What was wrong with the DAF tractor?
Ben Martins, automotive application specialist at Pico, follows the clues that led to a successful diagnosis and repair of issues afflicting a DAF tractor.
It’s been a while since I last worked on a tractor unit and I’d forgotten how impressive and how important these machines are. Keeping them on the road and working is imperative so when an issue arises getting direction quickly, as well as the fix, is vital.
The vehicle was presented with a load of warnings and alarms on the dashboard and so reaching for the scan tool was important to get an insight as to what is triggering the warnings. Whilst the scan tool was busy – scanning – I took the opportunity to chat to the driver. With no apparent drivability issues before the warning lights lit up, the only thing the driver mentioned was the lack of DEF (AdBlue), which was confirmed by the fault codes. Generally, a vehicle like this would get through a lot of DEF as it’s constantly on the road.
A large proportion of the fault codes were pointing towards the DEF system, though there were also some CAN related issues. But the fault code that was labelled as high priority was flagged as a DEF pump failure – Code:P3912. Despite all the DEF trouble codes being inactive, anything related to the DEF system can have a major impact on the engine operation and that would explain the engine protection and engine torque reduction fault codes. Taken together with the customer’s observation, this is a clear starting point for investigation.
As with everything, making sure you know how it should work is vitally important. Fortunately, despite the fact this is a Cummins after-treatment system, it shares a lot of similarities to the Bosch Denoxtronic 2.2 after-treatment system. Operationally, it is pretty much the same. There is a pressure building phase, injection events and then purging once the system is switched off. There are however some slight hardware differences in these areas:
- DEF reservoir
- Filter
- 4/2 flow control valve
- Single direction, fixed displacement pump
- Pressure relief valve
- Throttle/orifice to generate pressure by restricting flow
- Pressure sensor
- Injector
The biggest difference between the two systems is that the pressure sensor and the orifice are part of the injector module rather than within the pump. This makes the injector module a lot bigger than the Bosch unit but this means the pump assembly is smaller.
One of the most important checks we can make is simply to look over the area in question – but this is often neglected. You can often tell a lot just by using your eyes but don’t stop there: use all the senses you’ve been given. From the pictures above you will see that we have a slight DEF leak at a couple of the joints between the pump and the injector module. As we have a number of fault codes relating to the pressure being too high, this would make sense if the pressure stepped outside of the rating of the seals in the pipework.
You should also note the fittings on the injector module and the shape of the pipework, especially the bottom hose as it appears to be stretched. DEF is highly corrosive to some metals, so plastic push fit connectors are typical. Could the original fittings have been replaced or modified at some point? It’s good practice to make notes of everything you see, hear, smell and feel.
Almost certain we need to look at the after-treatment system, in particular the DEF pressure. Looking at the wiring diagram we could get these signals at the injector module. The plug is an eight-pin connector but only six-pins are used and, with the positioning of the connector, it was easier to use the breakout leads for ease. This isn’t always practical or typically advisable because disturbing the wiring could actually ‘fix’ the fault being reported, cancelling the fault code and leaving the original issue untouched. But sometimes this is necessary and, in this case, it was the best option for getting the signals we required.
First up then is to take a look at the pressure sensor. We can see if the injector is trying to operate by using a current clamp. By referencing the ground supplied to the injection module we only needed to graph the supply and the output. Should there be an issue with the ground supply we wouldn’t have a stable 5V supply. Unfortunately, just by having the vehicle idling with all the fault codes still present, meant the DEF system would not automatically prime itself. In order to test the unit we had to clear the codes and then manually activate the pump through the actuator tests through the scan tool. This allowed us to perform a priming test as shown.
From this capture we can see that the pressure sensor is responding to the changing pressure where it starts just above 0.5V and then sharply rises during the purging phase and gradually continues to rise before the test is stopped and then the drop in pressure. The pressure sensor is exactly 5V which, as I’m referencing to the sensor ground, gave me confidence that the ground is good.
What is interesting is that the scan tool was reporting a pressure between 10 and 12 bar. The 10 bar reading is slightly above the traditional 9 bar we would expect to be seeing for a DEF system suggesting that the fault codes are correct in reporting over pressurisation.
Using the scan tool again to perform an injection test gave further interesting results. This time though we connected to the pump control to see how the ecu reacts to the over-pressure.
