One thing I love about our industries is that there is always something new to learn. Despite what some may claim, there will always be something to surprise you!
This rang true with a 2017 Mercedes Arocs truck with 309,114km on the clock, which flagged a fuel pressure-related fault. When the fault occurred, the engine warning lights would appear, followed by the truck going into limp mode. The issue had been ongoing for around 3 months, where it was difficult to diagnose, due to its intermittency making it difficult to replicate. As the fault developed it was easy to replicate the issue and diagnose correctly.
The main fault codes present were:-
- 9D000A – upper-level limit value for rail pressure regulation has been reached
- A4000F – high-pressure fuel circuit not OK
Interestingly, the freeze frame data that came with the fault code indicated that there was around 200 bar of pressure difference between the desired and actual. It is also worth noting, that there was no indication of an issue with the low-pressure supply side when the fault was set.
Prior to this visit, the truck had already been in and out of the workshop for several other issues. These issues resulted in parts being correctly replaced, although it is easy to assume the parts cannon had been fired, however, this was not the case. The first problem was the rail pressure sensor, which was replaced due to reading 13 bar of pressure when the engine wasn’t running. Secondly, there was an issue with the idle speed as the engine would hunt and not idle correctly. In addition to this, the truck wouldn’t run through the high-pressure fuel test available through Xentry. The issue was found to be the inlet metering valve which had become stuck, causing the erratic idle. A replacement high-pressure pump solved this issue and allowed the test to be complete. Following the high-pressure test it was reported that injectors 1 and 6 required replacing. The injectors were replaced and the high-pressure test ran again. This time, the injectors had passed and all seemed OK. However, the fault returned.
One thing that was noticed during the high idle, was a squeak coming from the high-pressure injector common rail. This happened at the same time as there was a change in the rail pressure. Once the deviation between the actual pressure and the desired pressure was outside the limit, the fault codes were set.
The next task on the list was to determine if there is a relationship between the noise and the rail pressure. During the setup, we tapped into the spill return from the cylinder head and added the sight block. By doing this, it would allow us to measure the pressure and observe the fuel flow when the fault was present. The thought here is the possibility of air in the fuel system, which may be the cause of the noise and having the sight block in place would allow us to see any air bubbles.
In order to help identify the source of the noise, we made use of the accelerometer from the NVH kit to help capture the noise and isolate it to an area.
As you can see in the images below, we have captured each injector, along with the rail pressure sensor voltage and current from injector 1.
The accelerometer was attached to each injector pipe, and when the noise was present a reference for each cylinder was made. Here, we can see that there is high-frequency noise present on cylinder 1, cylinder 3 and cylinder 4. In addition to this, we can also see that there is a repeating pattern in the fuel rail pressure sensor, which is showing some very erratic behavior. Given we have some direction, we included the spill return pressure whilst observing the flow through the sight block.
A word of warning on these engines, an additional power booster is fitted to each injector, which will allow injection at higher pressures when commanded by the ECM. This power booster has a separate return gallery, so making sure the sight block is on the right outlet is key to ensuring we are looking at the injector needle return.
Adding pressure from the spill return provided some interesting results.
We can see there is a repeating pattern and zooming in shows this more clearly:
We do have to be careful when chasing fuelling issues on modern vehicles, as it isn’t quite as easy as using the firing order to work out which cylinder(s) we should be looking at. Given the ECU can control the timing of injection, it can also compensate for poor-performing cylinders and correct fuelling for the other cylinders. If we add the firing order to the above capture we could say that there is something odd with the spill return for cylinders 2, 3, 4 and possibly 6. What follows on from this, is gaining an understanding of the waveform. From the information displayed in the waveforms, we should consider two options, a bigger pulse which indicates a higher flow showing a failing injector? Or, where there is less of a pulse because the injector is over-fuelling?
Given that the sight block is in place, we were able to see that there is a considerable amount of air present in the return.
From this, are the bubbles associated with the squeaking noise found earlier, and how are the bubbles getting into the return line.
With the sight block still in place, along with the stop tap, we tried restricting the return. As the WPS500X was already in place, we could ensure the pressure never got too high. What we discovered, was that when the pressure started to increase, the noise stopped and the rail pressure stabilized. What we also saw was an odd pattern in the return line that repeated.
Whilst analyzing this in more detail, there are some comparisons we can make, as shown below.
- We can confirm that when the noise occurs there are bubbles in the return.
- The rail pressure fluctuates heavily when there are bubbles in the return.
- The pressure in the return line when the noise occurs is around 0.5 bar.
- When the return is capped and the pressure rises to around 1 bar, the noise stops and the rail pressure stabilizes.
Due to the rail pressure stabilizing, we’re now less inclined to think that there is air getting in from the supply side, which means air is getting into the return from the high-pressure side. Other than the high-pressure pump (which has already been replaced) and a porous head (not impossible) the only place remaining is the injectors. Before going down this route, we thought it might be worthwhile to give someone a call. Sometimes product knowledge really helps and after the call, we learnt that this noise is typical when an injector sleeve fails. When a sleeve fails it will allow air from the cylinder to enter the fuel return gallery. During the compression stroke, the air pressure can creep into the fuel line, alter the pressures inside the injector, thus resulting in a noise, and of course, introduce air into the return.
The big question though, is which injector, or injectors, are causing the issue? In order to replace an injector sleeve, the cooling system needs to be drained down, so the decision was to do all 6 sleeves whilst the coolant was drained. This would allow us to take a look and inspect each injector. This resulted in us finding something interesting with injector4 – which was one of our initial suspects.
For comparison, the image shows injector 2 on the left and injector 4 on the right. You can see that injector 4 has some discoloration, however, it was only when we removed the sealing washer we noticed the following:
In the image on the right you can see a crack in the injector body. This may or may not have had any effect on fuel pressure or the return pressure, but either way, it needs to be replaced.
With injector 4 and all 6 injector sleeves replaced, the truck was reassembled, and with the cooling and fuel system bled, it was ready to start. Whilst attempting to keep everything the same, the rail pressure and fuel return pressure are on different channels, but the results prove that something has now changed.
The display image indicates that there is no more erratic fuel rail pressure (now in green). The return pressure is also showing as stable, with some more defined pulses, which likely tie up with the injector spill and confirmed there were no more bubbles in the return line. You could say that cylinder 6 has a slightly higher pulse than the others.
We should also note the return pressure is now a steady 1 bar. As we saw when capping the return, we would get to 1 bar of pressure and the noise would stop. So, how come when the vehicle was faulty, the pressure wasn’t at 1 bar? We have previously mentioned that pressure is the restriction to flow. Given we had a lot of bubbles in the return line and the air is much more compressible, we now have less fuel flow and space taken up by air that can be squeezed easier. All of this results in less pressure being created.
Knowing what we now know, we can review the original captures to see if there is anything more that could be revealed that could have pointed towards cylinder 4. Looking at the first capture, where we used an accelerometer, we have added a view to show the data in spectrum mode.
What is interesting to see, is that if you separate out each injector, it becomes clear which is the one of interest.
This is where being able to review the waveforms with PicoScope 7 Automotive for analysis, can help you to reveal more about the waveform you originally thought held little or confusing data. Additionally, it means that there is now an improved way to test for this fault, which is relatively non-intrusive, as we are just capturing signals from an accelerometer attached to an injector.
I hope this helps and thanks to Lee Sharp at L&D Commercials Ltd for inviting me along for this adventure.
