Select a technical image to view it in detail.
Everyone who owns a car with an N43 or N53 petrol engine has probably seen these engine names in error message descriptions, usually alongside complaints about uneven running. There is a legend that these engines run smoothly, but it is mission impossible.
The engines do have ignition coil trouble. When replacing coils and spark plugs, keep the environment clean and follow the replacement procedure strictly. The coils should also be upgraded with snubbers, so that voltage overshoot does not repeatedly damage them (see the post about ignition coils).
But it is not that simple. I had suspicions for some time, and I decided to confirm or reject them, which took some effort.
Preparing the test
First I bought an ELM327 adapter.
Next I installed software on my computer that can read OBD2 mode 06. Neither step caused any problems, and both the adapter and the software worked immediately after installation.
Why this is necessary
The MSD80 detects misfire from changes in flywheel speed (Bosch, for example, uses an accelerometer instead). Siemens introduced this misfire detection system in the MS43 management module, which was used with the M54 engine.
Unfortunately, neither INPA nor ISTA D displays the misfire count for each cylinder, so there is no online monitoring.
My suspicion was this: what happens when a BMW (here an E60 LCI with N53B30, manual gearbox, M suspension and 19″ alloy wheels) is driven on a quite uneven road? Can the road surface cause false engine misfire error messages?
The experiment
In the city I switched on the ELM327 adapter and made sure the misfire counts for all cylinders were 0. I drove through the whole city, and the engine ran in all its modes:
- Homogeneous
- Homogeneous lean (Lambda = 1.5)
- Stratified charge, with Lambda up to 3.0–3.2
After 20 km in total I stopped and checked: the misfire count was still 0 for every cylinder.
Then I turned onto a well-known road with a “very good” surface, engaged cruise control in 6th gear at about 75 km/h. I did not want to go faster, because I feel sorry for my car.
The experiment was over after several minutes. The MSD80 had counted 17 (!) misfire points in total. Since each point needs several detections, uneven running was detected at least 30–40 times (see the explanation below). It then switched off Stratified charge and stopped counting misfires. Problem found, so off to the service center? But I did not feel any incorrect engine behavior, and certainly no mass misfire.
According to this, the engine is completely broken, with regular problems in cylinders 1, 3 and 5. Cylinders 3 and 5 already have brand new injectors, in addition to new ignition coils and spark plugs, so it is hard to believe they are faulty. If you check with ISTA D, it will have records for cylinders 1, 3 and 5. Its cold injector test will show that the coils and plugs are in working order, and it will suggest replacing the injectors on those cylinders.
On the way back I repeated the experiment with different software. Again, all cylinders started at 0 events, and then came the “crappy” road:
In this image the misfire counters are the 12 bottom lines, two for each cylinder. This time there were “only” 14 legit misfire points, which means flywheel unevenness was detected 20–30 times.
The manufacturer is not stupid
BMW did foresee that roads can be uneven. For this purpose the DSC/DTC module sends a “Rough road” signal, and misfire detection is switched off for a while, as it is under variable load or when the throttle angle changes. When you brake often and turn the steering wheel (or the wheels slip), the MSD80 receives the signal that the road is uneven. What the Germans did not foresee is that our roads are so uneven, across their whole width, that it makes no sense to brake or change direction.
Consequences
The bad news:
- On mid- or low-quality roads the MSD80 records uneven-running error messages.
- The misfire counters behind this error message (as seen in the ISTA D misfire test procedures) do not match the real situation.
- When the N43/N53 “catches” this error, Stratified charge is switched off for a while. Homogeneous lean injection continues at Lambda around 1.5, and fuel consumption rises slightly.
- When the engine starts a new session, it starts without Stratified charge. This is not ideal, since Homogeneous lean is the most unpleasant mode for fuel combustion in a lukewarm engine. Stratified charge comes back after 5–15 minutes, which is the good news.
The good news:
- The N43/N53 restores Stratified charge after a while if you do not change the driving mode, or after the driving mode changes (and after the individual cylinder corrections are recalibrated). Note that a much lower misfire count is then needed to switch Stratified charge off again.
- This error message, in neither active nor passive status, stops long-term trim formation or causes any noticeable problems in engine performance, apart from slightly increased fuel consumption.
Note: if uneven running (real or false) is also detected in Homogeneous lean mode, the engine switches to Homogeneous injection and starts re-creating the bank and individual cylinder fuel adaptations.
How the guilty cylinder is picked
The flywheel is a quite massive piece of iron. If a misfire occurs at idle, the faulty cylinder is identified quite reliably. At higher RPM it is hard to pin a disturbance on one cylinder in time, because the exact moments when the problem starts and stops are difficult to recognize. So the possible “guilty” cylinders receive coefficients. For example, one cylinder may be guilty for 70% and the next one in time for 30%. When any suspect cumulatively reaches a virtual probability of 100%, its misfire count grows by 1.
Some cylinder is always the first suspect, even in a very chaotic distribution, and that cylinder has a chance to become the dominant “guilty” one. The MSD80 records the list of possible culprits and their coefficients and keeps it between sessions, to make it easier to find the culprit in later sessions.
There is an interesting coincidence here. The firing order includes 1/3/5, so all the “defective” cylinders are consecutive in time, and their working cycles together make a full circle.
Siemens monitors misfire events by density, that is, by counting cases of flywheel uneven running. The MS43, for example, counts the events of each cylinder over 200 crankshaft turns and the total number of events over 1000 crankshaft turns. A separate, rare misfire is therefore not counted or recorded as an error. From the MS43 description, I understand that the individual memory buffer for each cylinder is 5–10 seconds long, and the total register up to 1 minute.
There is some relatively good news. For the misfire error status to become active, the road has to be very uneven for a longer while, and short uneven sections will not activate it. But as my experiment shows, in real conditions the misfire count reaches several tens within a few minutes.
The moral of the story
Under our driving conditions these misfire detection systems give false error messages. The cylinders recognized as damaged are NOT damaged, because the system is reacting to the unevenness of the road. Even engines in perfect condition will show misfires in one or several cylinders. The method BMW recommends, swapping coils, spark plugs or injectors between the “good” and “bad” cylinders, will also give false results.
This is the result after half an hour of driving on a normal road surface:
Be careful: some MSV/MSD software releases have problems with diagnostic programs. Misfire is usually shown in cylinder 1, sometimes accompanied by real shaking or rough running.
Below is the result after 10 km of driving with INPA connected in live data mode. No real shaking was noticed.
As you can see, cylinder 1 has 86 (!) misfire points counted.




