Best of Deals Car Reviews Repair Shops Cars A-Z Radio Show

Could a transmission fluid leak cause O2 sensor failure?

Here’s the background.

About a year ago, the Check Engine Light came on in our 2000 Montana. It’s got well over 260,000 on it, so that’s not exactly a big surprise when it happens. Pulling the codes and checking the live data, everything pointed to a simple failure of the B1S2 O2 sensor - the one that sits after the cat, sticking up about halfway back under the vehicle. It’s by far the easier of the two to change on this beast, and it took all of 5 minutes to swap it out. I figured that the sensor, which may have been the original one from the factory, had just finally worn out. (I’d never replaced it before, as far as I could recall.).

About a month later, the CEL was back, and the codes and live data were both showing indications of a bad sensor again. Same symptoms as the first time; the sensor output was at a steady 1.105 V with the sensor connected (which is what the sensor would read if the engine was running so dead rich as to use all of the O2 in the intake stream) and the voltage in the live data dropped to 0 if I unplugged the sensor from the harness. I checked for heater voltage at the plug (it was normal; system voltage across the two pins) and I further tested the harness for shorts - of which I found none. This didn’t preclude the possibility of an intermittent short, of course, but I saw no indication of one when testing, nor did I find any place where the harness looked like it might have been pinched at any time.

Failing to find any other issue, I just chalked it up to “random bad replacement part” and swapped it out again. It was a Bosch sensor that had been installed as a replacement. I’ve never had a defect in one of theirs in the past, and was very surprised that this one seemed to have failed. It took me another couple of months to get around to replacing the failed unit, during which no new codes cropped up, and nothing presented itself as an obvious source of the problem.

Two weeks later, the second replacement was doing the same thing as the first one (and the original, for that matter), and since I was flat broke and out of ideas, and not on board to simply keep assassinating expensive sensors by setting them up like tin cans in a shooting gallery, I put the whole thing on Ignore for a while. Since the downstream sensor is only an emissions inspection thing anyway, and has no bearing on the engine’s operation, ignoring it only produced a financial hazard and not a mechanical one. (And I’ve been lucky so far on the now-expired sticker.)

Anyhow, when I finally went back to the problem last month, knowing that I had a bad sensor in there, I rechecked everything before swapping in a third replacement. And I verified that it was working; the voltages read correctly, and the switching speed and range looked normal. So I drove it. For about three days. And the sensor output first started going intermittently 1.105V steady, and then fixed at 1.105V steady; failed AGAIN.

Now, I have not applied a drop of non-sensor-safe RTV on this thing anywhere, nor did I apply any added anti-seize beyond what Bosch supplied on the threads - and the upstream sensor has soldiered along without a hiccup through the whole debacle. If there was a problem with a coolant leak or contaminated fuel, or excessive oil consumption, or anything of that nature, I’d think that anything in the exhaust stream bad enough to eat three post-converter sensors would have popped the upstream at least once as well. And it hasn’t.

Which brings me to the only thing I can find that might, just possibly, maybe be the cause of the repeated failures. There has been a steadily increasing transmission fluid leak from a source that I finally located today; the o-rings in the snap-together connectors at the radiator weren’t sealing anymore. The airflow under the vehicle has carried the leaking fluid back as far as a couple of feet past the cat under the car, and there’s evidence of it on the heat shield above the cat and the sensor - but the exhaust, and the sensor, which both get really hot in use, show no sign of this contamination. Be that as it may, trans fluid has got lots of Stuff in it, including some dissolved metallic compounds, and I’m starting to wonder if the fumes from the trans fluid burning off of the converter might have been killing the O2 sensors. The upstream sensor is in a location where it would never get a whiff of that stench, which would explain why that one hasn’t failed. But searching the Net has failed to find even a single reference to this being a potential issue, and as often as trans fluid leaks happen, i would have thought that if this was a real possibility, somebody would have run into it and posted a heads-up somewhere. But as far as I can tell, nobody has.

