It would be fair to say that we all hate it when we approach a diagnostic problem with preconceived solutions that do not work out, and we then spend days looking for a solution we’d thought would be obvious. Just such a situation arose when this writer was recently invited to consult on a case concerning a fuel level indicator on a 2011 BMW X3 that should have been easy to solve, but wasn’t. This article is about that case and the lessons we can learn from it, but let us start by stating the-
When this writer was called in to consult on the case, the workshop that had received the vehicle (that belonged to a regular client of theirs), was too busy to attend to the vehicle right away, so they obtained the owner’s permission to call in a consultant to speed things up.
When this writer questioned the vehicle owner, he stated that the vehicle belonged to his wife, and he only drove it that day because his wife was complaining about something “weird” going on with the vehicle’s fuel gauge. However, before he could identify the problem or see what she was complaining about, the vehicle suddenly stalled, and it would not start again although it cranked normally. His initial thought was that it had run out of fuel, but apparently, that was not the case, as the fuel gauge showed that the tank was half-full.
This was borne out by the initial inspection, so the obvious thing to do was to check for proper fuel pressure, followed by a comprehensive scan to check for ignition system issues. No ignition codes were found, but a manual check of the fuel pressure showed that there was no pressure in the fuel rail. Therefore, the possible causes of the problem were obvious, the fuel pump had stopped working, or, the fuel filter had clogged up completely, or, equally likely, there was a problem with the fuel pressure control system, although the high-end scan tool that was used initially failed to find relevant trouble codes.
These possible causes made perfect sense at the time, so based on this writer’s assurances that he’d have the X3 on the road again in no time at all, both the workshop’s owner the vehicle owner’s permission was obtained to remove the vehicle to this writer’s workshop, which was granted, but that was also when the trouble started. Fortunately, though, the vehicle's driveshaft was already disconnected to allow the staff of the present workshop to push the vehicle into a bay since the transmission could not be set to "N" if the engine was not running, as is the case with many late-model BMW's.
Nonetheless, and long story short, the vehicle was removed to this writer’s premises without incident, and the following is-
This writer’s resident diagnostician confirmed the total lack of fuel pressure on the N52T engine, but on a hunch, he put the X3 on a hoist so he could more easily perform his famous "bang" test on the fuel tank. Based on his experience that fuel gauges sometimes read inaccurately, he always banged a piece of 50mm PVC piping against fuel tanks on vehicles with suspect fuel gauges.
In this case, it turned out that the fuel tank was empty, despite the fuel gauge indicating a half-full tank. Therefore, the fuel level indicator in the tank was defective and to prove his point, he despatched an apprentice to fetch 20 litres of fuel from the filling station across the road, and to then decant the fuel into the X3. The bang test confirmed that there was now indeed fuel in the tank, and as expected, the BMW started up almost immediately and ran at a perfect idle.
From his position under the vehicle though, the diagnostician could hear that the fuel pump was rather noisy as a result of having run dry, but although a significant quantity of fuel had been added to the tank, the fuel gauge still only read half-full, and the needle had not moved at all after the fuel had been added.
At this point, the problem was clear. The fuel level indicator was demonstrably defective, so based on this assumption and the noisy fuel pump, the diagnostician called up the vehicle’s owner to explain his findings, and obtained approval to replace both the fuel pump and the level indicator*, which was integrated into the fuel pump canister in the fuel tank.
*According to the salesperson at the dealership, the fuel level indicator could not be purchased separately, which later turned out not to be true. Not that it mattered much at that point, since we were going to replace both the pump and the level indicator, but more about this later.
In the meantime, the diagnostician roped in two apprentices to replace the vehicle's driveshaft, and to remove the vehicle's back seat to give him access to the fuel pump. Removing the old pump took most of an hour, so when the replacement pump arrived from the dealership later that day, everything was ready for the diagnostician to install the pump canister himself to ensure that nothing leaked.
However, replacing the fuel level indicator did nothing to fix the problem. The fuel gauge needle still read half-full, and a subsequent scan with a high-end scan tool still did not turn up any fault codes, although the new fuel pump ran quietly- so there was that, at least, which brings us to-
Image source: https://www.autoparts24.at/bmw-x3/tank/1/
This image shows a fuel tank from a BMX X3 with the F25 Chassis, which is what we had in the workshop that day. In this example, the big round hole is where the fuel pump canister fits into the tank, the tube terminating in the light blue fitting is the fuel return hose, and the large white fitting at the bottom right is where the filler hose attaches to the tank. The connector in the centre of the tank immediately to the right of the blue fitting is where the fuel gauges’ wiring connects to the tank.
