N53B30. When adaptations fail

This post best describes N43/N53 series engines – their complicated engine management, specific nuances in management performance and diagnostics.

The patient: N53B30. For few hundreds kilometers the engine became manifestly very whimsical – vibration in idle, rough switch of performance modes (for example, from low load/Stratified charge to open throttle/Homogeneous mode), subjectively it also felt that the torque has became much modest; all symptoms mentioned before progressed very rapidly. Misfires appeared in different modes. In other words situation can be describing as degradation.

 

How much would cost to fix such defect? EUR 100? EUR 500? EUR 1500? Depends on diagnostic specialist’s qualification.

First stage: check the memory for error messages. The error message regarding control Lambda probe of second bank has appeared. The status of error message – passive, the engine performs also in Stratified charge mode, everything look’s OK. The status of error message regarding uneven running – also passive, Stratified charge performs with Lambda till 3,0 .. 3,2.

Detailed description of error message say’s that in case of rich mixture (which is required by MSD80 and approved by control probe) the voltage of control probe was only 0,10 V, which of course, is out of range (voltage has to be above 0,80 V).

Checking the menu of long term trims, we manage to catch a moment, when the reading of wide-band probe and control probe are contradictory.

Wide-band probe reports Lambda below 1 (it means, the mixture is rich), but voltage of control probe is only 0,17 V.

Has something happened to control probe of second bank?

Perform the test of probes. Here are the results:

As seen from test results, the control probe can generate proper voltage, also range of wide-band probe is within the normal range. The problem, obviously, is somewhere else!

 

When we check the history memory of error messages, one more very interesting error message appears: the error message regarding insufficient performance of second bank’s CO catalytic converter. The catalytic converter also damaged? Let’s keep a little intrigue, other error messages are covered – we will come back to this a bit later.

Keeping in mind the experience with diagnostics of N series engines, I don’t hurry to blame the CO catalytic converter. Of course, the dynamic resistance of Nernst cells for all probes – they are within normal range, the heating of probes – as it is supposed to be. Lernbits: all segments adapted (including individual corrections of Lambda of cylinders).

As you can see in the previous images, both offset and multiplicative long term trims are within normal range, integrators do their work: the engine performs in closed loop mode.

 

But strange things continue to appear.

As you can see in Rough run menu, indicator bar of 5nd cylinder is very different from 0 (cylinders are marked in firing order): efficiency is lowered.

Allow the engine to run idle for 5 minutes,  but situation doesn’t changes to better – the engine don’t even tries to eliminate uneveness detected by itself!

When we check the individual long term-trim data (multiplicative corrections) for fuel injectors, the instantaneous trim of injector No.6 looks very striking. It is very close to max value (83%), in some conditions even overreaches it.

In the same time, the opening time of injector for cylinder 5 is relatively short! How is it possible (keeping in mind, that measured efficiency is lowered – why MSD doesn’t fix this problem?)?

 

Here we have to keep in mind two significant nuances, which are not documented:

1. Positive shift from 0 in Rough run menu bar actually means insufficient mechanical activity of cylinder (not increased, as it seems more logical);

2. The order of injector Rough run menu shows firing order, not the order of placement. Obviously, INPA shows cylinders in the same order as MSD80 MCU registers cylinders.

So the sequence of cylinders in Rough run menu is following:

1=1

2=5

3=3

4=6

5=2

6=4

(Note: for N43 engine, accordingly, data will be shown for cylinders 1/3/5/6, which correspond to 1/3/2/4).

This all brings to conclusion: 5-nd cylinder has some problems, it’s efficiency is insufficient, and also the 6-th cylinder has huge additional trim. Injectors? Valves? I don’t think so, because, for example, Stratified charge in idle run (with Lambda 2,0), the situation is normal (both mechanical efficiency of 5th cylinder and short-term trims for injector of 6the cylinder, and the engine runs even).

When we check individual data of instantaneous short-term trims of injectors, the situations becomes even more interesting. As we see, data of all cylinders are significantly different from 0; cylinder 5 – actual efficiency -16,1%! Cylinder No.6: +11%… Such difference from the moment of previous test (see the description of algorithm below)? And in the same time – long-term trims are completed? Absolutely abnormal situation!

 

MSD measured (and confirmed via Rough run data) insufficient efficiency of cylinder 5, but the opening time still is so short?

It comes out, that this huge additional short-term trim (as in the case with cylinder No.5 and 6) actually is not applicable! It means, INPA (more precisely – MSD80) shows desired as existing! We can make sure about this when we start a new session: the long opening time for injector of cylinder 5 is “lost”- this huge additional short-term trim actually was not applied!

 

Brief about performance of integrating (cumulative) long-term trims of additional corrections.

In different performance modes the engine regularly performs measuring of individual “contribution” of injectors (cylinders), it means, temporarily enriches the fuel mixture in some of cylinders and measures the contribution of this current cylinder (changes in contribution) in total Lambda and calculates the difference between real and ideal situation. The results of tests are summarized in first 3 lines (menu: Adaptation/injectors for Lambda regulation, ../F5/Shift + F6/F3). After the test ends, these values (in “spare” time) are summarized with corresponding group of data of each injector (for example, data measured while the opening is short – to group of data of short opening time etc.). After deleting the long-term trims (and during the process of creating new ones), while the ideal situation and actually measured situation are totally different – corresponding learnbit is passive (not colored – the creating of initial long-term trims is in the progress), if the situation is close to perfect (it means, bars of first 3 lines are close to 0,0) – learbit is colored (the long-term trim is completed successfully).

