When checking my car with the B58 engine, I noticed the following picture:

All cylinders have regular misfires! Yeah, I had the feeling that by very low required torque, the misfires could be felt, but I blamed uneven road or my paranoia after regular work with N43/N53 series engines.

I checked – the displayed misfire counters are of the actual session. When starting a new session, data are cleared, and their (misfire) number increases during driving time. I had no doubt that DME really registers the misfires in low load mode. 

I was wrong.

At first, I decided to check if my suspicions regarding misfires on the low requested torque mode were true. I turned on the Expert mode, chose Rough run data, and went to a drive. 

Notice: for these engines, the DME can detect the mechanical efficiency of each cylinder during ALL ranges of RPM! This is a fundamental upgrade after MSD80 (which can do it only in idle) and MSD85/87 (these DMEs measured the mechanical efficiency of cylinders till 1500 RPM). 

Here, the unevenness of the engine performance (cylinders) in idle.

On the left side of the image, you can see how I switched the gearbox to a neutral position; the rev-matching system reduces the RPM till idle (blue curve). At this moment, a very short unevenness of the engine performance can be seen. The driving speed of the car (green curve) gradually drops to 0.

Here, the unevenness of the engine performance, while driving with speed around 70 .. 90 km/h; gear 6, released accelerator pedal (in Overrun mode, the fuel is not injected in cylinders – I made sure about it):

The only slight unevenness on the right side/third of the image – I (again) switched the gear to idle. 

And here, the same road, the same speed, and gear, only – I drive with a very low required torque (the econometer shows only 2 .. 4 l/100km; the position of the accelerator pedal is indicated by the red curve):

The legend of the picture – below in the blog entry.

As we see, the performance of the engine becomes VERY uneven! It becomes very uneven in the second part of the image – as you see, the accelerator pedal (red curve) is almost completely released. On the first side of the image, the econometer indicated 5 .. 7 l/100km; on the second side: 2 .. 4 l/100km. And as we see from all the graphs, the uneven road is not the one to blame. Both in idle and Overryn mode (so – DME “can feel” all evennesses of the road); in larger requested torque conditions (driving the same road), the engine runs evenly, and that can be seen in the graphs above. So – in this: low torque driving, and there actually are problems. Problems, how it looks, affect all cylinders – it is impossible to identify one or two culprits. The fuel mixture is correct (I checked – Lambda 1.0; control probes confirm a non-lean fuel mixture). So: in combustion chambers 2 .. 3 times less (comparing driving conditions/requested torque on both sides of the graph) fuel is injected, but the unevenness of the cylinder performance increases significantly (at least for 3+ times). These are misfires; there are no other options! Unfortunately, DME did not identify ANY misfire!

There are two stages for comparison:

I drive with a very low required torque (on the left); on the right side – I drive idle. The red curve – position of the accelerator pedal. 

As we see, the engine’s performance in idle is even; all unevenness starts when the engine works with lowered required torque.

The next step – the analysis of Combustion timing. 0 ms ignition time should confirm the misfires (moreover – misfires ”outside” spark plug). Here is the graph.

Interestingly, Combustion time data did not indicate real misfires (caused by current leakages outside the combustion chamber). Yes, the jittering was as if it increased, but 0 ms ignition/combustion time (which would suggest that the spark hasn’t jumped the electrodes of the spark plug) I did not see. 

But the misfire counters indicate that there are misfires registered. When does it happen? Here is one of the moments when the misfire data grows rapidly:

When zooming the Y-axis, make sure that misfire events are registered: 

Here, the legend:

In the graph, you can see that when the accelerator pedal was released four times (the red curve), the speed choice graph indicates (the grey curve) that gear 1, then 2, 3, 4, and finally – gear 6 is chosen. At these moments, I switched the gears. On each gear switching, the misfire is identified (pink, yellow, and brown curves)! Fantastic! Instead, when the misfires happen in real (the shivering of the engine can be felt and seen in the Rough run data), misfire events are NOT registered!

Why are the misfires recorded when the gears are switched? The reason is simple: at the moments of gear switching, the rev-matching system “affords” around 5% lower RPM than the existing ones. Accordingly, when releasing the clutch, DME notices a slightly lower engine efficiency. The accelerator pedal is “activated” a moment after releasing the clutch. Between both these events – releasing the clutch and activation of the accelerator – the engine efficiency is slightly lowered. This is enough to detect the reduced efficiency of some cylinders: misfire.

Summary:

a. in ISTA call-up functions, misfire counters are cleared 3..5 seconds after the event (misfire);

b. for a low/very low requested torque combustion timing misfires are not identified;

c. if the car is equipped with a manual gearbox, one misfire is recorded every time the gear is switched; misfire counters can be used as counters of how many times the gears are switched;

Seems the programmers have “forgotten” one “small thing” – put the gear-changing procedure in the Transient event list. During Transient processes (for example, swiftly pressing or releasing the accelerator pedal, pressing the brake pedal, etc.), in a normal case, misfire identification is stopped for a moment. But not in this case.

Does it seem to me alone that the software of the heart of the car – the most important unit has been written very perfunctory? I have suspicions that Mr/Robert Bosch has turned in his grave because of shame. 

This time, the problem causes can be:

a. worn spark plugs;

b. low-quality engine software or fundamental problems in this mode.

I don’t rule out the b option. In this mode, the fuel mixes with the air so poorly that every 3-rd cycle goes “in vain”. Or – the engineers have “forgotten” to activate the multi-ignition mode (because misfires are not identified). Or – there could also be some fault with ignition timing (for example, wrongly received knock data). There could be many options. I would not be surprised by anything.