Select a technical image to view it in detail.
The first part of the experiment is here.
In this part, I look at which fault messages were recorded in the DME fault memory.
What the fault memory shows

Some things are unchanged: the fault memory again points to “several” cylinders, and the turbocharger is switched off.
The most interesting point is that the DME correctly identified combustion time problems in the right cylinder, the 2nd.
Here are the fault messages in more detail:

I do not know how the allowed combustion time difference between cylinders is calculated. However, the alarm was raised when the combustion time of the damaged cylinder was around four times shorter than that of the cylinders running correctly.

The “combustion time too short” message is simple. Its error threshold is 0.5 ms, and everything looks correct, which surprised me.
The time stamps are more interesting, though.
Time stamps and the catalytic converter warm-up phase

The DME noticed misfires of the 2nd cylinder twice:
- when the engine started to run on 5 cylinders during the CO catalytic converter warm-up phase;
- at 3:02, when I ran the experiment exactly during the CO catalytic converter warm-up phase. Here is how it looks:

23 seconds after starting the engine, I began pulling the ignition coil out of its seat to imitate misfires. I did not pull it out far enough for the DME to register an inappropriate combustion time. No fault message about it was recorded, and that is significant. For about 10 seconds the DME pretended not to notice anything, then suddenly started to “notice” misfires.
Here is the continuation:

The misfires were identified at a rate of 5 to 6 events per second, which means the DME again successfully identified ALL of them.
For contrast, here is what happened when I started provoking misfires after the CO catalytic converter warm-up phase:

Over 15 seconds I provoked group misfires, and the DME registered only THREE of them. The actual number of misfires was roughly 10 times higher.
I have not the slightest doubt that Bosch is cheating again here: during the CO catalytic converter warm-up phase, the crankshaft rotation data are being evaluated.
The combustion time tolerance and absolute value fault messages, on the other hand, are fine. They are registered at exactly the right moments.
Conclusions
- The DME measures combustion time correctly and correctly registers fault messages about its inconsistencies.
- Bosch is not allowed to measure changes in crankshaft rotation speed (and identify misfires from them), so it works the combustion time data twice as hard. However, the tolerance error corridor is wide and almost coincides with the minimum combustion time, so the added value of the tolerance error is questionable.
- During the CO catalytic converter warm-up phase, the Bosch DME still “mystically” sees the misfires. It looks as if cheating is going on here: in parallel to the combustion time and accelerometer data, the mechanical efficiency of the cylinders seems to be checked as well.
- Once it identifies the combustion time problem (and the misfires during the CO catalytic converter warm-up phase), the DME immediately and correctly switches off the injector of the damaged cylinder.
Bosch’s fight seems quite desperate. Until now the DME reaction did not surprise me, but if there were no surprises, there would be no third part of this experiment. Read it here.