During this test we saw a pattern in which the system never reaches the injection stage. Instead, the after-treatment system primes the DEF, the pressure again rises to around 3V then the system switches the 2/2 control valve to drop the pressure by purging the system and then, once the pressure has dropped, the whole cycle is repeated. We observed this cycle several times and it only stopped when the test ended. The scan tool provided the explanation: the test wouldn’t proceed to the injection stage if the pressure is too low. Interesting that we have a high pressure issue and so would assume that the test wouldn’t continue for both too high a pressure as well as low.
With having seen both the voltage live on Pico and the pressure reading on the scan tool we can say that the sensor is outputting a value that reflects the fault. Considering there are no burst pipes, just some small weeping around the seals, we have to believe there is a blockage. Looking back at the schematic, one place had our attention.
Based on its position, the filter inside the injector module would be first on the action list. It sits after the pressure sensor and, if we remember the golden rule that pressure is the resistance to flow, then a blocked filter here would result in the pressure sensor reading a high pressure. So, we removed the DEF lines from the injector module – easier said than done – so we could see the filter.
Not only was it so blocked that it had burst but it also appeared to be contaminated with a dark brown/reddish substance like rust. But how can there be high pressure when the filter is burst? Given that pressure is just a restriction to flow, we have in effect created an opening so some fluid can pass through the split in the filter but it is still restricting flow. With the DEF pipes now disconnected and any spillage being caught in a container, we sampled the fluid from the tank to see if there was an issue there.
Using a refractometer designed for DEF we can see we have a clear fluid which is sitting around 33%. Ever so slightly above the target of 32.5% but, given we don’t have a quality issue, it’s not something to be concerned about.
With the driver still with us at this point, the job was more about continuing with the diagnosis and seeing if this was the problem. Fitting the injector module together without the filter is risky and a thorough cleaning of ports was needed. We were glad that we did clean the module as we did find some finer particles which may have also been adding to the restriction but could also cause a blockage at the injector. When we were happy that we’d cleaned the module thoroughly and that the fluid coming from the tank was clear then we ran through the scan tool test, hoping that the DEF pressure had been reduced.
Here we can see that the initial pressure gets to 2.7V and then drops away to around 2.3V and then we see we have activity from the injector. I’m not 100% sure when the after-treatment ECU determines a high pressure or whether it waits to see if the pressure drops once the pump duty has been dropped to 10%. If we compare the before and after we can see some clear differences.
Using reference waveforms we can bring in the pressure trace captured when the filter was still fitted. Here it’s apparent that without the filter there is a clear drop in pressure yet before the pressure continues to rise. With activation now taking place we carried out a quantity test to make sure there were no further blockages.
The extra detail you can see with Pico compared to the reports given by serial tools allow you to see much more and give you an overview of the health of the after-treatment system. By zooming in to the waveform we can see the sawtooth effect that is taking place. This is caused by the pressure drop after an injection event has taken place. This can be useful in determining whether an injector is partially blocked without removing the unit from the exhaust.
By zooming in further again, you can pick out the ripple generated by the pump. Here, we would be looking for something that looked different or a repeating pattern.
Having checked that the rest of the system worked correctly and that the blocked filter was the sole cause of the fault codes and warnings all we had to was order a new filter plus the necessary gaskets and bolts. Once all was refitted and the codes cleared, the vehicle was left to run and observed with Pico to ensure that DEF was injected – which it did. Thankfully for this vehicle the filters are available separately through the VM.
What I find extraordinary is that the same system is used by other manufacturers yet you cannot buy these filters on their own – you have to replace the whole module, which seems extremely wasteful considering you can prove the rest of the system is working correctly. By replacing just the filter we saved the customer just under £1500.
The question still remains though: what blocked the filter in the first place? Given how clean the fluid was from the tank to the injector module was this something breaking down in the module?
One theory brings us back to our observations at the beginning of the diagnosis. The fittings used to join the pipework from the tank to the injector module appear not to be original. We do know that DEF has corrosive properties that may have caused a reaction to these components which has meant the filter has become blocked by doing its job and filtering this contamination.
We didn’t have anything to repair or replace these fittings and with a customer that was waiting the best thing we could do was advise that this may happen again in the future and the remedy would be to replace the pipework from the tank to the injector module.
With all the DEF faults I’ve seen more often than not it’s caused by something else and for the most part the after-treatment systems are pretty reliable. Fortunately the systems have everything already built in which means using existing sensors, feedback and control circuits can all be easily seen with Pico.
However, ensuring you know how something works is also vital when it comes to diagnosis. Pico is there to support and capture the evidence required to back up any repair.
