Today I had the chance to be present in quite interesting case, where several mechanics were not able to find the cause of problem.
Vehicle: BMW 6er, engine N52.
Symptoms: periodical error messages regarding trim in both banks (especially in first bank), uneven run, increased CO level in exhaust gases.
Many different parts have been changed, including both wide-band Lambda probes and catalytic converters. Unfortunately even replacement of catalytic converters has not decreased CO level in exhaust gases.
Examination of engine’s mechanical issues showed, that there are problems with hermetic of crankcase ventilation tubes. After elimination of this defect, the examination was continued – the experience suggests, that there should be also other cause of problem.
As we see, almost everything looks very nice and neat. Both offset and proportional long-term trims are close to 0, integrators are stable, and system works in closed loop mode.
Only thing, which is inappropriate – Lambda value in first bank. 0,96, which is significant difference from 1.00 and it also dissociates with fact, that integrator don’t try to correct the situation. Together with inappropriate Lambda value, offset long-term trims of 1-st bank are creating more rich fuel mixture than in 2-nd bank. This obstacle confirms the assumption, that 0,96 could be the true value of Lambda (not, for example, short term deviation from ideal).
To clarify the cause of problem, we check the dynamic resistance of Nernst cells of Lambda probes:
As seen in the image, both wide-band probes show correct values of dynamic resistance of Nernst cells, in 2-nd bank the dynamic resistance ir either normal or there is a short circuit (true value is around 100 Ohm, MSV ADC in case of correct values of Nernst cell dynamic resistance can show value 0/256/512 Ohm), but for the 1-st bank the dynamic resistance of control probe is abnormal: several tenth times above norm. It means, that this probe, most probably, is damaged. There can be several reasons for elevated dynamic resistance of Nernst cells: including – inappropriate heating/temperature (as seen in first image, the heating of probe is with correct PWM cycle; it really happens, the probe is good/long enough heated – this is not the cause of the problem) or the result of damaged wires/connections (they have been checked).
As seen in in image, the voltage of control probe of 2-nd bank is appropriate Lambda 1,00 .. 1,01. So this probe works normally (suspicions regarding possible short circuit are not confirmed).
But voltage of 1-st banks control probe even with Lambda 0,96 (reported value of wide-band probe) don’t reaches over 0,2V (and keeps steady, which indirectly also excludes abnormal unbalance of 1st banks cylinder mixture), which is inappropriate in case of rich fuel mixture.
Average Lambda value was also controlled via external Lambda meter and confirmed problems with 1st banks Lambda (average Lambda value in exhaust was lowered – around 0,98).
Here we have to take in account, that MSV70/80, MSD80, when calibrating the wide-band probe, changes the mixture within certain limits to get the acknowledgement from control probe, that the stoihiometric mixture has been reached. For this reason, performance of control probes for MSV/MSD control units has to be monitored very close.
The lowered voltage from 1-st banks control probe excludes also possibility of bad connection (increased series resistance, around 4K Ohm) in connection “probe to MSV” – input impedance of MSV/MSD is high, they don’t reduce the EDS voltage, generated by probe.
Note: several scenarios are possible (one of them – as described above), when in case of damaged control probe the MSV/MSD control unit falsely identifies the wide-band probe of corresponding bank as damaged component (because it hasn’t finished the calibration procedure of wide-band reference value successfully).
Self-diagnostics of MSV70 failed regadless of several obvious problems/symptoms:
- abnormal Nernst resistance of control probe (4K Ohms instead 512 Ohms max);
- lowered EDS of control probe (0,2V instead 0,8V min);
- inability to calibrate wide-band probe’s Lambda reference value;
- incorrect value of average Lambda, reported by wide-band probe of corresponding bank.