the snow off the car at my mechanics, started the car- sounded good, went in and paid, came out...........
burning ignition coil!!!

Here's what he put in: new plugs (NGK BKR6E-11), new wires (NGK FX58- mechanic's choice), and a new subaru/diamond ignition coil.

Is there any part of the system that would tell the ECU to fire the ignitor more than it should- perhaps in cold weather???

So really I think this is the end... again. Not really any further along- I'm thinking maybe the ECU is messed up... but not sure if it's worth the $60 (used ECU) and $35 (used ignition coil) gamble.

Thoughts? Anyone want a nice project car?

[plain OM]

. . . .burning ignition coil!!!

What was burning? Was the coil hot, showing signs of melting, smoking? All in just a few minutes?

If the coil was hot, showing signs of melting, or smoking (or similar over-heating) then the problem could well be the ECM as we discussed above. (I take it the car was really cold if the snow had to be cleared off the parked car.)

Is the 1-2 secondary winding now open circuit or the resistance significantly off what it was/should be?

I'm going to stretch this to say that the ECM is faulty. It's causing too much current to flow through the primary of the 1-2 coil, or more precisely, it's turning on the current for too long a time (or never turning it off completely). The primary is designed to handle a pulsing on-off current. But if the ECM is allowing the current to stay on for too long, the coil primary will overheat. While this doesn't directly explain the secondary side failures, it might be because of the heat generated, causing physical expansion and breaking of the fine wire secondary.

Is there any part of the system that would tell the ECU to fire the ignitor more than it should- perhaps in cold weather???

The ECM controls the timing of the spark, and this can be different when the car is first started cold, than after it's running at normal temperature. But this control affects all four cylinders, not just 1-2.

The ECM turns on the igniter transistor for, say 1-2, and this causes current to flow through the primary of the 1-2 coil. The current will not reach maximum instantaneously -- it will increase to maximum over a short time -- a few milliseconds. Then, when the current just reaches the maximum stable level, and the ECM wants the 1-2 plugs to fire, it turns off the current. Without going into a long explanation of this, the effect of shutting off the primary side current is to cause the secondary side to develop very high voltage between the 1 and 2 towers. This is conveyed to the plug where it jumps the gaps and fires one of the two cylinders.

When cold, it's possible that the ECM turns on the current for a slightly longer time than when hot, because the current build up could be slower, for example if the battery voltage is low. The critical thing, again, is to get the primary current up to its stable level, but, and this is important, not allow it to"dwell" at the stable current for too long.

This sort of brings it back to more measurements on the primary side, perhaps at the 4 pin connector to the coil. I think you did measure voltages there with the key on, engine off and there wasn't anything unusual. But I presume that was with the ECM at a reasonable temperature. The question is what happens when it gets cold, and the engine is started? Any idea how cold it might have been when you started it today?

I appreciate the frustration: there's been a few threads here lately where a number of us have been toughing it out against very elusive gremlins. But one now seems to have been drawn out of hiding and will be addressed; another still needs some verification. Sometimes it does take a lot of steps, especially when working at a distance, but there has to be something, and it's just a matter of sticking it out till it's found.

If your mechanic is unable to scope or measure the signals coming from the 1-2 and 3-4 ECM wires and compare them for obvious differences (waveshape, voltages), then perhaps the investment in a used ECM (from the identical car, to avoid introducing any other variables) is worth it. Another shop that might be able to do this type of diagnostics is another possibility, but that would probably involve a minimum charge as well.

If the car is otherwise fine, you like it, and would want to keep it for a good time yet, then why not give it a chance? It's still far less than the cost of replacing it.

[CNY_Dave]

If it has not yet been on an ignition analyzer scope, I think you haven't done the single best test yet.

Compare the coil output that does not cook to the coil that does, there almost *has* to be a visible difference.

It should really be a cheap test as well- 30 or 60 minutes shop time, max.

[Sackett]

Of all the issues I have seen, you need a (any) shop that can hook up a scope to the coil and ecu. you need to see all values; voltage and time on/off to the coil. I think this will answer your questions.
I am leaning to the ecu also, but that is a wild ass guess.

[SubieDrew]

Plain OM, CNY_Dave and Sackett (welcome to the coil conundrum- thanks for chiming in). Sorry for the delayed reply- I didn't get the car back until yesterday PM. Drove it the mile or so to my house under the power of a couple cylinders- while the coil pack smoked

So first and foremost- yes I totally agree that it needs to go to a shop who can do the proper ignition diagnosis. Just couldn't drive it far- and would have needed to get it towed- so thought I'd give my mechanic one more shot. Well- that didn't work out.

