Select a technical image to view it in detail.
This case happened to my “office brother”. If I hadn’t seen the problem myself, I would have been very skeptical, and that is why I wrote this entry. I will try to keep it compact, with few images.
The vehicle and its history
The customer’s car is an F10 with the N20 engine, built in 2013. That means MEVD 17.2.9 (upgraded software) with the electrical wastegate.
The owner had been repairing the car himself. Some time ago he noticed strong engine vibration. Diagnostics (more precisely, reading the fault memory) pointed to the HPFP: an error relating to a drop in rail pressure was stored. He replaced the pump with a new one, but the problem remained. He then replaced all the injectors with new ones. The problem still persisted.
At that point I was asked to “just take a look”.
Symptoms
I re-adapted the engine. There were no error messages in the DME fault memory.
On a cold start the engine starts normally. As soon as the wideband probe starts working (the DME switches to closed-loop mode):
- Idle RPM begins to fluctuate strongly.
- Sometimes the engine even shuts down.
- The mechanical efficiency of the individual cylinders fluctuates, but all cylinders behave alike. The average mechanical efficiency is close to 0 (leveled), and no misfires are observed.
All the essential parameters (air mass, RPM, even lambda) fluctuate so strongly that it is practically impossible to tell where the cause lies.
Once the engine is warm it runs more or less acceptably, but cold it runs very badly. Even a cheap car should not behave like this, let alone a BMW with a new HPFP and new injectors.
First recommendations and diagnostic limits
One of the two basic parameters, the amount of air or the amount of fuel, is clearly completely wrong. My recommendations were:
- Measure the true rail pressure, compare it with the sensor data, and evaluate the stability of the fuel pressure.
- Check the air-tightness of the CCV and the intake manifold.
The aim was to get some reliable basic parameters to rely on. BMW AG has “made it easier” for diagnostic specialists on these engines:
- No offset-type LTFT data are available.
- No integrator data are available.
- No lambda values are available (only voltage is displayed).
- No LPFP pressure data are available (there is no sensor to measure it).
- No data on injector corrections or adaptations are available.
In this case it was not even clear whether the initial mixture was too rich or too lean. The DME managed to reach a correct mixture, but the size and polarity of its corrections were unknown.
The injectors
A day later my colleague called. Blocking the CCV inlet had not given the expected result. But there was other news: the owner had installed injectors intended for EU5, which do not suit this upgraded engine, instead of the required EU6. Presumably he wanted to save some money.
My colleague suspected the injectors. I was very skeptical, because the basic parameters (for example, flow rate) of both injector types, EU5 and EU6, are identical. And I was wrong.
The following tests confirmed that the injectors caused the cold-start problems:
- With new EU6 injectors installed and the engine re-adapted, it runs perfectly.
- With two old EU6 injectors and two new EU6 injectors, the engine runs perfectly. Among the old injectors, two were free of defects; one was strongly leaking and one was dripping, so those were judged damaged.
- With two old EU6 injectors and two EU5 injectors, the engine shows all the typical symptoms.
- With two old EU6 injectors, two new EU6 injectors and the old HPFP, the engine runs perfectly.
So the owner replaced the HPFP and two injectors without justification, and on top of that wasted money on four unsuitable injectors. For me there was no doubt left: the injectors caused the idle fluctuation during the cold start.
But where exactly is the cause?

Asymmetric spray beams
Then I remembered the BMW AG material on the beam characteristics of the new EU6 injectors. The nozzle has 6 beams, but their flow rates differ:
- The two beams directed toward the exhaust side have a reduced flow rate.
- The other four beams have an increased flow rate. The difference between the two is 30%.
While the engine is warming up, the N20 DME creates two injection pulses. The second pulse comes very late, so that part of the fuel reaches the CO catalytic converter and warms it up more quickly.
Evidently, under these conditions the changed injection geometry has such a large effect, compared with injectors of symmetrical flow rate, that it produces very different mechanical efficiency and mixture in the individual cylinders, and the DME cannot correct for it.
It is possible that the newer EU6 injectors can be used in an “old” N20. But it is clear that EU5 injectors can NOT be used in the “new” N20/B48/B58.
Open questions
One more interesting detail: these EU5 injectors were manufactured in 2019, so the “old” injectors are still in production.
Many questions remain unanswered, for example:
- Are these EU5 injectors rejected (untrimmed) EU6 injectors?
- Are the EU6 injectors a “middle generation” with asymmetrical beam flow rates?