The term “leaking injector” is used to describe different defects of injectors. But this time we will talk about the most popular of possible cases – some hard microparticle getting in the nozzle of the injector.


Even premium quality petrol happens to have some additions – sand, rubbish, additions of oil products etc., not for no reason the pumps in the petrol stations and also cars itself are equipped with fuel filters. Unfortunately, not all foreign objects are filtered, some of them reach the fuel supply system and also injectors. Some of them are successfully “injected” in the combustion chamber but in some case they stack in the injector itself.

If the foreign object gets stuck in the nozzle of the injector, the is not able to close completely – also in the closed condition the fuel still leaks via a nozzle.

Sometimes these foreign objects are “flushed out” by time and the problem disappears as quickly as it appeared.

If the defects were observed for an only short period of time, the car owner can not even notice this problem; if the defect was for a longer time, most probably, MSD has switched to one of the emergency modes, which requires re-adapting of the engine to restore its normal performance. In the last scenario – it’s enough to re-adapt the engine, no replacing of the injector is required.


This case is very instructive because suddenly the injector with less wear of all started to leak. The current engine has nozzles, which have performed for over 300’000 km, but the defect appeared exactly for this one with the wear of around 120’000 km.


Beginning of the problem.

Increased permanent vibration in idle was observed, after a while – the idle became uneven, separate misfires were observed.

When opening INPA, was discovered: no error messages recorded (yet), but following abnormalities are spotted:

offset value of the 2nd banks LTFT is strongly negative (around -1.30 mg/stk, contrary to offset LTFT of the other bank), which means – for some reason the total amount of the fuel has to be decreased for all bank 2. This is one of the symptoms, which says – one of the injectors is injecting too much fuel.

Of course, in the next step, we check adaptation data of the injectors.

Opening the adaptation data for injectors in idle, Stratified charge (the engine has installed NOXEM, and not taking in account the respectable mileage, it still runs in Stratified charge), we see, that the adaptations of injector of cylinder No.5 are strongly negative, it means – it’s opening time is strongly reduced. This is a symptom, which indicates the injector, who has leaking problems.


The engine has switched to Homogenous mode, we continue the inspection with Rough run data.

As we see, the mechanical efficiency of cylinder No.5 (menu: cylinder No.2 by firing order) is strongly increased. For this current software release, the 0 efficiency corresponds to around +6 units, so – the efficiency of cylinder No.5 is increased by around 50 % (compared with cylinders 4 and 6). Obviously, the fuel amount, injected in this cylinder, is significantly higher than in other cylinders.


At the same time – live data confirm, that the opening of the injector is even smaller (for 5 .. 10 %) than for other cylinders. Obviously, due to total reduced fuel amount for a 2nd bank, in the cylinder No.4 or 6 misfires have started (because the fuel mixture in these cylinders is lean), permanent vibration – due to the increased mechanical efficiency of cylinder No.5.


Switching the engine is Stratified charge:

Rough run shows, that the efficiency of cylinders has almost evened (subjectively – the vibration has disappeared), but live data confirm, that the opening time of cylinder No.5 in this mode is also 2 times smaller (0.10 ms in the opposite of 0.20 ms for two other cylinders). So, with around 2 times shorter opening the efficiency of the cylinder is equal to “normal” cylinders!


After a while, the error messages regarding fuel mixture maintaining problems of the 2nd bank and reaching of the max correction range of cylinder No.5 are recorded.


Here we have to take into account following nuances in the performance of MSD80:

  • algorithms of creating the adaptations of the injectors for Stratified charge and Homogeneous modes are completely separated;
  • for Stratified charge mode, wider “corridors” of allowed adaptations are intended, swifter change of the flowrate (and, accordingly, also in the opening) are allowed;
  • for Homogeneous mode, more narrow adaptation “corridors” are intended, creation of new individual adaptations in idle are immediately turned off by swifter changes of injector parameters – MSD works with last available “correct” data of injectors.

So – we have several parameters, which confirm, that the injector of cylinder No.5 injects an increased amount of fuel in the mode of short openings (for example, in idle mode). Hoping, that the injector will be able to get rid of the foreign object by itself, it was decided to wait for a little – no swift movement made.


