Select a technical image to view it in detail.
B58 misfire detection experiment, part 1
Given the B58’s problems with identifying misfires, I ran a series of experiments. The first goal was to see how the B58 handles misfires with different causes. The second was to check my suspicion, based on the symptoms, that Bosch still uses crankshaft rotation unevenness to detect misfires while the catalytic converter is warming up (the first minute after engine start).
Misfires can come from an incorrect fuel mixture or from weak ignition. Ignition problems have several variants of their own, such as a defective coil or a flooded spark plug. Depending on the defect, the timing of spark creation may or may not change, and the DME could also use an abnormal spark timing to identify misfires. Does it? Read on.
Warning: do not try to repeat these experiments. You can get a fatal electric shock.
Test setup
I used a brand new Magneti Marelli ignition coil on cylinder 2. I connected it correctly but did not fix it in place. To provoke misfires, I gradually pulled the coil out of its seat. This increased the losses in the coil’s high-voltage circuit, because part of the energy is lost as discharge into the air.
Pulling the coil out: no effect at first
First I pulled the coil out by about 1 cm:

There was no difference at all in engine performance, including during the catalytic converter warm-up phase. The engine ran perfectly.
Then I pulled the coil out further, to about 2 cm from its intended position:

Again no effect. I increased the air gap by a few more centimeters, still with no reaction. I began to wonder whether this engine needs ignition at all.
My first conclusion is that the energy of these coils is enormous, and the reserve is huge. This engine is not unique in that; every industrially developed engine has a substantial spark energy margin. So if an engine misfires, fitting more powerful ignition coils is not the answer. It only fights an unsolved problem elsewhere.
Adding load to provoke misfires
Since simple methods did not cause misfires, I did the following:

I increased the air gap and added a pair of pliers as an extra “load”, placed on the rail and thereby grounded. Part of the spark energy should “bump” into the pliers.
This works, and the first misfires finally appear.
Misfires during catalytic converter warm-up
Right after the start (at the 133rd second) everything looks fine:

After a short while, the first misfires appear:

At the 165th second a full misfire occurs, and the DME notices it. At the 171st second there is a partial misfire, which the DME misses. To be fair, the catalytic converter is still warming up and the engine runs less evenly than it normally does at idle. Such partial misfires cannot be considered a problem.

At the 182nd second the mechanical energy of the “damaged” cylinder is INCREASED, and the DME identifies a misfire (the last seconds of the warm-up phase). Why is the energy increased? The other cylinders are not to blame, because the vector-type idle smoothing is not used in this mode. Was there a premature ignition? Did the DME identify an incorrect combustion time? At this point I have more questions than answers. Still, one conclusion holds: during the catalytic converter warm-up phase, the DME recognizes all suspicious situations quite correctly.
Here is a zoomed-in view:

There is a solid spike in energy. A very exotic explanation could be that the fuel ignited at the beginning of the dwell time rather than the end. The reason would be that these coils contain no diodes, which were an integral part of the coils on the previous, pre-direct-injection generation of engines. This theory should be checked.

Exactly one minute after the start, at the end of the catalytic converter heating period (which is significant), the misfire counter of cylinder 2 is cleared.
Misfires after warm-up

At the 252nd second there is a partial misfire, and the DME does not notice it. Once the warm-up phase is over, the DME suddenly does much worse at detecting misfires.
Here are the next several tens of seconds, zoomed in:

We see several partial misfires:

The DME notices none of them.
It looks as if the DME becomes very sharp-eyed during the catalytic converter heating phase. My suspicion was justified.
Group and continuous misfires
Next I checked how the DME handles group misfires. I increased the air gap a bit more, and the engine now ran on 5 cylinders:

The engine started at the 196th second. Cylinder 2 misfires continuously. Starting at the 200th second (4 seconds after the start), the DME began registering misfires. The counter rises at around 10 misfires per second, which means the DME recognizes every one of them. In addition, no false misfires were registered in any other cylinder.
Unfortunately, continuous misfiring cannot be sustained for long, because the DME switches off the injector of the “damaged” cylinder. So I continued with relatively rare misfires, provoking them from time to time:

The engine is started at second 0. By the 20th second there are 4 events of incorrect performance of cylinder 2, and the DME notices all of them.

At the 21st and 26th seconds cylinder 2 performs incorrectly again, and the DME notices both events. At the 31st second there is a controversial partial misfire, which the DME misses.

At the 46th and 54th seconds there are two misfires, and the DME notices both. More interesting is that at the 56th second the misfire counter of cylinder 2 is cleared, again exactly at the end of the catalytic converter warm-up phase.
The next image shows how the DME recognizes misfires after the warm-up phase:

There are group misfires from the 194th to the 196th second. The DME notices about 6 of them, while the graph shows at least 15 to 25. From the 199th to the 201st second there are continuous group misfires (at least 8 to 9), and the DME notices only 3. From the 204th to the 206th second there are some separate misfires, and the DME handles these better.
All these misfires are caused by “ignition defects”, which also appear as a significantly shorter combustion time. If the combustion time stays at least partially correct, the DME at idle, AFTER the catalytic converter warm-up phase, does not recognize such misfires at all. More precisely, it stubbornly pretends not to recognize them for more than a minute (see here).
When an “ignition problem” misfire does get detected, the DME acts correctly. It immediately switches off the injector of the damaged cylinder in the background, so there is no smoke and no pungent smell from the exhaust. Even so, some surprises awaited me after such misfires, but more on that in the continuation.
Combustion time measurement
Finally, here is how the DME measures combustion times.

Until the 28th second, the cylinder 2 coil was in its intended position. At the 32nd second I pulled it out by about 1 to 1.5 cm. At the 41st second it was pulled out by about 2 cm.

At the 44th second the coil is pulled out by about 3 cm. The DME identifies a stable and significantly reduced combustion time.

At the 74th second I started moving the coil back toward its correct position, and at the 77th second it was back in place. Everything is in order again, and the DME measures the combustion time correctly.
Read the continuation here.