So here’s the poser: Can trans fluid fumes, or trans fluid itself, kill an O2 sensor just by getting to it on the outside? Anybody know for sure?

Please post the exact code you’re getting

Not the interpretation, just the code . . . P0130, for example

Thank you

The only code is P0138, which is indicating exactly what was observed; voltage high from sensor 2 bank 1; that’s the sign of a fouled O2 sensor that is no longer able to register oxygen from one side of the element. (Usually, this is caused by the exhaust stream side being fouled, while the atmosphere side is OK - but it would do the same thing if the fouling is solely on the outer section. That is something I have verified.) If the voltage had been excessive, as would have been the case for a short to heater supply voltage, there’s a different code to indicate a sensor circuit problem - but this is the only code I’ve got. The captured live data from the CPU at the time of the code getting set also reflects that the voltage was at exactly 1.105V, which is fouled-sensor data. As noted, the upstream sensor is working just fine, so a source of the failures that is internal to the engine seems really unlikely. That’s why my poser was so specific; I’ve narrowed this down a lot, and the only potential cause that seems to be present is one for which I can’t find anything documented that would confirm that it is possibly relevant, and while my knowledge of the sensors and the fluid seems to say that this is probably it, I’d like to find out if that’s known to be a killer before I blow another $40 on a fourth replacement. Dead ones don’t even make interesting Christmas ornaments - not that we bother with a tree, having 8 cats.

I can’t see any possible way it could get to and harm the sensing surface, but if it were to manage to find its way onto the plug contacts it could theoretically by creating a contact resistance affect the signal sensed by the ECU. I’m really, really out on a limb with this theory, but if a good flushing with contact cleaner (without lubricant) solve the problem, well… it’s a cheap and easy experiment anyway.

On the connector; been there, tried that. And trans fluid isn’t electrically conductive enough to foul the signal - but if it were causing a loss of current flow to the CPU via a near-short across the sensor leads, the result would be a reduction in the measured voltage anyway. That’s not happening; the voltage is maxed, ergo no short there. If the conduction was acting to couple voltage across from the heater circuit, the chances of getting a result of precisely 1.105V seem too far-fetched to even worry about - but the plug and socket have both been cleaned repeatedly to no effect, so that seems at least as unlikely. What we’ve got is the max voltage output that the sensor can deliver - which these sensors only produce when there’s some O2 in contact with the reactive surface at one end of the element but none at the other. And according to several tech diagrams, there is supposed to be a small duct from the area of the cable ingress point down into the body of the sensor, designed to be resistant to water intrusion but present to allow the contact with atmosphere that the sensor apparently can’t work without. Trans fluid is extremely invasive; it will spread via capillary action over nearly any surface it contacts. That alone makes me think fluid contamination of the atmosphere side of the element is likely to be the culprit, but as stated at the top level, I’d expect this to be a known issue somewhere if that’s what is going on - and there isn’t any mention of such an issue anywhere that I can find. Ditto for trans-fluid-burnoff-fume-generated poisoning of the atmosphere side of the element. One factor that makes me particularly suspect fluid intrusion is that the sensor location is toward the vertical here, with the duct presumably pointed up at the same roughly 45 degree angle as the sensor. In that orientation, the chances of the fluid getting into the unit and on to the sensor element are high enough, if the duct actually exists. (In examining one of the failed ones with a magnifying glass, however, I can’t see that opening; either it’s tiny enough to be easily plugged, or the clever folks at Bosch found a way to do without it.)

I probably need to get an engineer involved to answer this question.

Thinking out of the box, does the 2000 Montana use vacuum modulation to control shifting? If so, is it possible that the vacuum modulator diaphragm is leaking, allowing ingestion of the tranny fluid?

I did not miss you comments that any erroneous exhaust component should also contaminate the upstream sensor, and I agree, but I thought I’d post the thought anyway. Leave no stone unturned, as it were.

