A couple of months ago an acquaintance of my bought a second-hand Audi A8. These are nice cars with lot of luxury and suppleness. The car is 2,5 year old and with today’s technique lot’s of kilometres can yet be driven with it. The car is equipped with an V8 4,2 litre engine with automatic transmission and has driven 65000 kilometres. After owning the car for one day, the man noticed that there was something wrong with starting the engine. There were no problems starting a cold engine, but if the engine was started after it had done some work, it misfired. With this story the acquaintance called the dealer and the car was examined at the garage. After a day he got his car back, but it looked that the problem wasn't solved. My acquaintance had lost his trust in the dealer, because the problem probably already existed when the car was sold to him and because they hadn't fixed the problem (yet).

This acquaintance happens to be the manufacturer of our ATS labscopes, so making an appointment for examining this problem was done quickly. After doing some tests the problem got clear. The starting problems were in the cooling down period at 10-30 minutes after the engine had been at operating temperature. A relative simple assignment, because the measuring can be done with a good functioning engine (starting and immediately shutting down the engine) and when the problem occurs (after 10 minutes standing still). First we have done some measurements to figure out if a vital control was missing, such as controlling of the ignition coil and/or injectors at the breaking down period. The used ignition system has on every spark plug a separate ignition coil which has an integrated end module. Measurement of high voltages isn't that simple. So first we have measured the control pulse going to the end module of a ignition coil so we get a indication on what is going on and because these are the easiest to get to. In the starting period the pulse signals existed, see picture 1. Sometimes these pulses can be very small and the pulses of this car are also not higher than 4 Volt. The width of the pulses are about 5 milliseconds, what is also the contactangle which is calculated by the control device. Both findings were not abnormal so the cause of the problem had to be found somewhere else.

Picture 1: Ignition

The next step was the measurement of the fuel injectors. Because we have 2 measurement channels available we could measure the signal of the fuel injector and the crankshaft sensor at the same time. At the first time measuring the signals we already know that the signal of the crankshaft sensor had to be fine and also the signal of the camshaft sensor. We know that the engine control unit (ECU) must have both signals to determine the position of the piston. If one of this signals is missing the ECU then does not know the current position of the specific piston and will stop signalling the ignition coil to prevent damage to the engine. The measurement done with the injector and the cranksshaftsensor revealed immediately the cause of the problem. The injector was controlled open much too long (found after measuring: 300 milliseconds) which causes the engine to “drown”. Normally a injector, at a normal stationary engine speed, is controlled open for a period between 2-5 milliseconds and when starting the engine, that is already warm, the open period is between 5-10 times the time when the engine is cold. It should be noticed that when a engine doesn't start immediately the time that the injector is opened gets shorter to prevent the engine from “drowning”. Picture 2 shows that the time that the injector is opened is too long. This picture also shows that the signal of the cranksshaftsensor has a offset voltage. This is seen more often with modern cars. The reason for this is for getting an even better self-diagnosis of this sensor, in this case short-circuiting. The value of the “offset” is around the 2,5 volt and this is the imaginary zero-line of the alternate current signal.

Picture 2: After 10 minutes

We had to wait for half an hour before the engine started again. At that moment the opening time of the injectors had shortened so much, that after several times starting the engine, it got “clean” and the right amount of fuel was reached. Picture 3 shows this phase where in the beginning the engine still got too much fuel (approximately 50 milliseconds) and did not start. This scope-image shows a virtually black signal beam (bottom signal) that does not clearly reflects the alternate current. although, in the beginning there are 2 small openings visible which are the missing spikes from the flywheel (position recognition). The scope of the scopeimage can be easily made bigger so that a clear crankshaft signal is visible. The start of the engine is in the signal of the cranksshaft sensor visible at the moment that the signal is rising. Also a increased engine speed can be observed whereby the signal voltage is relative high for several milliseconds and then stabilize in the following time. An inductive cranksshaft sensor is just like a bicycle dynamo. It produces a higher voltage at higher engine speed. The voltages produced by these sensors can be, at a high engine speed, as high as 60 volts.

Picture 3: After 20 minutes

Figure 4 shows the injector signal together with the signal of the cranksshaftsensor with a smoothly starting engine. It is clearly visible that the time, that the injector is opened, not longer is then 15 milliseconds. This image does not show a previous injector pulse (due the fact that the scope image is enlarged), on which the engine already started. This can be seen at the rising cranksshaftsensor signal. So far we only observed the facts, but still had not found the cause. although, we had an idea where the cause of the malfunction could be found. A coolant temperaturesensor can, when an engine is very cold, cause a very long injectiontime. Because the car was still in its warranty period the dealer was informed about our findings. Because the system probably had not given any faultcodes was it much harder for the dealer to find the problem. We advise the garage to take a look at the data, that can be retrieved with there serial communication tester, and especially at the values of the coolant temperature sensor. Indeed this showed that the ECU saw an engine temperature of -50 degrees Celsius. The sensor showed a very cold engine, but was not fully interrupted, because if this was the fact then a fault code was retrieved from the system. The sensor is replaced and the problem was solved. A malfunction in this sensor circuit would also be found with the ATS Labscope, because the voltage of the coolant temperature sensor with an warm engine is mostly under the 1 Volt. In the situation of malfunctioning the voltage would be probably between the 4, 5 en 4, 8 volt and is therefore rejected. If these high voltages are also measured while the engine is warm it is recommended to also measure the ground of the sensor with respect to the ground of the battery.

Picture 4: Normal firing

R.H.M. Metzelaar
www.gmto.com