Misfire detection algorithm and it’s fails

Many people are asking – why the misfire algorithm tests for different software releases were performed? Why it is so important?

 

Very short about theory. If one of the cylinders has misfire, the changes in flywheel speed are insignificant. They are small and with fluctuation effect.

As a sample – image from OVERVIEW OF ENGINE MISFIRE DETECTION METHODS USED IN ON BOARD DIAGNOSTICS. Journal of Kones. Combustion Engines, Vol8, No 1-2, 2001.

As we see, the changes of torque for current test sample on 2000 RPM is even hard to see, not even trying to identify the “guilty” cylinder!

It can be seen, checking the live data via INPA in the situation, when some of the cylinders has bee turned off.

In this sample the cylinder No.1 has been switched off. As we see, the cylinder No.1 affects also values for other cylinders: cylinder No.5 (next according to firing order) has seemingly higher efficiency, No.3 (next after No.5) – with slightly increased, etc.

As can be seen here, the self-diagnostics of MSD80 for many releases (especially for older ones) are making gross mistakes, when detecting fluctuation process differences (in case of damage of any cylinder) from ideal performance.

In case, when the MSD80 Rough run algorithm functions incorrectly, sometime the truly “guilty” cylinder shows not decreased, but, quite the opposite – increased activity (as a result of mistake in self-diagnostics).

Accordingly, in case of incorrect performance, as “guilty” is marked the next cylinder (in firing order) – in case, if the polarity of all transition process has been defined incorrectly, or next after next (in firing order), if the efficiency of damaged cylinder has been detected incorrectly, but the polarity of the rest of process – correctly.

 

What does it means in case with real engine?

Let’s suppose, that the ignition coil of cylinder No.1 is damaged. But, as the MSD80 has old (for example, 7611396) software release, the algorithm marks “guilty” cylinders:

No.1 – in cases, when the algorithm performs correctly;

No.3 – (next after next, in firing order) – in cases, when the efficiency of “guilty” cylinder has been detected incorrectly (close to nominal value);

No.5 – (next, in firing order) – in cases, when the polarity of process has been mixed up.

 

In a typical case (release No.7611396): if the cylinder No.1 during one hour has been missed ignition cycle for 10 times, we can get following (typical) results:

1-st cylinder: 5 misfire points;

3-st cylinder: 2 misfire points;

5-th cylinder: 3 misfire points.

 

These data are approximate – they are obtained by provoking misfire in exact cylinder. As if everything is OK – the cylinder No.1 (really damaged) is dominating. But, in real life, each of the cylinders has been detected as damaged for several times even without real evenness (for example, the affect of uneven road).

 

Taking in account the rough road factor (influence), even in case of even road the misfire counters become more similar, for example (the same release 7611396):

1-st cylinder: 7 misfire points;

3-nd cylinder: 5 misfire points;

5-th cylinder: 6 misfire points.

 

Which cylinder(s) is(are) damaged? And if the changes in counters due to Rough road are “in favor” of cylinder No.5 (for this scenario even several additional one-time/random events are enough)? Or – one or even all three counters of cylinders are exceeding threshold and give the reason for ISTA D count them as “damaged” during performance of test?

The experience shows, that the existence of defect for one cylinder (even not real existence, but, for example, sensitivity of 1-st cylinder to data logger connection) creates messages for at least 2 “damaged” cylinders!

One more significant reason, why the Rough run self-diagnostics system for early releases of MSD not only don’t helps to identify the damaged cylinder, but actually does the opposite: several “healthy” cylinders are identified as damaged!

 

What to do? Carefully follow the recommendations described here, especially:

1. check misfire counter with ELM adapter, don’t rely on values, detected by ISTA D;

2. perform short driving sessions on good quality roads, not long ones on bad quality roads;

3. during observation of misfire counters, turn off the diagnostic tools (turn it on only when reading data);

4. if possible, upgrade the MSD80 software release. As the best possible (during tests) are recommended:

8603156 N53B30 UO E60/E61

8603178 N53B30 OO E60/E61

8603182 .. 188 N53B30 OO E63 (additionally carefully check the fuel offset LTFT, their correct creation).

These (newer) releases has tiny number of falsely detected “damaged” cylinders, and self-diagnostics is performed reliably. If the ELM adapter shows large misfire numbers for any of cylinders during driving on even road – check the spark plugs and ignition coils of “guilty” cylinder.

 

Related entries:

Uneven running. Rough run. Smooth run.

Uneven running. Rough run. Smooth run. EP 2

BMW AG. Problem solved. Or not?

MSD80 software releases