It is perhaps worth mentioning at this point that the diagnostician was nothing if not a walking encyclopaedia of electrical circuits and circuit testing methods, but in this case, he was stumped. He knew that he could not start testing resistances in the fuel gauge circuits without running the risk of damaging one or more control modules, so instead, he removed the fuel pump from the tank to see if he had missed any connections, or damaged anything during the installation.
He did not find anything that was obviously wrong, but as he said later, he was certain that since the tank had two distinct parts (each part overhanging the drive shaft) each side had to have a dedicated level indicator. He was also convinced that the two level indicators had to be somehow connected to provide a single signal to whatever control module supplied the combined indicators' reference voltage. However, he could not see another level indicator in the tank, and he could not determine where the wire shown below went in the tank, or what its purpose was-
Image source: https://www.searchautoparts.com/sites/www.searchautoparts.com/files/images/Ma1218-09.jpg
As shown here, the wire only just reached the opening where the fuel pump canister went into the tank, but he could not see where and how it disappeared into the tank, even with a strong flashlight and a swivelling mirror. More frustratingly though, there was no attachment point for the connector on the fuel pump canister, so he must have dislodged it somehow during removal and installation of the pump, and now he could not see where it was supposed to go.
This writer had only rarely seen the diagnostician in a flustered state, but he was severely discombobulated when he requested that we contact the dealership to make sure there was not a second level indicator in the tank. We did just that, but the salesperson insisted that his system does not list a second fuel level indicator (in the tank) for the VIN we gave him. He also volunteered the information that he had checked, and could not find TSB that addressed our problem, so the matter was out of his hands.
Clearly, what we needed was specific service and/or repair information, so we bit the bullet, and purchased a rather expensive 2-day subscription from a third-party online resource that included a high-resolution wiring diagram of the fuel gauge and its associated wiring. For various reasons, we cannot reproduce this diagram here, but one highlighted section of the diagram clearly showed a second fuel level indicator on the left side of the fuel tank.
Faced with this information, the salesperson at the dealership called his supervisor, who baldly stated that yes; some models in that particular production run did indeed have a second fuel level indicator in the tank, but his system did not show the second indicator. He also said that the blue wire the diagnostician had found connected to the connector on the inside of the tank, and while it was difficult to reconnect this wire, it could be done by a person with small hands.
So now, we knew there was a second indicator, but what we did not know was a) whether one or both indicators were defective, and b), if we replaced the right-hand side sensor needlessly along with the fuel pump. We also did not know which control module supplied the reference voltage, so we began to delve into our expensive service information to try to figure this out. The news was not good, and although we cannot reproduce the relevant screenshots of the BMW-specific ISTA scan tool* here, we can paraphrase what each page told us. Here is what we learned about-
*The service information we purchased consisted of a series of screenshots taken from a BMW-specific ISTA (Integrated Service Technical Application/Programming) scan tool.
The first thing we learned was that BMW saddle fuel tanks use a complicated system of venturis in the terminal fitting of the return hose. As the returning fuel passes through this fitting, the fuel flow sets up suction in tubes that connect the two halves of the tank in two directions. Thus, as long as fuel flowed through the return line, fuel would be "sucked up" from one half of the tank, to be deposited in the other half to keep the fuel in both halves of the tank at an equal level, which explained the need for a second level indicator in the left half of the tank. Next, we learned about-
Fuel level indicator fault codes
We suspected that our high-end scan tools could not access some of the X3's CAN serial communications systems, and this suspicion was confirmed by our service information that told us that only BMW-compliant scan tools can extract fault codes relating to the instrument cluster. Therefore, since it was late in the day, we waited until the next morning to rent a fully updated BMW-specific ISTA scan tool from a local workshop supply vendor. Here is what it told us-
This tool came with guided tests, so when we got into the relevant system, the scan tool displayed fault code E12C03 – "Fuel level sensor, right short circuit to B+”. It also told us that-
Most importantly though the scan tool told us that depending on the VIN, one of three control modules could supply the level indicator's reference voltage. These included the JBE (Junction Box Electronics) module*, or the REM (Rear Electronics Module), or the BDC (Body Domain Controller). On hybrids, the reference voltage is supplied by the TFE (Hybrid Pressure Refuelling Electronic Control Unit), but since this was not a hybrid, we could safely discount that module.
* On this particular vehicle, it turned out that the JBE (Junction Box Electronics) module controlled and managed all electrical signals to and from the fuel level indicators, before passing a calculated fuel level on to the actual fuel gauge via the CAN bus system.
Since we now thought that we were on the right track to resolving the problem, we returned to our service information to see what it said about actually-
The service information included a section on focused diagnostic procedures, but since this is a long and involved process, we will give the short version here. Essentially, the diagnostic procedure starts by removing the pump from the tank and then disconnecting the fuel level indicator's wiring from the connector on the tank. This is followed by jury-rigging a suitable reference voltage to the level indicator from an external power source, and then checking the signal circuit with an ohmmeter to see if the output voltage changes if the level indicator's arm is manipulated by hand.
