Select a technical image to view it in detail.
N53: Cylinder 6 Shut Off by the DME, Traced to a Rusty Fuel Rail
A brief retrospective of the problem, and how it was solved.
The symptoms
On an N53 engine, the DME switches off cylinder 6 with no apparent reason. The EML (Check Engine) lamp lights up in the KOMBI, and the smell of unburned fuel can be noticed inside the car.
Directly before and after the event, all injector and adaptation data are perfect, and so is the live data. The problem disappears after restarting the engine. The fuel leakage symptom (more precisely, live data confirming an incorrect Lambda) also disappears by itself after 5 to 10 seconds, without a restart. The problem was progressing. Most recently the defect appeared every 10 to 30 km.
Live data shows that the problem occurs when the engine runs in Stratified charge mode, typically in the low to very low torque range:
- Lambda in bank 2 drops from 2.5–3 to 1.2–1.6.
- After 2 to 3 seconds, the DME confirms the fault and switches to Homogeneous mode.
- Lambda in bank 2 then drops to 0.7, the minimum value reported by the DME. The NOXEM reads 0.78, also a minimum.
Ruling out the DME and the injector
With such stable symptoms, I was convinced the DME was to blame. Installing another DME did not solve the problem. At that point I got over my laziness and connected the oscilloscope to check the injector control. The signal analysis showed that the DME drives the injector correctly. I even replaced the injector plug to exclude the connection completely.
I then replaced the injector with a new one, and the problem remained. The main question became: what makes an injector stick half-open in this particular injection strategy? My hypothesis was rust building up inside the rail.
Rust in the rail
This is the rail from the outside:


Unscrewing the pressure sensor revealed the following:

Signs of rust are clearly visible on the thread. They are also visible on the inner wall of the rail’s fuel passage (note the “circle”). This sensor had never been removed since I bought the car, so the rust has obviously been building for a long time.

Here are rust deposits on the sensor itself. The “circle” corresponds to the area filled with fuel, so the rust is inside the rail.
Finally, here is the seat of one of the injectors in the rail:

There are two types of dirt here. The black dirt is soft in texture. It could be paint from the fuel tank or leftovers from the coating of plastic parts. It is not the main culprit, because it does not form in the rail itself. The second type of dirt is rust. The rail rusts from the outside, where oxygen is available for oxidation, and from the inside too, as visible here.
On this basis I decided to buy a new rail.
Result after installing a new rail
After 3000 km, the engine regularly uses Stratified charge and the defect (the jamming injector) has disappeared. Unfortunately, the inside of the rail cannot be inspected, because its bore is only about 3 mm in diameter. Evidently the rail was rusting intensively inside, near the cylinder 6 injector.
Why the defect looked the way it did
This is the control signal the DME uses in the problematic mode:

The DME used three very short injection pulses. During such short pulses the injector cannot open fully, so the opening is not only short but also very narrow. The pulse length is close to the theoretical minimum (100 µs, with rise and fall times of about 150 µs).
The rust microparticles produced by oxidation are so tiny that they slip quickly through the injector nozzle when the injector works with longer openings. When the opening is narrow, they get stuck.
Why the replacement DME seemed to work
There is an obvious objection. After I installed the replacement DME, I drove about 1000 km without any sign of the defect. That was before I tidied up the FA and recorded the correct FR. How is that possible?
The replacement DME had a “historic” FR, 75XXXXX, recorded directly after installation. The next FR group was 76XXXXX, and the last, 86XXXXX, is the current one. Only the last part of the previous FR group is “usable”. During the test drive I found that the DME regularly records misfires, caused by the uneven road and a suboptimal injection strategy.
That explains it. The DME did not use the injection strategies with very short openings, so the defect did not show up. It sometimes used Homogeneous lean mode, but it refused Stratified charge mode during the first kilometers after starting the engine.
Conclusion
This shows how capricious the injection system of these engines is. I hope this entry strengthens your belief that fuel additives and cleaning aids of any kind are a “death sentence” for these engines.