The oxygen sensors (Lambda probes) perform the control of exhaust gases and are one of the most important parts of petrol engine. The modern BMW engines use 2, 3 or 4 oxygen sensors.
4 cylinder engines: wide-band oxygen sensor for each bank (for example, N43 for 1; 3; 5 series; X20i) or one wide-band probe for all cylinders (for example, N43; 1; 3 series; 116i; X18i) and a control probe (narrow-band), common for both banks (for example: all N43 series engines) or two control probes (N46 series engine).
6 cylinder engines: each bank has a wide-band oxygen sensor and also each bank has a control (narrow-band) sensor.
Wide-band Lambda probes are placed on the beginning of exhaust system, before CO catalytic converters.
Control Lambda probes are placed after CO catalytic converters.
The exhaust system of N43 series engine (with one wide-band probe, one CO catalytic converter and one control probe).
The exhaust system of N43 series engine (with two wide-band probes, two CO catalytic converters and one control probe).
The exhaust system of N46 series (4 cylinder) engine and placement of probes.
The exhaust system of N52/N53 series (6 cylinder) engine and placement of probes.
Technical documentation of BMW “keeps silent” regarding algorithms of probe management. Regarding less complicated (technically) issues, for example, a system of cooling there are block diagrams, also a description of some systems (for example, electric water pump), but electronic control systems are not described. Obviously, BMW believes self-diagnostic system can handle diagnostic of probes. Unfortunately, the system is not reliable enough – the practice shows that. In more than half from cases, when there are suspicions regarding problems of exhaust gas mixture control, there are no error messages regarding probes or, if they are, they are quite inadequate. Usual solution – change of all probes. Four probes costs more than EUR 300, and in my opinion, replacement of elements, before their performance is evaluated, is not the mark of professionalism.
Accordingly – material, mentioned below, can have derogation from real situation (even more – some nuances or functional algorithms can differ also from software release).
Narrow-band probe serves following purposes in modern petrol engines:
- determination of reference value of Lambda 1,00 for wide-band probe (calibration of wide-band probe);
- control of the performance of CO catalytic converters.
As we know (very simple explanation), narrow-band probe generates voltage, if no oxygen is present in exhaust gases. So, if Lambda > 1, 00 ( it means, the mixture is lean, there is oxygen present in the gases), the probe don’t generates the voltage; if Lambda < 1,00 (it means, the mixture is rich, no oxygen is present in gases): generates voltage. In very narrow area, the probe works also in linear mode – voltage, generated by it, changes depending from Lambda.
Voltage of narrow-band probe – dependence of mixture (Lambda value).
Wide-band probe allows to measure Lambda (proportion of oxygen/fuel) in wide range: from 0,65 (very rich mixture) to atmosphere air composition (Lambda 30 .. 100). More information can be found in internet, here I will mention only some nuances.
It’s not possible to read the values (reported Lambda) of wide-band probes, using multimeter or oscilloscope. Using these measuring tools (measuring voltage between Pump cell pins), it’s possible to check only, is the mixture lean or rich (if/when Lambda probe is connected to correctly working ECU). Using oscilloscope, it’s possible to see also dynamic changes (for example, unbalance of cylinders fuel mixture), but absolute values can be seen only with special technical solution. This solution stabilizes the voltage of Nernst cell around 450 mV, changing the current via Pump cell. This current recalculated according to the data tables (provided by manufacturer) is the value, which is interesting for us: Lambda. Unfortunately, it’s not possible to measure Lambda by several devices simultaneously, what means – it’s not possible to connect in parallel to working MSV/MSD unit and see, what Lambda actually (before additional reference value corrections – see description below) is measured by wide-band probe.
Control of performance of CO catalytic converter
If the engine performs correctly, in case of Homogeneous mixture, the narrow-band probes (with Lambda 1,00) are maintained not in the middle-point of curve (0,45 .. 0,50 V) as could be expected, but by generated voltage around 0,7 .. 0,9 V. No official explanation from BMW for such situation exists, but, knowing special features of Lambda probes performance, we can assume: in the middle of linear range the sensitivity of probe (voltage vs changes of Lambda) is very high (to both rich and lean mixture), even in case of perfect catalytic converter the voltage generated by probe (even with well balanced cylinders) will look like rectangle. Moving the working range to top end of curve, Lambda probe becomes insensitive to enriched mixture, but its sensitivity to unburned oxygen is on max. Moreover, bit more rich mixture (Lambda = 0,995) provides, that if the little amount of unburned oxygen gets into exhaust system, the catalytic converter is supposed to burn it (because there is a fuel, which to burn). If the catalytic converter doesn’t function (don’t burns the oxygen), after it there can be short-term excess of oxygen, which will create voltage drop in signal sent to DME.