If the level indicator works correctly, it should show a resistance of 50 ohms (+/- 5 ohms) when the arm is at its upper limit, and a resistance of 990 ohms (+/- 20 ohms) when the arm is at its lowest position. These values represent maximum and minimum allowable fuel levels in the tank, respectively, and the right-hand level indicator passed the test. In fact, when we reconnected the right-hand level indicator and ran some additional guided tests (on the implicated control module) on the scan tool, the display told us “The right fuel level sensor is OK. No fault can be currently found in the component group tested.” (Italics added)
These tests were repeated on the left-hand fuel level indicator, but this time, we got a reading of 455 ohms, which confirmed that the fault occurred when the tank was about half full. However, despite our best efforts, we could not figure out a way to manipulate the arm of the left-hand indicator because it was completely inaccessible from the only entry point into the tank we had, so we returned to the service information to see what it said about this.
Considering that even though the scan tool had indicated a short to battery positive on the right-hand indicator but that the level indicator demonstrably worked as designed, we had to agree with the service information that the left-hand level indicator was the problem. However, since we could neither test the operation of the fuel gauge itself nor find a way to manipulate the arm of the left-hand indicator, we had to come up with a way to simulate the output signals from both level indicators at the same time.
We needed to do this to verify that the fault was indeed with the left-hand level indicator, despite the contradictory presence of a short circuit code on the right-hand level indicator, which brings us to our-
Decade boxes are also known as sensor simulators, and if they are used correctly, they can be used to simulate the output of almost any sensor on any vehicle. Essentially, the sensor is “spliced” into the wiring of a particular circuit in such a way that the simulator replaces the sensor under investigation.
However, since we had only one simulator in the workshop, we were obliged to purchase a second one to run this test, which, we were not sure would work because according to our service information the code could not be cleared without first fixing the underlying cause of the code, but at this point, we were willing to try anything.
Thus, with the two sensor simulators installed between the level indicators and the correct control module, we set both to 100 ohms (to simulate an almost empty fuel tank) and to our surprise, the needle on the fuel gauge responded immediately. With some experimentation, we could produce almost any reading on the fuel gauge simply by changing the settings on the simulators to any point between 50 ohms and 990 ohms, which confirmed that the left-hand fuel level indicator was indeed defective. It simply had to be the problem, since the right-hand indicator had earlier passed scrutiny by the scan tool- despite the presence of a short circuit code, which brings us to-
Our expensive service information told us that the only available remedy for defective left-hand fuel indicators on BMW X3 models is the replacement of the fuel tank, since the level indicator is incorporated into the tank during the tank’s manufacturing process, and can therefore not be removed from the tank.
According to official BMW sources, there is simply no way to remove and replace this indicator, but since the cost of a new tank ran to several thousand dollars, and the (recommended) labour involved to replace the tank was seven to eight hours, the total bill for the only possible repair was enormous.
We were not sure how the vehicle’s owner was going to take this extremely bad news, so we turned the problem over to the original workshop to explain to the customer. However, before the original workshop called the owner, they sent over their chief mechanic to run through our diagnostic process with us to confirm our findings, which they accepted rather graciously despite our earlier assurances that we would “have the vehicle back on the road in no time at all”- not three days later.
Long story short though; to make things right with the customer, we returned the vehicle to the original workshop, and we gave the new fuel pump and level indicator to the original workshop at cost price. We also waived our not-insignificant diagnostic fee, which waiver we later heard the original workshop had passed along to their customer, which leaves us with this-
We later learned that the final repair bill came to an eye-watering six thousand dollars, which the customer gladly paid, based on his opinion that the dealership would have charged him more for the repair, but it is not for this writer to express an opinion on dealerships' labour charges and parts mark-ups.
Nonetheless, if we had (re)learned a valuable lesson during this affair, the lesson would be not to give assurances on either diagnoses or repair outcomes prematurely. In this case, we had diagnosed a problem and promised a timely fix when we did not have a complete understanding of the problem. Moreover, while the customer was ultimately satisfied with the outcome, he does have some reason to accuse us of unprofessional conduct, and therein lies the biggest lesson of all, which is never to give anyone any reason to attack your professionalism, because this usually costs you money. We clearly failed that test.
As for the fault code on the right-hand fuel level indicator that was in perfect working order? Well, despite extensive research into this issue, we never learned how that happened, or how that was possible. Faulty coding or programming in the JBE (Junction Box Electronics) module seems like the most likely explanation, but we don't know for sure, because the fault code cleared itself and it has not returned, so we will likely never know all the circumstances that re-taught us this very expensive diagnostics lesson.
