by Bryan Pollock, Willco Auto Care, Sanborn, NY, USA
This Dodge 2500 vehicle was originally towed in with a crank non-start and fault code P0087, Indicating rail pressure too low.
After evaluation, it was found that the CP3 fuel system had failed.
Contaminated fuel caused the CP3 pump to fail and metal debris spread through the entire fuel system. The fuel system was replaced, including the lift pump, high-pressure pump, injectors, lines, and the fuel tank was professionally cleaned.
After the parts were replaced and the fuel system was bled, the engine started and ran for 20 minutes. However, after 20 minutes, the vehicle hiccuped, misfired, knocked and then shut off.
The engine was then back to a crank non-start with low rail pressure, now a loud combustion knock with a regular cadence was heard as the engine was spun over during cranking. The P0087 fault code for fuel rail pressure too low had returned.
Scantool live data shows cranking rail pressure is approximately 75 bar, however, 275-300 bar is required for the engine to start.
This indicates there is a loss of pressure in the system. But the knocking while cranking indicates that fuel is getting into the cylinder. Since the labor involved in removing injectors to inspect the cylinders for fuel accumulation or capping off injectors one at a time is very intensive. I wanted a way to try and isolate which cylinder(s) may have fuel leaking into them as least intrusively as possible.
I decided to use a technique I have had success with in the past. First, I decided it would be useful to establish a few pieces of information about the Cummins 6.7L diesel engine.
During cranking, this engine uses near TDC injection strategy. Meaning that when the piston is just before TDC on the compression stroke, fuel is injected into the cylinder. This means we can use an injector trigger signal as a reference for the cylinder compression stroke.
The firing order for this engine is 1, 5, 3, 6, 2, 4.
Now on to the process to narrow down our affected cylinder(s). The theory in this case is that since I have a knock event while cranking, this means I have a partial combustion event. This would only be possible if fuel were getting into a cylinder.
I have an old GM single-wire knock sensor that I previously used to make a diagnostic tool. I want to use the knock sensor as a probe to pick up the partial combustion event while the engine is cranking.
The knock sensor is to be mounted to the cylinder head stud for the intake manifold near the intake heater grid. This will be connected to Channel A of the PicoScope.
Now, if there is a combustion event, I should be able to see it on Channel A with the knock sensor signal. Next, I need to be able to correlate that event to a potential cylinder.
Since the truck is already a crank non-start, I don’t need to disable the fuel system or jump the starter for the cranking event. I only need to disconnect the fuel rail pressure sensor so the engine control module will fire the injectors with the actual pressure far below the minimum specification. I can current ramp Injector 1 activation as a trigger. This will allow me to see the parade of cylinder compression events. I hooked my TA018 low amp clamp to Channel D.
If there are any knock/combustion events, the knock sensor will pick it up.
Since I have a cylinder trigger and know that injection takes place just before compression, I can accurately divide the waveform by the firing order based on the trigger. With the knock sensor in place and the current probe on cylinder 1 injector activation wire, we are ready to crank the engine and capture a waveform.
What is captured is pretty conclusive. But I need to edit the capture just a bit to make an accurate analysis. I used the phase rulers with a custom label of “COMP”, abbreviation for compression event, and set 6 partitions. I then set the partitions to start at one injector signal event and end at the next. I can then see events I need to focus on.
Next, I used the “Annotations” in PS7 to label each partition of the phase rulers with the cylinder number based on the firing order of 1, 5, 3, 6, 2, 4.
I can see below that it appears as if cylinder 4 is where my combustion event is coming from. The knock that is picked up happens almost immediately following the #4 injector activation event, based on the edge of the phase ruler and the firing order.
Now that we know what cylinder is causing our issue, we are able to block off the #4 injector and of course, the engine starts and runs.
We confidentially filed for warranty replacement of the remanufactured injector that had just been installed in the vehicle. A quick conversation with the remanufacturing company and sending over some screenshots helped expedite their decision to warranty the injector for us.
After replacing the #4 injector for the 2nd time, the engine now runs smoothly and the P0087 fault has not returned.
By taking the approach I did in this situation, I was able to avoid the time-consuming process of capping off one injector at a time and reduce the engine tear-down to just one time to replace the exact component required, instead of attempting to verify visually, which could prove labor-intensive and challenging.