However, I do think that the issue is getting more focused. Here's what more I know now.
Plain OM- the temps were not crazy cold, upper 20's maybe. We got about 3-4 inches of snow on Tuesday PM/Weds AM. So when I got to it on Weds PM it needed to be cleared off. So maybe it's one of the factors, but hard to say.

It was smoking from the front/left (as you face it) side of the coil pack- under the cyl 1 tower. I really think that it is the ignitor that is getting too much voltage. You can see a bulge in the side of the pack right at that area. The cyl 1+2 coil was feeling hotter than cyl 3+4.

But here's something odd? the only time the coil pack has smoked is when a brand new coil has been put in. It hasn't smoked when a used coil pack was installed...

And more interesting- the secondaries both read 11.28 so the secondary on this coil pack hasn't yet been shorted. The spec says 12.8 +/- 15% which would be as low as 10.88. So looks like that is ok...Points to ignitor issue issue or primary...

I've tried to get any reading on the primaries on this style coil pack but never been able to get any reading- I believe you said you can't actually check the primaries on these coil packs with integrated ignitors, Plain OM- can anyone confirm that? I know the FSM says you can and has a very "helpful" illustration that shows how to do it- but I've never gotten a good continuity reading on know good or failed coils.

So to the point of testing the input from the ECU to the coil pack while the engine is running myself. Since getting to a shop may be challenging. To to this I would need to get my tester on wire one and ground to see the wave, is that correct? I suppose I could try to get a small wire in from the back of the connector to the coil pack and connect to that and to ground to watch the fluctuation in voltage. Or remove a small part of the wire insulation...

The I/O chart I have says it should be 1- 3.4V. Plain OM you said a digital multi-meter will show an average? I do have an analog meter too.

My digital does have a "dwell" setting- maybe that would give a reading in "degrees" that I could compare to cyl3-4?

Also I have the spark plugs that were in there. They have about 14,600 miles on them. See photos- Cylinder 1 plug is darker than the other three for sure... that's indicating a rich mixture on that cylinder, right? Maybe trying to compensate for a bad signal to that cylinder? or a stuck injector?


I agree- getting this fixed sure seems cheaper than getting a different vehicle- new or used. But once this is resolved there are the other typical issue with a 12 year old subie- external head gasket seep, external seeping valve cover gaskets (not inside spark plug tubes), oil pan gasket leak, center differential needs fixing...

But there are so many memories in this car as we've had it since it was new.. and it should go past 200K easy... but the wife has lost patience with the car and my obsession It's a balancing act for sure But heck for $100 dollars at the scrap yard I could swap the ECU and a used coil pack... almost cheaper than getting it to the shop (flawed logic and throwing more parts at it, I know- but this is such an evil temptress).

Only other thought would be a hidden short way up in the firewall before the wire come into the engine compartment. Maybe just run dedicated wires straight to the coil pack from the ECU?

Curious to hear your thoughts.. It sits in the garage until next steps...
Thanks again for all your help and logic.

[CNY_Dave]

Huh, I would have expected 2 sooted plugs, not one.

[SubieDrew]

Hi CNY_Dave,
I know, Right? weird...
So does that point more to injector for the soot issue- not so much for the coil issue? Although AA1car.com does say that "repeated coil failures on rare occasion can be caused by a lean fuel condition (dirty injectors, vacuum leak, or EGR valve leak)" - but there is NO EGR on this car so that's ruled out

If the #1 injector is messed up wouldn't it try to over compensate? either by trying to pump more fuel or retard the spark?

I do have the good ol' P0420 code recurring. Maybe cyl 1 is dumping too much fuel killing or killed the cat... but I digress.

[plain OM]

I believe you said you can't actually check the primaries on these coil packs with integrated ignitors, . .

See the first attached diagram. The primary of the coils have one end on the 12 V supply, which is on pin 2 of the 4-pin connector. (Could be a yellow wire.) Coming down through the coil, we end up at the collector of a transistor switch. (This is probably more than a simple transistor -- perhaps a mosfet or darlington.) The resistance between the collector and emitter (lower connection) of a transistor is normally open circuit when the transistor isn't being turned on. The connection between the coil and transistor is embedded inside the coil assembly, and therefore is not accessible.

If the transistor switch shorts, there might be continuity between the 12 V supply pin and pin 3, which is the ground. But if there were a short in the transistor switch, it's not likely there would be any controlled sparking; in other words, cylinders 1 and 2 would have massive misfires.

Nevertheless, you could try measuring resistance between pin 2 and pin 3 on the coil using the digital meter. Do it with the positive (red) test lead at #2 and the black at #3, and then reverse the leads. The resistance readings should be very high in both. (In this case, the measurement is of both the 1-2 and 3-4 igniter/coils.)

You could also measure the resistance between pin 1 to 3 and again pin 4 to 3, measuring one way and then reversing the test leads as above.

The problem is that the coil might show differences between the 1-2 and 3-4 sides, but is this due to the coil or external factors that have caused the difference. It's highly unlikely that all the coils would have failed in exactly the same way unless something is causing it, and that has to be outside.

The I/O chart I have says it should be 1- 3.4V. Plain OM you said a digital multi-meter will show an average?

Yes, most digital meters will indicate an average for a voltage that is varying. In this case, the signal from the ECM is probably a voltage that switches on and off between zero to something like 2 - 5 Volts. If the on and off times are equal, and when on it's at 5 V, the meter would likely read around 2.5. However, the voltage probably isn't a clean on-off (there could be pulses) and it's probably not equal time on and off. Each meter design will react to these more complex signals differently, so it's hard to say exactly what your meter should read. But comparing 1-2 to 3-4 might reveal whether there is a significant difference.

My digital does have a "dwell" setting- maybe that would give a reading in "degrees" that I could compare to cyl3-4?

There's no spec on the "dwell" (degrees) in these electronic control systems, but 1-2 comparison to 3-4 could also be indicative, regardless of what the actual reading is. Worth a try. But check the meter's owners manual to be sure the input impedance when in the "dwell" function is high, in order to avoid inadvertently loading the ECM. (Typical digitals have a 10 mOhm input on Voltage measurement ranges, but I don't know about the dwell ranges.)

Then again, if the cause is "active" only when the car (ECM?) is cold, getting valid/useful measurements might be a challenge. But it's all worth a try.

[SubieDrew]

Well, sorry to be off the post for a couple months. Needed to attend to other pressing issues in life. The car has been sitting in my garage until I have the time to do something with it. Still need to find a mechanic who can test the ignition system.

But in the mean time- I finally got a chance to do some resistance readings on the two most recent ignition coils as per your advice Plain OM. My multi meter is an Actron CP7677 with decent ohm range- and a diode/continuity setting that give and audible tone and a numeric value. Actron®

Here's what I found testing both coil packs each pin combination both ways (red-black, black to red):
Failed New Coil in Car JAN 2013
Pin 2 12v-Pin 3 Ground
Pin 2 red - Pin 3 black 1.9Ω
Pin 2 black - Pin 3 red 1.9Ω

Pin 1 cyl1+2 -Pin 3 Ground
Pin 1 red - Pin 3 black infinite
Pin 1 black - Pin 3 red infinite

Pin 4 cyl3+4 -Pin 3 Ground
Pin 4 red - Pin 3 black 1952 on diode setting/ infinite on 200Ω setting
Pin 4 black - Pin 3 red 736 on diode setting/ infinite on 200Ω setting

Across Plug wire towers
1+2 11.53kΩ both direction
3+4 11.71kΩ both directions


Failed Used Coil taken out in DEC 2012
Pin 2 12v-Pin 3 Ground
Pin 2 red - Pin 3 black infinite
Pin 2 black - Pin 3 red .981MΩ

Pin 1 cyl1+2 -Pin 3 Ground
Pin 1 red - Pin 3 black 1877 on diode setting/ infinite on 200Ω setting
Pin 1 black - Pin 3 red 575 on diode setting/ 181.3 kΩ

Pin 4 cyl3+4 -Pin 3 Ground
Pin 4 red - Pin 3 black 1872 on diode setting/ infinite on Ω setting
Pin 4 black - Pin 3 red 568 on diode setting/ 170.2kΩ

Across Plug Wire Towers
1+2 infinite both direction
3+4 12.31kΩ both directions

Ok so if I'm understanding this info (at least a little bit)... these two coils failed in different ways.
The failed new coil has a short in the transistor switch and a failure in the primary coil for Cyl 1+2.
The failed used coil only failed in the secondary for Cyl 1+2.

Plain OM can you chime in with your thoughts? Does this help to narrow down anything?

I'm going to try to test voltage next-
can I try to crank over the engine with the 4 pin connector to the coild pack unplugged?
Will I be able to get some reading from pin 1-3 and 3-4 respectively? or will the ECU not send a good pulse with that disconnected?

I'm sure it would be better to test with it plugged in and start the car- I can get some small wires in from the back of the plug to jumper into the wires. Would you think I only need to check voltage pulse from pin1-pin3 and pin3-pin4?
Do I need to check the 12V part of the system?

One concern with this method of testing is the coil in the car isn't operating properly already... might that in turn tell the ECU to send a different signal to cyl 1+2?

Oh yeah and once it's running- the coil will probably start to smoke so this has to be a quick test
Thanks in advance for any thoughts and help. You guys have been great.