In this situation learnbits are colored – the individual additional corrections of injectors “are” successfully completed.

In current situation, MSD80 has detected huge differences between desired and existing, but, for unclear reasons, hasn’t added the values to corresponding group of data. Even more – it requires +16,1% additional corrections to ” defective” cylinder No.5, but the opening time still is very short. It is obvious, that something does not work as it should.

 

It’s clear, that these huge differences of Lambda between cylinders can be the reason, why fake error messages both for CO catalytic converters and Lambda probes are diagnosed.

In this current case the difference of fuel mixtures between cylinders reaches even 30%, no Lambda probe is able to keep the optimum fuel mixture in such conditions, no CO catalytic converter – to ensure low CO/CH.

Now it’s time to reveal the rest of error messages in error message history. It turns out, that before several hundreds of km, while performing the maintenance of car’s exhaust system, the NOx sensor was disconnected for several times. The time-out didn’t reached over 5 minutes, but, as you can see, the error message memory contains several messages (logically – regarding time-out: the data exchange with sensor was interrupted). After the maintenance the error messages were deleted and didn’t appeared anymore (there was no reason for them to appear – the sensor works correctly).

But, as experience with N43/N53 shows: any error messages (even if they are deleted) leaves an impression to creating and continuing of long-term trims.

This is a very good sample, how after completely unrelated error messages of NOx system (in addition – after solving the problem and deleting the error message successfully):

* application of individual corrections of injects is stopped;

* when unbalance of cylinders grows, misleading error messages regarding CO catalytic converter and performance of control Lambda probe are recorded;

* the registers of injector corrections “fly away” uncontrollably out the permissible limits (and actually they are not even applied – the same as in case with non-coded injectors);

* there are no symptoms (no error messages, no learnbits), that something is out of state;

* the engine starts to vibrate, shiver, it looses power,  the performance is uneven.

This situation is one step from trim error messages of banks, error messages regarding wide-band probes, error messages for cylinder trims and error messages regarding misfire.

Exactly in this way all problems of N43/N53 does start, if there are error messages (even with passive status, even deleted even deleted and the engine is in perfect technical state) regarding NOx catalytic converter and/or NOx sensor.

 

In such situation, most common strategy of repair is to replace injectors and ignition coils, although they are in excellent condition.

Besides, as mentioned in the beginning of this post – the NOx system is in perfect order, the engine performs in Stratified charge.

 

After evaluating the situation, the decision is made – corresponding long-term trims of 2-nd group are deleted, using INPA (long term trims of Lambda Probes, individual long-term lambda trims). Like in many other cases, also in this – no information from producer, what status bits are deleted, they are not also visible. As experience shows, deleting of old long-term trims renews also very specific status bits, which, obviously, delete error messages regarding creating of these long-term trims, overfilled registers etc.

After deleting long term trims and performing 2 short (each 5 minutes) driving sessions, the amount of correction for cylinder No.5 and 6 normalize, the amount of individually applied short term-trims normalize, appropriate learnbits are colored.

 

Right after deleting long-term trims, the engine starts to measure the contribution of injectors:

The results of repeated test in the same driving conditions gives following results:

As we see, the data of injectors are measured correctly, difference form perfect situation converges to 0,0; it means, long-term trim data are correctly applied and restored.

Scrolling the menu down, we see:

The engine has performed the measurements of injectors productivity both for short and long opening. Both test results are close to perfect, it means, the differences converge to 0,0 (not reaching above 0,3%) – the long term trims work correctly.

Rough run menu shows correct performance of all cylinders in idle run Homogeneous mode.

The amount of corrections for No.5 and 6 after creating of new long-term trims:

The engine runs smooth and even again, previous performance & power are back!

 

Conclusion: nor Lambda probe, nor CO catalytic converter was damaged (there were error messages regarding these issues). The injectors were not damaged – but the engine applied huge individual short-term trims. The ignition coils were not damaged, although the misfire shows the problems of ignition system. For MSD80 is “not enough” with “clean” error message memory – after specific error messages complete deleting of according long-term trims is required, only then normal algorithm of the engine performance restores.

 

Average service center would replace (unjustified) following spare parts:

1. control Lambda probe (or even both – control and wide-band Lambda probe) of 2-nd bank, because the self-diagnostic system points directly to it (and live data confirms it);

2. CO catalytic converters (because the self-diagnostic says, that their performance is weak, also the level of CO and HC in exhaust gasses would be increased, but – not because of damaged CO catalytic converters, but the reason would be abnormal fuel mixture in cylinders);

3. performing the test with ISTA D, at least one (typically – several) cylinder will be marked as “guilty” for misfire, the ignition coil would be replaced (and typically – also spark plugs). When there will be no result (because the ignition coil, which was replaced as damaged, actually is in technical order), other “guilty” injector would be replaced.

If the long-term trims would be deleted after replacement of injector (as required by ISTA D), of course, the situation would improve, because the normal creation of new long-term trims will restore. And there would be one more case, which was “solved” only with complicated and expensive repair works.

 

Related entries: