In this entry – about an interesting situation, which I have encountered already often enough to make it worth to mention.
Patients: N43 series engines in the simplest configuration: with one wideband probe.
Situation: during the cold start (rarer – for warm engine), directly after starting the engine, relatively small/acceptable disbalance of the mechanical efficiency of cylinders is noticed, but when the wideband probe starts to work, this disbalance does not disappear or stays unchanged, but increases more and more. Soon misfires start, the engine switches off as if damaged cylinders. The error message regarding disconnected cylinders, their misfires are recorded in the error message memory of DME. Components of the disconnected cylinders (spark plugs, ignition coils, injectors) are not damaged. Typically – no other error messages in the DME error message memory.
Example: directly after cold start (while the wideband probe does not works) situation looks as following:
“Left side” (1-st and 4-th by placement) cylinders indicate for about 10% reduced mechanical efficiency;
“Right side” (2-nd and 3-rd by placement) cylinders indicate for about 15% increased mechanical efficiency.
To exclude the impact of the individual adaptations of injectors, all adaptations are deleted (it’s confirmed by the flywheel learnbits in the current menu).
When the wideband probe starts to work, this difference slowly (during 5 .. 10 seconds) increases to +/0.080 units and even more (+/-40%), after what the cylinders No.1 and 4 (by placement) start to misfire and are disconnected.
What happens in this situation?
DME has lost its mind? It does everything to “break” more or less correct performance of the engine?
Cause of the problem: this time the injectors of the “right side” are leaking for a cold engine.
Till the moment, when the Lambda probe starts to work, the fuel mixture in the cylinders of the “left side” is more or less correct (the engine works in open-loop mode, the fuel mixture is injected according to the energetical model and the adaptation maps), on the “right side” – rich. Cylinders of the right side have slightly higher mechanical efficiency because there is a fuel leftover. Rich fuel mixture burns well, no misfires.
When the Lambda probe starts to work, it sees the fuel leftover (total, measured in all cylinders) and starts to reduce fuel in all cylinders proportionally. In the cylinders of the left side, the fuel mixture gets more lean, on the right side – more correct (it is still rich).
At the same time, indicating, that the situation with fuel mixture has changed radically, DME forcibly rewrites offset type LTFT (with a strongly negative value, typically: -1.0mg/st.), Integrators indicate negative values, but – can reach more or less correct values after rewriting LTFT.
Smoothing out the average fuel mixture, it becomes, that the fuel mixture in the cylinders of the left side is so lean (Lambda above 1.3 .. 1.4), so the misfires start in them. When detecting the mass misfires, DME switches off “damaged” cylinders.
In all similar cases, the reason for the situation was leaking injectors. As if simple. But, there is one more nuance – the version of the software. In all cases, the methods of the solution “improved”, when the software was updated.
What do the newest SW releases do in this situation?
1. Did not perform swift rewriting of the offset type LTFT to the negative values, making diagnostics easier and not making the situation worse in the next time of starting (if the injectors are not leaking at that moment). Not-rewritten offset type LTFT did not allow the performance of leaner fuel mixture when Integrators try to “lean” the fuel mixture;
2. did not allow such deep total fuel mixture corrections to the “lean” side – did not reduce Integrators till -30% (min value limit) and then was checking, what is happening to the fuel mixture, but by already relatively correct Integrator values (for example, -10%) when analyzing measured Lambda and reacted on time. For example, if by Integrator -10% Lambda is below 0.80, it is clear – even by -30% of Integrator will not be able to reach the Stoichiometric mixture. It is not correct to continue to reduce the amount of fuel, because it can lead to misfires – obviously, there is some fundamental problem in any of cylinders. In this situation, DME immediately recorded the error message regarding the fuel mixture (too rich), disconnected closed-loop mode and switched to the mode of using only the adaptation maps. As a result – the fuel mixture of correctly working cylinders stayed correct, all cylinders continued to work.
It must be recognized, that these improvements improved the situation radically.
At first – in the DME error message memory, correct error messages regarding rich mixtureare recorded. It is clear – so: the injector(s) was leaking.
At second – DME did not provoke misfires in other (correctly working) cylinders but ensured the more or less normal performance of the engine.
Sentence of this entry: it is worth to upgrade the SW of DME for N43 with one Lambda probe, as well as remember – if the engine has only one wideband probe, it is common for all cylinders, not taking into account, that DME as it creates two banks. Actually, these banks are virtual, hardware problems are equally “harsh” on all cylinders.