Misfires and Lambda probes

This time about nuance, which seems confusing to many. I received several e-mails with questions regarding the situation, described above, so I will try to explain it.


Description of problem: if the engine has misfired in any of cylinders, the corresponding banks have the following problems:

a) wideband probes are incorrectly trimmed (typically – the mixture is rich);

b) the error messages regarding trim of wideband probes;

c) the error messages regarding the signal of control probes;

d) the error messages regarding fuel mixture even in a situation, which actually is correct

are possible.


Why misfires have such catastrophic consequences?

The situation is especially dramatical if the wideband probes are used as regulating ones (modern engines has exactly that).


The main cause of this problem is:

a) wideband probes “see” both the oxygen in the exhaust and unburned fuel;

b) narrowband probes “see” only the oxygen in the exhaust.


Example: readings of Lambda probes of MSD80:

1. Lambda value (wideband probe);

2. voltage (wideband probe, output of measuring circuit);

3. voltage (narrowband probe; input signal).


In case of correctly working engine (N series, including N43/N53, working in Homogeneous mixture mode), when driving evenly, DME (description – simplified, just part of algorithms) controls and manages Lambda probes in following way:

a) tries to maintain the average fuel mixture of each bank, so the voltage of the control probe would be around 0.7 .. 0.8 V;

b) customizes the module of electronic wideband probe management in such way, that the voltage in conditions, mentioned before, corresponds 1.50 V (Bosch DME) or 2.00 V (Siemens DME). The current via Pump cell, in this case, is 0.0 mA + correction (added by DME) and Lambda, measured by the wideband probe, is 1.00.

In the same time DME controls indications of probes:

c) evaluates correction, applied to a management module of the wideband probe, mentioned before. These data, unfortunately, are not available for users. Only in case, if the correct indications are not reached with max allowed corrections, the error message regarding wideband trim problems is recorded;

d) if DME considers, that the wideband probe is trimmed correctly, but, for example, in case of the rich mixture the control probe signals regarding lean mixture (its voltage is close to 0), the error message regarding signal of control probe (DME considers, that the probe is damaged) is stored.


And now let’s look at the situation, when any of cylinders works incorrectly – it has, for example, damaged ignition coil.

As an example, we will look at 4 cylinders N43 series engine with misfires in cylinder No.1 (of course, the situation will be similar for all N series engines).


If cylinder No.1 has misfires, at this moment in the exhaust of 1st bank gets:

a) 50% (of all injected) burned fuel and 50% (of all sucked in) “burned” oxygen – because cylinder No.2 works correctly;

b) other 50% (of all injected) fuel and 50% of unburned fuel – unburned fuel mixture of cylinder No.1.

As the wideband Lambda probe “sees” both fuel and oxygen, it’s indications will be somewhere “in the middle” – it means, it will take into account all unburned components. Of course, the precision of measurements is very low (it’s not a normal composition of exhaust gases, which should be measured by Lambda probes), but wideband probes are able to measure such composition.


Further, the unburned fuel and air (which contains the oxygen) get to CO catalytic converter, where the part of the fuel is burned. Of course, the “power” of the catalytic converter is much lower than would be necessary to burn 50% of fuel, and as a result, part of fuel and oxygen gets to control probe.

As the control probe “sees” only oxygen, it reports, that the mixture in the exhaust is lean – the voltage drops below 0.1 V.


At this moment DME, seeing that the wideband probe displays appropriately correct fuel mixture, but control probe – lean, takes a decision to recalibrate the wideband probe. DME slowly enriches the mixture, to reach such proportion of air/fuel, so the control probe would confirm, that the mixture is Stoichiometric. In case of the Stoichiometric mixture, the voltage of control probe increases to 0.7 .. 0.8 V.


If misfires continue, the enrichening of fuel mixture doesn’t help – a large amount of the oxygen gets to the control probe, and it still indicates, that the fuel mixture is lean. Step by step, DME get to the margin of a wideband trim algorithm, when the max allowed correction is reached (for DME of MSV/MSD series this threshold is around +/-0.30 V) and – records the error message regarding trim problems of corresponding wideband probes.


DME can “go” even further in its conclusions – when it sees, that even in case of strongly enriched fuel mixture the voltage of control probe stays low (close to 0), it can make a conclusion, that the control probe is damaged and the error message regarding damage (incorrect signal) of the control probe is recorded.


Depending on software and coincidence of consequences of current DME (for example, if misfires are in both banks and control probes of both banks indicates as if a lean mixture), DME will also make a different conclusion – the fuel mixture actually is lean! And the error message regarding is recorded!


Management units of Siemens have better success with the processing of misfires – they better indicate misfires and in case of them reacts more correctly. Bosh management modules are worse in this – very often they record unjustified error messages regarding Lamda probes, individual adaptations of cylinders are “collapsing” much quicker, permanent disbalance of mechanical efficiency is created. It is very possible, that the difference hides in the obstacle, that Siemens owns the patent, which determines, how to use flywheel speed to indicate misfires.


The situation gets even more interesting, when we realize, that no misfire counters are available for N series engines, using INPA or ISTA D. Starting with N43/N53, misfire counters can be seen, using an ELM327 adapter (OBD mode 6), but for an older engine – these data are not available!


Sentence of this entry – if any of cylinders has misfires, it’s just normal, that DME has recorded false error messages regarding probes and fuel mixture.


What to do in a situation, when the DME memory is full of error messages regarding cylinder misfires, Lambda probes, fuel mixture?


If the misfire counters are available, check them. Accordingly, if there are any cylinders, in which misfires (relatively again others) are not observed, the first conclusion – obviously, the fuel mixture in the bank limits is correct enough, with a chance of 99%: second (others) cylinders of the same bank has/have ignition problems. The remaining % is “given away” for leaking injector.


If misfire counters are not available, eliminate as much as possible, the visible problems with ignition system (replace spark plugs, if they are worn; replace ignition coils, if the defects are visible).


Check performance of all Lambda probes, performing driving session, as described here:

a) check the ability of the control probe(s) to generate the voltage;

b) check the voltage of wideband probes.

If voltages of all control probes correspond to normal, there is a reason to believe, that the error messages regarding probes are consequences of misfires, but not – the error messages are the consequences of incorrect fuel mixture (defects of probes).


How to indicate the damaged cylinder, if misfire counters are not available?

a) delete error messages, perform quick re-adaptation of the engine;

b) try to ensure for the engine the working conditions (temperature), when the misfires are present in idle. Observe the indications of control probes. In the bank, where the damaged cylinder is located, the voltage of the control probe will drop (some time will be close) to 0 V;

c) observe Rough run data. Cylinder, which has misfires, will indicate the short-term efficiency of 0.

Note: before interpreting the Rough run menu, find out the sequence of cylinders, “polarity” of a bar, and values, to which 0 efficiency corresponds. Description of Rough run you will find here.


Unfortunately, the viewing of Rough run data requires patience, the engineer has “catch the moment” when the misfires are present. When the suspicious cylinder is indicated, switch its spark plug and ignition coil with any of cylinders of the another bank. If the problem, confirmed by control probes, will switch the bank – the guilty cylinder has been found!


Unfortunately, if both banks have problems – misfires are in at least 2 cylinders and analysis don’t give any clarity: replace all ignition coils and spark plugs (if they are worn).