Select a technical image to view it in detail.
This case is an unusual one. The car has an N43 engine, the low-power version, and its full functionality had already been restored: the NOx system was repaired and stratified charge works again.
Complaint and first data
The owner complained of occasional symptoms, both while driving and at idle, and of increased, permanent vibration with a “diesel” feeling. These symptoms typically point to injector problems, and that defect is quite common. I asked the owner to send me diagnostic data. The DME error memory contained no fault messages.
The fuel mixture looked like this:

This is correct. The DME has not created two virtual banks, because the car has only one wideband probe, but otherwise I have no objections.

The NOx sensor also confirms the correct fuel mixture. In addition, the overrun test from the last visit was fine, so the injectors were not leaking.
Injector corrections in homogeneous mode
The latest injector tests in homogeneous mode, however, indicate serious problems:

The injectors in this menu are listed in the order 1/4/3/2. They are split as if by banks: the injectors of the 1st bank receive positive flow-rate corrections, and those of the 2nd bank receive negative ones.
The live data of the injectors:

This confirms that a huge flow-rate correction is being applied, in places around 25 to 30%.
It is unlikely that two injectors of one bank have both changed their flow rate in the same direction. The flow-rate corrections in stratified charge, on the other hand, are entirely normal:

This case is bizarre. Injectors with atomization or spray-pattern problems normally show increased corrections in stratified charge. Here it is the opposite: the injectors have problems only in homogeneous mode.
The control probe and the cause
The readings of the control probe also look suspicious:

Is the probe faulty? A test in kick-down mode showed that the lambda probe works correctly. Yet the mixture appears correct while the probe reads lean, so could there be additional air in the exhaust?
I asked the owner to check for this. It did not take long to find: the exhaust was not airtight, and the extra air suction was detected close to the exhaust manifold. Once the leak was fixed, all the problems disappeared.
How such a defect produces these symptoms
The engine has only one wideband probe, yet the DME works as if it had two independent banks: cylinders 1/4 and 2/3. Within each virtual bank, the DME trims the wideband probe, both at idle and on the move, using the readings of the control probe.
If air is sucked in at cylinders 2 or 3, as in this case, the wideband probe is trimmed incorrectly for those cylinders, toward a much richer mixture. When the cylinder efficiency tests run, the DME then identifies increased activity in both cylinders. This is logical, given the incorrectly trimmed probe. The DME tries to reduce that activity by applying negative flow-rate corrections to the injectors.
The mechanical defect, the air suction, is unstable because of engine vibration, temperature changes and so on. The result is a significantly different fuel mixture and an imbalance of cylinder efficiency between the banks. The “diesel” feeling is a natural outcome, as is an increased rate of misfires caused by the incorrect individual fuel mixture of the cylinders.
Note on engines with two wideband probes
The same defect can occur on the N43 (and also the N53) with two wideband probes. With a strong, constant air leak, there will be error messages about trimming the wideband probes. However, if the DME manages to trim the probes, there will be no error messages about their performance.