Topic: 2cyl intermittent death... WTF

[darthekai]

Occasionally, my car falls on its face and it sounds like its hitting the rev limiter (but its not). At first, if I backed right off the gas, it would go away and I couldnt make it happen on purpose.
Then one day it stuck around for a while. I thought my car was toast, but after sitting parked for a while it started up great and normally.
 it seemed like it would happen at first when my car was either cold or I had turned it off, gone into a store, and came back out. It's gradually been happening more and more often.

I tried to reproduce the effect by unplugging things. The only way I could get it to happen was removing TWO spark plug leads.
Then one day it happened in a parking lot and I was able to do a little more diagnosing. While it was struggling to stay alive (and shaking like a mad man), I removed cyl 1 and 3 wires. No change. I put them back on and removed cyl 2 and 4 wires. Car died as soon as I removed 2, and would not start with no 2 or 4.

I figured it was the coil, because thats what this points to. I ordered a new coil, plug leads, and plugs off rockauto for about $100.
Just replaced them all tonight (the old coil actually was a junkyard takeoff) and IT STILL DOES IT.
Anyone have any sweet ideas?
Was thinking maybe injectors but like... why two at once?
The reason I was thinking injectors is because the AFR gauge reads VERY lean while its sounding like its on the rev limiter. I am pretty sure this is because theres a decent sized exhaust leak in my header but I cant really be sure.
But maybe injectors? Maybe injector wiring?
Maybe FPR? I dunno that doesnt make much sense.
I did take my kickplate off and wiggle the ignition wires and injector wires to zero effect...
Not quite sure how to diagnose btw injectors, injector wiring, FPR, and ignition wiring. No fuel pressure gauge on hand...could maybe splice one in and wait till it happens again and confirm fuel pressure?


PFA

[seat safety switch]

I mean the coilpack is probably 2cyl wasted spark, right? I feel like it's still gotta be ignition. Have you visually checked the coilpack drive harness? Maybe squeeze the wiring and see if there's some crunchy bits or mouse-eaten insulation.

[darthekai]

yeah. I thought the same thing. Theres no way 2cyls at once and 1&3 is a coincidence. But then I thought, would I even recognize the signs of failing injectors or a FPR?

[seat safety switch]

You can flow-test injectors on your bench but if it's an intermittent failure...

An FPR is easier to test (with a fuel pressure gauge plumbed inline) but if you're getting these failures while driving it's pretty hard/irresponsible to try to read that gauge. How much is a replacement FPR? It might be in the "uckfay it" price range.

I don't think injectors on the Subaru ECU are batch fired, are they?

[RockThePylon]

The reason I was thinking injectors is because the AFR gauge reads VERY lean while its sounding like its on the rev limiter.

AFR reads oxygen. No sparky, no burny, lots of oxygen, along with lots of fuel.

[darthekai]

AFR reads oxygen. No sparky, no burny, lots of oxygen, along with lots of fuel.

Now here's something I dont understand.
wideband O2 sensors.
Its a black box for me.
If it only reads oxygen, how does it read anything richer than stoic since all of the oxygen is used up.
You'd think it goes from lean - stoic and maybe a but further but not much.
I started looking at how the actual sensor works, with reference cells and so on, and just decided to classify it as a black box that spits out magic numbers.

[diamondedge]

Now here's something I dont understand.
wideband O2 sensors.
Its a black box for me.
If it only reads oxygen, how does it read anything richer than stoic since all of the oxygen is used up.
You'd think it goes from lean - stoic and maybe a but further but not much.
I started looking at how the actual sensor works, with reference cells and so on, and just decided to classify it as a black box that spits out magic numbers.

Not all the oxygen is all consumed in combustion. Even at richer AFR's you still have some sneaking by.

[darthekai]

So I wasted some company $$$ and sort of learned how they work

I think.

So after reading a bunch, what I can gather is a wideband has a reference cell that essentially is 14.7 product gas (and sealed) and a pump that pumps straight oxygen (? via a membrane?) into or out of the measured exhaust gas. The amount of oxygen that must be pumped to maintain an ORP across the two cells of 0v is proportional to the current used to pump it. This value gives you your "AFR". Sometimes I wish you could set it to like a molar AFR.

So if I had misfires, would this mean the gauge would read lean? I have no idea. Theres way less CO2 in it, way more air, and way more fuel. Probably, because the oxygen is the strongest contributor to the ORP.

[seat safety switch]

So after reading a bunch, what I can gather is a wideband has a reference cell that essentially is 14.7 product gas (and sealed) and a pump that pumps straight oxygen (? via a membrane?) into or out of the measured exhaust gas. The amount of oxygen that must be pumped to maintain an ORP across the two cells of 0v is proportional to the current used to pump it. This value gives you your "AFR". Sometimes I wish you could set it to like a molar AFR.

On narrowband, at least, the reference oxygen comes in through one of the wires, which is hollow. This is why soldering the wires on an O2 sensor breaks it.

[darthekai]

apparently bosch o2 sensors all have a resistor with some very specific value in the plug head that tunes the specific reference cell to output correctly without re-calibrating the whole system.
anyways.
back to my problem.

How the uckfay can I track down an intermittent ignition failure? Without cutting the harness open?
Should I just run new trigger wires and hope for the best?

[darthekai]

well maybe what I'll do is bypass my harness and run a separate wire to the ecu. Cut just after the connector, solder and heat shrink the new cable on.
If that fixes it, I will need to expose the whole harness and check for additional damage.
Any suggestions for 4c automotive cable? I would guess it needs to be shielded as well.