During the previous experiment (description here), I explored how the B58 series engine recognizes misfires in idle. Not it’s time for the next experiment – on the go, partial load mode.
As in the previous experiment, I used a spare ignition coil, which I gradually pulled out the shaft to increase the losses in the spark plug circuit.

This image shows how large the losses I had to create for the misfires to start. The air gap of more than 2.5 .. 3 cm! I can only repeat – the ignition coils of these engines are very powerful!
On the 342nd second, I finally found the optimal position of the ignition coil and turned on the engine. The car runs idle.

On the 388th second, we see the misfire of the 1st cylinder. DME correctly notices this misfire. Here I have to remark that the engine runs in the warm-up algorithm (as mentioned in the previous entry, all symptoms say that in this mode, Bosch is cheating and uses the data of the flywheel to identify the misfires). On the 389th and 390th seconds, we see partial misfires – DME does not set them.

On the 410th second, when the warm-up stage is completed, the misfire counter of the 1st cylinder is cleared. Around the 415th second, DME switches on the vector mode to maintain the idle.

On the 430th second, DME (after detecting that misfires are finished) starts restoring/renewing the injector adaptations necessary for the vector method – the idle becomes more stable.

I started to drive. By medium torque (accelerator position around 40%), on the 489th second, cylinder No.1 has group misfires. RPM of the engine: 1500; gear 4; speed – around 55 km/h. DME correctly identifies these misfires.

Second 591st. Accelerator position around 20%. Gear 6. RPM around 1400. One misfire. DME identifies it correctly.

To increase the density of the misfires, I slightly increase the air (loss) gap. I started the engine on the 723rd second. Yes, now there are a lot of misfires! As expected, during the warm-up stage, DME identified them correctly.

On the 790th second, when the warm-up stage is completed, DME deletes the misfire counter. As we see, misfires continue, and the counter starts from the beginning.

It is interesting that in such a situation (if some cylinder has performed poorly during the warm-up stage), after the warm-up stage, DME also “understands” which cylinder is “the sinner.” It looks that here Bosch cheats again: based on the initially gathered data, indicates the possible culprit “by inertia”.

On the 865th second, the misfire counter is deleted again. And again – it starts the counting from the beginning.

I started the movement. On the 940th second, there is a misfire (at least – partial) in cylinder No.1 – DME misses it. The accelerator position is around 15%. RPM: close to idle. Gear 6.

On the 958th second – a fake misfire: actually the cylinder works correctly. The accelerator is around 20%, RPM – around 1300. On the 960th second, by accelerator position around 40%, the group misfires of cylinder No.1 start. DME notices them correctly.

Group misfires start on the 1003rd second, by accelerator position around 50%. DME notices them correctly. It is interesting that – if DME starts to notice misfires of the exact cylinder, it does not lose this ability even in case the accelerator position is swiftly reduced (I instinctively reduce the required torque in such situations).

On the 1014th and 1023rd seconds – the group misfires by accelerator position around 40 .. 45%. DME correctly identified them. On the 1016th second – fake misfire. This time, it is interesting that on the 1023rd second, directly after the group misfires, the misfire counter is cleared.

On the 1051st and 1052nd second – one misfire in each of them. DME recognizes only one of them. Required torque – very low.

On the 1077th second – the group misfires again. This time, DME does not lose the ability to recognize them when the accelerator drops to 10 .. 15%.

1106th second. What we already saw: if the accelerator position is around 40 .. 50%, DME immediately recognizes the misfires. This time, the misfire counter is deleted after 8 seconds.

1122nd second. Individual misfire. The accelerator position is around 10%. With all due respect: DME recognizes the misfire.

1133rd second. Group misfires by accelerator position of around 40%. DME works correctly.

1141st second. Several misfires by the accelerator position around 30% and 1700 RPM. This time, DME recognizes only one misfire.

1222nd second one misfire by the accelerator position of around 10%. DME recognizes the misfire, but the counter is deleted after 2 seconds already.

On the 1250th second, there is one misfire by a low (around 10%) accelerator position. DME recognizes it. One second later, DME identifies a fake misfire, although all cylinders are working correctly.

On the 1361st second, one more misfire by low required torque (accelerator position around 15%; RPM around 1500, gear 4). DME recognizes it correctly.

Here, after all this cluster of innumerable misfires – no error messages regarding misfires of the 1st cylinder. There is only an error message regarding incorrect combustion time. Needless to say, no EML (Check engine) lights up.

What are my conclusions?
Idle – if any cylinder misfires in the warm-up stage, the DME continues to be “smart” after the warm-up stage. It is possible that – although DME can not understand from the vibrations of the engine which cylinder is exactly misfiring- DME’s reaction is based on the initial (warm-up stage) experience.
On the go, the misfire recognition is doing pretty well. A worse success is in the mode of lower required torque. It is understandable – in this mode, misfires create less vibration. If the accelerator position is 30+%, misfires are identified with quite stability.
If DME has correctly identified misfires by higher torque, it does not lose the “ability”, when the torque decreases. It looks like in this situation, DME takes into account past experience. Say – we already found the culprit, so all vibrations can be “blamed” on it.
Sometimes, fake misfires are identified (in situations when all cylinders work correctly). DME “assigns” some small vibrations to the cylinder, which was the culprit previously.
DME services (deletes) the misfire counters quite strangely. Sometimes, they are not deleted for even several tens of seconds; sometimes – they are deleted directly after misfires.
The situation with the misfire provoking is interesting, too. When provoking misfires with the existing method (change of the ignition coil position), misfires start faster “classically” (as for the “old” port injection petrol engines) – in the range of lower RPM and high required torque. The mode of the very low torque turns out to be stable. This is one more confirmation that the problems in the low required torque mode are created exactly by the spark plugs (the spark, which “jumps” in an inappropriate place), not by the lower spark energy (due to ignition coil problems).