In the picture: the middle point of the narrow-band probe’s active range (corresponds 450 .. 500 mV):
The active range, which is used to control the performance of catalytic converter, is marked (corresponds 0,7 .. 0,9 V):
In case of good performance of CO catalytic converter, the average value of DC of the control probe signal is around 0,7 .. 0,9 V, AC component: insignificant.
In case of poor performance:
a) the amplitude of AC component significantly increases; short voltage drops appear;
b) the voltage of control probe drops down quickly after enriched mixture (high voltage ‘delay’ disappear).
Note: MSV/MSD engine control units can’t differ (quick) pulsation of Lambda caused by poor performance of CO catalytic converter from, as an example, pulsation of Lambda due to uneven cylinders performance. So, if the cylinders individual fuel adjustments (adaptations) are incorrect, or some of cylinders are misfiring, MSV/MSD can register false error messages regarding the insufficient performance of CO catalytic converter.
Calibration of wide-band probes (adapting the reference value)
At the same time with control of CO catalytic converter, narrow-band Lambda probes perform also one more very important task: they calibrate the reference values of wide-band probes (it means, calibrate the wide-band probes, if/when Lambda = 1,00).
There are several reasons, why such calibration is necessary: to avoid technical inaccuracies of probes itself; deviations of electronics (analog part of DME), etc. During this calibration DME adjusts fuel mixture until narrow-band probe report 0.75 V. At this moment reference voltage of wide-band probe is modified in such way the probe reports Lambda close to 1.00.
Correct performance of N-series engine in Homogeneous (with Lambda = 1,0) mixture.
Both wide-band probes are calibrated: narrow-band probes generate around 0,75 V.
If there are suspicions regarding incorrect control of fuel mixture: misfire, trim errors, error messages regarding CO catalytic converters (error messages regarding probes, of course), control activities, as described below, have to be performed:
1. Check the dynamic resistance of Nernst cells. They are displayed in analogue values, choosing OBD and 6-th menu (../F5/F2/F6).
The dynamic resistance of Nernst cells shows the electron-chemical activity of probe – the higher the activity, the less is resistance. Moreover, this resistance is inversely proportional to temperature (in newest generation of engines the dynamic resistance of Nernst cells is used to stabilize their temperature). Correct dynamic resistance – several hundreds Ohm. Unfortunately INPA displays values with step of 256 Ohm (it’s very rough), which don’t allows to see true value of dynamic resistance of Nernst cells. Correct (reported by INPA) values of Nernst cells: 0; 256; 512 Ohm (acceptable: 768/1024 Ohm). Additional control: for cold (unheated) engine resistance between Nernst and Pump cells connections (measuring by multimeter) is several MOhm (MegaOhm), which means, INPA has to display max value (65 KOhm) for narrow-band probe and 8 KOhm for wide-band probes. During several dozens of seconds from the beginning of heating of probe the resistance has to decrease to mentioned above 0 .. 512 Ohm.
Possible problem situations:
a) several dozens of seconds after starting the heating, the dynamic resistance of Nernst cells is above 512 Ohm: the probe has aged, the electron-chemical processes are inefficient. The probe has to be replaced;
b) for cold engine (just started in idle run), for some probe the dynamic resistance of Nernst cells is shown lower than max values mentioned before: there is a current leak, short circuit – if the resistance is shown in several hundreds of Ohm) – the electrical circuit has to be checked. If it is in working condition, the resistance of probe’s sensor has to be measured with multimeter, if the resistance is lowered – the probe has to be replaced.
2. The value of Lambda in exhaust gases has to be measured, using a additional measuring tool. For this cause (true though, with lower reliability level) for N43/N53 engines the NOx sensor values at ../F5/Shift+F2/F1 has to be used (it has to be checked, that sensor is in online mode, because sensor: when turned off, as default value shows 1,00). If the measuring tool shows Lambda value, different from 1,00 (in the same time all probes shows as if correct values), it has to be considered, that some of the probes (or only one existing) gives incorrect data regarding fuel mixture (and wide-band probe is calibrated incorrectly). Of course, it has to checked, that the exhaust system is airtight – but, I believe, that all specialists can do that without any problems.