No vacuum modulator. Not looking forward to trying the replace-sensor-and-see-what-happens gambit, but I’m getting low on confidence that I can skate along on this expired sticker much longer. It the replacement will work for three days, though, maybe I can buy a year’s worth of sticker to ignore the problem for a while with.

Question: Either you didn’t mention it, or I missed where you did, but after you pulled the old sensors out, did you test them and make sure they’re actually bad?

Is it possible that coolant is leaking into the EGR valve mounted on the upper intake manifold, and this coolant is dripping down the EGR tube into the exhaust pipe after the upstream O2 sensor?

It takes very little coolant to damage an O2 sensor.


Testing the sensor definitively off-car is difficult, but yes, with some ingenuity I borrowed a compatible working vehicle to swap one of them into - and the sensor immediately came up bad.

On this 3.4L, the EGR tube hits the manifold well to the driver’s side, and there’s nothing leaking coolant anywhere near it. If coolant were running down it, the destination would be the steering rack. The tube’s not cracked, either, as far as I can see. (I had it off a while back.) But the only coolant seepage this vehicle has had lately was from a bypass hose at the opposite corner of the engine, well away from anything that could have allowed it to be sucked into the engine, and that was only enough of a leak to lose about half a teaspoon a month. I know because it habitually accumulated in a small depression, and I could watch that depression fill slowly over the course of a month. Like a lot of 3.4L GM mills, this one had intake gasket failure at about two-thirds of the current mileage, and they’ve been replaced with the more robust design from Fel-Pro; no significant coolant loss has been noticed since. And I run non-silicate coolant.

Oxygen sensors do have a vent to the outside world.
That’s how they get oxygen on one side of the active element vs exhaust gas on the other side.
It’s possible that trans fluid or the fumes from burning it on the exhaust system are getting into the sensor and contaminating it.

Articles that talk about O2 sensor vents:

My suspicions are aligning with @circuitsmith. Either that, or the transmission leak and the repeated O2 sensor failures are caused by separate problems.

Can you post the long term and short term fuel trims when the fault code sets?

Do you have a way to watch the rear O2 sensor activity from a cold start?

Circuitsmith, that was what I had suspected. I’ve already ordered up another sensor now that the leak is fixed, and if that one survives, I will post the results back here for public reference. The only reason I could see for the observed behavior was contamination on the atmosphere side, but the specific issue I had just wasn’t given as a possibility anywhere - and I’m much happier when I can find confirmation of a plausible but hard-to-verify guess about something like this. I’ve watched the rear sensor from a cold start; the reading goes direct to 1.105V as soon as it heats. It’s hard to distinguish from “immediately”, as my tool has a slight delay in establishing communications after engine start, but I’ve seen it go from 0 to 1.105V in one step as fast as I got the live data stream up. I haven’t tried building a tap harness to check it directly.

And there’s a fairly important clue in that second link of Circuitsmith’s, where it discusses the fact that the atmosphere vent is via the wires themselves; the plug at the end of the sensor’s harness had trans fluid on it, and as invasive as that stuff is, I have little doubt that it had wended its way into the wiring. I’ve thoroughly sprayed out the vehicle’s harness plug with residueless contact cleaner now, in preparation for the next replacement sensor. At some point when I have nothing better to do, I may try swapping a new plug onto one of the “failed” sensors’ harnesses to see if they mysteriously begin to work once vented again.

asesmith - I don’t have that data saved at the moment, sorry. It will doubtless set another P0138 before the replacement sensor shows up, though, and if I have time, I’ll post the captured snapshot of those two back here.

I’d be interested to see the rear sensor activity from a cold start, as the car warms up the sensor should switch almost identical to the upstream sensor and then start to flatline (preferably low) as the catalyst lights off. The rear sensor should still respond to goosing the throttle a few times.

A flatline high is sometimes indicative of a rich running engine, perhaps not enough to set an active code yet, but fuel trim numbers will tell more.

FWIW, if I were working on that car the sensor installed would be AC Delco or Denso.