Next day.

Staring cold engine (+6 oC), everything looked very correctly.

Here, the opening of the injectors:

Here, Rough run data:

As if everything looked very acceptable, but when the engine warmed up, the situation was getting worse. The explanation is very simple – cold fuel has higher viscosity, it flows “not so good”. Accordingly – it flows badly also via the damaged injector. Additionally, cold engine required even 2 .. 3 times more fuel to maintain the idle, accordingly the effects of the defect reduces for 2 .. 3 times. When the engine warms up, the opening times are reducing, the fuel gets “more liquid”, the impact of the defect – increases.

When the engine warmed up, the defect fully restored and didn’t disappear during following several motor-hours, so the decision to replace the injector was made. The existing injector was cleaned and stored as emergency replacement.


Several more nuances.

If the leaking of the injector is detected, check the live data of injectors, their multiplicative type adaptations. In this case – the value of multiple adaptations of cylinder No.5 is correct. That’s good news.

The second menu, worth to check – the chemical tests of the injectors:


Data in these menu shows relative differences (changes in cylinder tests). As we see, measuring injector (both for short and long openings) haven’t significantly changed since previous tests. These are two more confirmations for a fact, that the injector doesn’t have any serious defects – if the cleaning will succeed to get the foreign object out of the nozzle, the injector will be able to serve for a long time.


And finally, some nuances, which can mislead you.

At first – pay attention to the voltage of the control probe for the 2nd bank.

It is increased, because the correct voltage: around 0.75 V. Control probe indicates rich mixture (richer than Stoichiometric), because part of the fuel, which was injected in the exhaust (but was not mixed and indicated by wideband probe), has mixed in the exhaust, it has burned the small excess of the air in CO catalytic converters. Additionally, MSD, when detecting such situation (control probe indicates too rich mixture), recalibrates wideband probe in way, that it’s indications start to correspond to this enriched mixture, it means, it (wideband probe) indicates Lambda 1.00 (or voltage of 2.00 V), even if the real Lambda actually is reduced (the mixture is rich).


Here, confirmation for rich mixture:

The average Lambda is 0.96, it means: 2nd bank works with rich mixture (with Lambda around 0.92; at least 8% of fuel left-over).

And here – the moment of a test of the Lambda probes:

As we see, even by as if the correct indication of the wideband probe of Lambda 1.11 (lean mixture, air left-over around 11%), control probe still indicates rich mixture (it’s voltage around 0.80 V). This time it’s not the problem with probes, but only consequences of leaking injector: incorrect calibration of the wideband probe!


Recalibration of the wideband probes.

At the 2nd day of the defect, described above, just after warming up, the wideband probe of the 2nd bank indicated Stoichiometric mixture (2.00 V; Lambda 1.00), but control probe – slightly decreased voltage: around 0.60 V (the voltage is as if slightly leaner than necessary). This decreased voltage is a consequence of the impact of defect of the previous day – the wideband probe is incorrectly calibrated.


Situation after 2 minutes.

As we see, now the wideband probe is calibrated correctly: MSD has performed it’s task perfectly. It indicates Lambda 1.00 (2.00V), and the voltage of the control probe is exactly 0.75 V.


The sentence of this entry:

  • the hard microparticles, which are mixed in fuel, can cause the leaking of the injector. Piezo injectors are more “sensitive” to this than injectors of “old type”;
  • such problem (this is not even a defect of the injector) can happen at any moment, even for brand new injector directly after installation;
  • it’s quite reckless in case of any such problem immediately replace the injector, or even more – replace all injectors. If there is an option to clean the injector – use it. It the cleaning succeeds – the injector can be used, this defect doesn’t give any negative impact to further function of the injector;
  • leaking of the injector causes many related problems – incorrect signals of the probes, error messages regarding maintaining the fuel mixture, misfires, etc. So the diagnostics have to be performed carefully and thoroughly.

Correctly performed diagnostics in case of problems with injectors can save you several thousands of EUR! Additionally, the most common problem – micropart get in the piezo injector – usually can be solved with the cleaning of the injector and don’t leaves any negative impact on the further performance of the injector.


P.S. Some screenshots after repair (injector – replaced and encoded; engine – re-adapted)