3. Test procedure of probes has to be performed. In INPA loader 2.023 it can be chosen in diagnostics session (../F9/F3). During this test, injectors opening time will be changed +/-10% for several seconds. Essential things, which has to be observed:
a) min and max values of voltage of narrow-band probes. Min values has to be below 0,1 V; max – above 0,85 V. If any of values (typically – max value below 0,8 V) – the probe is damaged, it has to be replaced;
b) min and max Lambda values of wide-band probes. If min values significantly differ from 0,90, and/or max values significantly differ from 1,10: the probe is damaged and has to be replaced.
4. If test, mentioned before, is not possible by some reasons – check the narrow-band probe’s ability to produce necessary voltage, perform the driving session. In the moment of aggressive speed-up (kick-down mode) the fuel mixture has to become rich, narrow-band probe has to generate voltage above 0,85 V.
5. Evaluate differences of multiplicative and offset STFT + LTFT between banks (if engine has two banks). If the engine has large differences in these parameters (typically: in idle run/offset type LTFT + integrators) – there are reasonable suspicions regarding incorrect functioning of some bank’s Lambda probes. Typical causes – leaking injectors (and as the result – incorrect cylinders individual adaptations/fuel mixture).
6. Check the temperature of the engine. If the temperature is lower than necessary (+ 95 .. 115 oC), the richer mixture has got into exhaust. N42/N53 engines – additionally check the temperature values of exhaust gases and setpoint setting for wide-band probes at ../F9/Shift+F3/F1 (make sure, that it’s 1,00).
7. For N43/N53 engines choose diagnostics menu (../F9/F1) choose Homogeneous mode with Lambda 1,00 (pressing F4). Attention: it is very important not to forget to turn off this mode after test! To do it safely, it is recommended to do it with turned on (and checked) Stratified charge! (Choose Stratified charge: pressing F3, check, if it turns on; then press F5, to exit the menu).
The sample of successful test of Lambda probes:
To make the situation even more complicated:
1. MSV70, MSV80 and MSD80 loader 1,001 don’t shows correct offset and multiplicative STFT + LTFT (no sum is displayed, don’t shows equal/correct measurements etc.);
2. MSV70, MSV80 and MSD80 loader 1,001 and 2,023 don’t shows setpoint values of wide-band probes, if they are different from 1,00, for example, not fully heated engine can have enriched mixture (for example, 0,98), but nothing will show, that such (enriched) mixture is required;
3. MSD80, loader 2.023 shows multiplicative STFT + LTFT correctly, but, for example, don’t shows impact of additional adaptations, which degrades accuracy of these measurements; wide-band probe setpoint values are displayed, but one menu don’t displays: Lambda values of wide-band, narrow-band and NOx sensor, which complicates the viewing of data;
4. MSD80, loader 2.023 don’t shows offset STFT + LTFT (sum), and one menu don’t shows additional adjustments of each injector, which don’t allows to compare proportion of fuel mixture of both banks in idle run;
5. For no engine setpoint differences of wide-band probes from default values (this would help to identify the incorrectly working narrow-band probe);
6. In one menu no Lambda values of wide-band probes and NOx sensors for N43/N53 are displayed, no differences are shown (that would ease the job);
7. No menu shows status of narrow-band probes: do they work in active range or – regularly “falls out” of active range (it would lead to incorrect performance of whole system);
8. For no engine status bits are shown, if wide-band probes are calibrated successfully; there is no literature, under what circumstances/modes these procedures are performed; for example, N53 starts the calibration of wide-band probes in idle run and keeps narrow-band probes in necessary mode only when the temperature of engine has reached the nominal and there are signs, which shows that the movement has started (increased RPM);
9. No engine has control points set, where could be possible to connect to probes and check live data of them.
This list could be continued for long…
The situation (with no error messages stored in ECU memory), when/if any of wide-band probes shows value different from 1,00, but control probe/s – correct (0,7 .. 0,9 V), is more interesting. In this situation – allow engine run idle. If the mismatch don’t disappears for longer while (some 5 minutes) – by some reason the engine don’t performs the calibration of wide-band probes.
For N46/N52 and other “ordinary” engines – delete the long-term trims, create new ones; if the problem persists (and Lambda test shows as if correct results) – switch wide-band probes with places. It’s possible, that the correction range of probe’s (the ones, which sows incorrect value) reference point has reached maximum deviation;
Problems with performing long-term trims for N43/N53 engines are observed in several “interesting” cases: in case of incorrectly registered injectors; if the Stratified charge mode is disturbed (if the engine is left in any of fixed working modes; if the engine has problems with NOx system, and if fuel trim error messages are recorded).