E36 Fuel Gauge Stuck; first on half tank, then on empty

[HighandFast]

The referenced car is a 1997 E36 M3. Because I didn't find what I was looking for when I did my search when I got stuck, and I know somebody else is going to search this forum for the Peake codes below, I give you this:

If you get E3 and E4, which look like this:

E3 Oxy sensor adaption limit, Cyl #1-3;
E4 Oxy sensor adaption limit, Cyl #4-6;

your onboard diagnostics are telling you that the fuel/air mixture is out of the range the O₂ sensors can detect. That is, the fuel/air ratio is so rich or lean that you need to be looking for either 1) a problem with the fuel system, including an empty tank, or 2) a vacuum leak.

In my case, and totally understandably, I was also getting these two codes (and why not for the other cylinders? Dunno.):

EE Misfire detected, Cyl #1
F0 Misfire detected, Cyl #3

As it turned out, I was out of gas. Period. That simple. Why I didn’t know I was out of gas is not so simple.

The back story, if you’re interested, follows. Note that at the end there is a diagnosis of a problem with the fuel gauge:

Unbeknownst to me, though I had a suspicion, my fuel gauge was stuck at just above a 1/2 tank. I don't drive the car enough to know what it should read, unlike with my daily driver. And, because I disconnect the battery when not in use, I don't have the trip meter, which I reset with every fill-up, to rely on to know that laws of physics seem to be being defied by my car.

It all came to a head this morning. About a mile from my house I floored it to make a green light and it bogged down. Not a low-fuel sputter, but a vacuum leak kind of sluggish feel. That seemed to resolve itself and for the next ~8 miles of highway driving things seemed fine. Then they went bad again. And a half mile from my destination (11 mi from where it first bogged down) it sputtered and died while I was going up a pretty long hill after exiting the hwy. I coasted to a stop at the crest of the hill. After sitting for about 30 sec, I restarted it and it sort-of ran for 500 ft and it died just as I was turning into the parking lot. I got it running again long enough to go 20 ft to a parking spot, where the car sat for the next 30 minutes.

When I got back to the car from my appointment, I did a scan for failure codes. I keep a Peake scan tool with me (what, you don't??) and it said what I said at the beginning, E3, E4, EE, F0.

And that's all it said. What I didn't know for sure was that these codes were saying, especially in the absence of other codes, that there was something wrong with the fuel/air being supplied that was beyond the O₂ sensors' ability detect. As any good DIYer would do, I took the irrational action of disconnecting both O₂ sensors — because I was at a loss for anything rational to do — and tried starting the engine. It idled rough, but eventually smoothed out, so I took off toward home. One mile and three stalls later, I coasted into the parking lot of an Exxon (and Midas). Mind you, the fuel gauge was reading a little more than 1/2 full. Based on nothing, I decided it was a vacuum leak, called a tow truck, and $127 later it was in my garage.

When I got in the car at home, and to my surprise, the gauge read empty. And when I turned the ignition to ACC the low-fuel light came on! Adding fuel got the car running normally, but the gauge was now stuck on empty.

Gauge troubleshooting: I got bamboozled by the stuck fuel gauge. I now importantly know that, if there is a break in the connection to the fuel level sensors, the gauge will stick at whatever level it was at when the break occurred. However, it will eventually jump to empty (regardless of whether there is fuel) because the resistance at that sensor is now infinity (open circuit). How long it will stay at the last correct reading before showing empty is variable. Knowing this characteristic of the sensor system would not have saved me, but I now know that an open circuit in the fuel level sensor system will likely not be immediately apparent.

Normally when testing the sensors' resistances, you should get readings of between 10 ± 2 and 250 ± 5 ohms (empty and full, respectively), and both sides should read about the same. (Test the sensors by removing the white connectors from the tops of the pumps.) When I measured mine after adding gas and driving a mile, I got readings of 158 Ω and 172 Ω — pretty close and within the margin of error — so I knew it wasn't the sensors that were causing the gauge to read empty.

Then I found a break in a wire where I had spliced it while converting to a two-pump system some 5 yrs ago. When I reconnected it, the gauge showed 1/2 full. Then I disconnected to see what happened and it stayed at a half, even after cycling the ignition, but it would eventually go to empty. I fixed the bad splice and now the gauge is reading normally (actually, a bit high, but I'm confident that will straighten itself out).

The most painful, $127, lesson I learned was (remember, I was at a gas station when I called the tow truck) if the car exhibits fuel starvation characteristics, the first step in diagnosis should be to put a gallon or two of gas it the tank...no matter what the gauge says!!

[LuxoM3]

How many miles are on your car? I *think* at around 180k miles my fuel gauge (95 M3) went dead. I ended up replacing the fuel pump AND fuel filter. Was about $225 a few years back.

In your case, sounds like you aren't getting fuel. Would explain a your codes pretty easily.

[HighandFast]

In your case, sounds like you aren't getting fuel. Would explain a your codes pretty easily.

You think? LOL.

BTW, it's got right about the same miles as yours did, though, trust me, that is statistically insignificant.

[bmwdirtracer]

Hi my friend,
At the April event at V.I.R., going over the hump on the back straight, my M5 spit and coughed. Quick check of fuel: half a tank. Half way down the front straight, it did the same, then seemed to genuinely lose power going up the esses.

I came into the pits of course, and hooked the car up to a good diag computer (really....no paper charts... ) No codes.

Went to live data, and the scan tool showed that the fuel pump side of the tank was at 75%, the other side at 10%. Should be fine. Just to check everything again, I hooked the car to ISTA, and Abel (328 Power 04) connected remotely from Texas, to look over all the data. It all looked good, other than that the non-fuel pump side of the tank still had so little fuel.

I filled the tank with the absurdly expensive race-track gas, and VOILA, all was fixed. Turns out the instrument cluster doesn't know which side the fuel pump's on.....and, probably, my passive "sucking jet pump" has come adrift, and isn't sending fuel to the electric pump.

Of course, with two pumps, you've fixed that. But, like you, I'm going to keep my tank topped up, on track!

Missed you at the spring event!