So a few months ago the CEL on my 2004 Outback came on. I took it to a part
s store, had them run the codes, and it said my primary O2 sensor was bad.
The 300 price tag for the part at the store was more than I could stomach,
so I passed intending to find one online. I finally got around to buying on
e at Rockauto for about 110 bucks about a month ago.
I went to install it, and realized I had two primary O2 sensors. the code d
oesn't distinguish. I figured I'd try one, and if it didn't solve it, it sw
itch the new one to the other side and put the first-removed O2 sensor back
The new part had the wrong plug. It was the standard 4 wire plug vice the 6
wire plug that my car uses. So I figured I'd just solder it together. I go
t out the tools, clipped the wires and started stripping the insulation. Uh
oh. That looks like either aluminum or stainless. Both of which are a pain
A little flux and I had some questionable solder joints. Felt solid, but th
ey didn't look like the solder penetrated very far. I put on the car and it
ran a lot worse than it had before. It had surging at medium throttle open
ings when cold-that's indicative of a non-working O2 sensor.
I had to leave for two weeks on business during which time my wife drove th
e car daily. She said it got worse to the point that the idle would fluctua
te like it wanted to die, but never did. When I got back I could hear an ex
So I suspected I had a bad solder joint, and knew I had an exhaust leak. I
checked the solder joints, then crimped the wires any way. No luck. Still s
urged and had horrible mileage. So I put the old part back in, which involv
ed crimping it's original connector back on. The car ran better, but still
a lit CEL.
So at this point I'm back to square one parts wise but also have an exhaust
I got under the car and found it to be one of the bolts on the flange betwe
en the muffler and catalytic converter had rusted to failure. So I cut off
the flange on both sides of the joint, got a 2 inch ID joint from the parts
store, clamped it on, and started it up.
This was the amazing point. The car was noticeably quieter than I remember
it ever being. When I put my head out the door and rev the engine all I hea
rd was sound from the front of the car. Nothing from the exhaust. I suppose
this is normal, but apparently that flange had had a mild leak for a while
that steadily got worse until it suddenly got a lot worse.
I drove it, and at the fourth startup it had no CEL!!! Still on the origina
l O2 sensor.
So I'm drawing the conclusion that the original code wasn't due to a bad O2
sensor, but was because of an exhaust leak between the CC and the muffler.
My theory for the mechanism here is that the ECU determines primary O2 sens
or performance based on what it expects to see in certain scenarios like th
rottle lift or certain throttle settings, and the exhaust leak was reducing
back pressure enough that it was creating a different environment where th
e happenings in the exhaust manifold were different enough from expected th
at the ECU didn't think the O2 sensor was up to snuff. So fixing the leak g
ot the performance of the gasses in the exhaust back to where they should b
e, so the ECU no longer thought the sensor was sending bum info.
Just a theory.
The car has better throttle response now. I hadn't thought it had any hesit
ation on tip-in, as I thought it was just the torque converter's laziness,
but now it's so immediate that I think it did have a little hesitation.
Incidentally, the flange bolts and flange were unsalvageable, but the pipe
was fine. Still good wall thickness. The 12 bucks and two hours I spent wer
e far better than what a shop would likely have charged me since they'd pro
bably have replaced both the cat and muffler for tens of times more.