This time the story about classic of BMW engines.

Yesterday I took the opportunity to join an interesting case: E39 with M54B25 engine and, accordingly, MS43 Siemens control unit. I have heard a lot of foul words regarding Siemens control units from people performing diagnostics. For reasons of such negations you have to ask them, because, in my opinion, if you understand, how does it all works – it’s very interesting.

Symptoms: uneven run, misfire, regular error messages regarding one or both banks (fuel trim), Lambda probes – after deleting the long-term trims and attempts to create new long-term trims. In the same time there are cases, when long-term trims are created successfully and no error messages are present in the current moment.

I will start form other end – about, in my opinion, interesting nuances (related to the topic), which are not mentioned in description of MS42/MS43 engine control units.


Control of Lambda probes.

M54 engine uses 2 pieces of narrow-band regulating probes and 2 pieces of narrow-band control probes after catalytic converters (all 4 wire, with separated heating).

MS43 has PWM management for heating the probes. In the same time it is not fully used – the heating of probes is done with full “power”: output keys (by normal on-board voltage and in idle run) are fully open. As mentioned in description for MS42/MS43, issued by BMW AG, PWM is used to heat up the probes slowly (without thermal shock).

Intended probes are with heating elements, which, even if the coefficient of temperature is positive (which provides automatic stabilization of temperature), then it’s so unexpressed, that the probes reach the work temperature only in minute or two.This is much longer than for modern probes (7 .. 15 seconds).

From the other side – there is no place for embarrassment: this management solution is 20 (!) years old!

MS43 measures internal dynamic resistance of Nernst cells of probes (visible INPA ../F6/F3/F5 activation/probe test). For new probes MS43 “sees” this Z in limits of several tenths of Ohm, for used probes – in limits of several hundreds Ohm (heated probes). Unfortunately MS43 doesn’t performs any visible short-term trims (according to this Z) in idle run (will have to check in driving tests, if MS43 tries to not allow the overheating of probes, when Z drops down till abnormally low value, but I have no illusions, judging by wide spread of Z between probes). Even more – it don’t “understands”, when/if the probe shows inadequate Z. MS43 also don’t waits, till Z of probes will drop down to active status: short-term trims/long-term trims starts in exact timing. To complicate the situation even more, bar limits in INPA indications is set in limits of 20 Ohm, but even new probes in idle run don’t reach such level (and exceeded limit indirectly shows, that the performance is abnormal, there is an error or defect).

Accordingly, if the probe after fixed time will have inadequate Z (short circuit: Z close to 0, or chain break: Z close to 65k Ohm), MS43 will not see it and still will try to use this probe.

In the image: measurements of Z for Lambda probes circled with red:
BMW INPA Lambda probe system test menu

Conclusion: Nernst cells of Lambda probes has to be monitored – you have to be sure, that they are in working condition. Otherwise wrong conclusions could be made.
For cold probes: Z of Nernst cells has to be with max limit of range (65K Ohm, according to INPA data; R above 1M Ohm, measured with multimeter – between outputs of Nernst cell, for each output against body and/or heater).

For warmed up probes, idle run: 25..500 Ohm (depending of temperature and aging, INPA indication).


Short-term trims and long-term trims.

After deleting old long-term trims, MS43 starts to create new ones after heating up the probes; this moment can be seen in INPA, choosing ../F5/F6; short-term trim offsets start to change (become different from 0). MS43 uses integrator for STFT, it means, the speed of change (dv/dt) of them is limited in time (1..2% in second).

In the image: menu of short-term and long-term trims.

Top row: integrator of STFT (short-term trims);

Middle row: offset (LTFT) long-term trims for idle run;

Bottom row: proportional (LTFT) long-term trims.
BMW INPA adaptations menu

Initial max correction range of short-term trims +/-28%

Exceeding +/-10..15% limit longer than for 10 seconds (if/when balanced condition/Stoihiometric fuel mixture is reached), forced modification of long-term trims (LTFT) is in progress.

Exceeding +/-28% limit (and at the threshold of +/-28% the stoihiometric mixture is not reached), the emergency situation is detected:

a) system goes to open loop mode (exact bank stops the Lambda probe control, resets short-term trims and long-term trims of exact bank;

b) individual offset corrections of cylinders are switched off;

c) appropriate fuel trim error message is recorded.

It is interesting, that total “depth” of long-term trims is higher than +/-28%, but initial jump of them is limited with these +/-28%. I see following logic: after engine maintenance/repair (and forcing the engine to create new long-term trims) it has to be “more or less” in good condition, in the same time – during a period of time (components are wearing out), the limits of corrections are widened.

One more important number: in idle run correction of +/-28% corresponds +/-2,2 .. 2,4mg offset long-term trims. Respectively, if offset long-term trims, for example, of some bank shows +1.0mg, it means, that short-term trim equivalent is around +12%.

If the current moment short-term trims of current bank show, for example, +5,5%, then total correction is +12 + 5,5 = +17,5%.

Engine in operational condition has to be able to keep fuel mixture within the range by +/-15% long-term trims (the number is based on math and logic), which is around half of the allowed limit, understanding, that +/-10% will be added by short-term trims (without overwriting long-term trims), it means, offset long-term trims has to be in limits of +/- 1,2 .. 1,4mg. Higher long-term trims indicate to to serious inadequacy of originally planned fuel mixture, the reason has to be checked.

The current engine initially had inadequacy close to max level: +2,3mg offset for long-term trims, for this reason, MS43, performing initial long-term trims, regularly “fell out” the limit, but, if it managed not to reach it, fuel mixture control was performed as if normally. And, just to remind, the error message regarding problem (regulating Lambda probe of 1-st bank) is not one to blame.

One more nuance: as in this case initial fuel mixture was too lean, then probe, which actually was damaged (regulating probe of 2-nd bank, it’s internal resistance of Nernst cells, even when the probe was heated, stayed in level of 65K Ohm) and didn’t produced EDS, “behaved” as if normally. Of course, when the problem of fuel pressure (which was reason for lean mixture) was solved, the incorrect performance of damaged probe would be detected, but it would take much more time…

Literature: BMW AG. MS42/MS43 engine management systems.


Related entries:


Management of Lambda probes

Crankcase ventilation