A BMW that cranks and will not start, with the tester reporting no communication at the engine address, sends most people straight to the unit. On the N20 and the N55 that unit is the MEVD17, and it stays asleep until terminal 15 reaches it with a clean supply.
The tester lists the engine address and then fails to read it, while the gearbox and the body units still answer on the same line. That difference matters, because a dead bus takes every address with it and a single missing address points at one unit.
The cluster shows a message about the engine and the car turns over without catching, because the unit is not there to fire the coils. Do not read that message as a diagnosis. The unit is absent from the network, and the reason is electrical until you measure the pin that feeds it.
Start with the battery, measured at the posts with the ignition on and the headlamps loaded. A unit that sees a supply below its threshold keeps the processor down and answers nothing, which looks exactly like a failed board. Support the battery and read the addresses again.
Check the diagnostic socket as well, because one pushed pin gives an intermittent read on every address and makes the whole car look faulty. Measure the supply and the ground at the socket with the tester plugged in. When both sit where they belong, the fault is in the car.
KL15 is the ignition supply that wakes the engine unit, and here it arrives through the front power distribution rather than straight from the switch. A missing feed on that terminal leaves the unit asleep, so the tester sees a silent address and you see a crank with no start.
Measure KL15 at the unit connector with the key in the run position and the plug fitted, using fine probes through the seal. You are looking for battery voltage, and a reading several volts below the post voltage is a fault in itself. Note both numbers.
A meter shows an average and hides a supply that sags when a load comes on, so put the scope on the same pin and crank the engine. A feed that collapses as the starter turns is being loaded by another circuit, and the unit never gets a clean wake up.
Look at the shape of the trace at key on too, because a slow rise says a capacitor in the supply path is tired. A clean step from zero to battery voltage is what the unit expects. Anything rounded or stepped deserves chasing while the wiring is in front of you.
MEVD17 is the unit on the N20 and on the N55, and its frequent complaint is a silent address or one that drops out after a jump start. The injection driver stages sit on the same substrate as the processor, so a voltage spike reaches both. Ask how the car was started.
On the bench the unit runs from a stabilised supply at thirteen and a half volts, and current is read at switch on with nothing else connected. A board that pulls more than an amp at rest has a short, and one that pulls almost nothing has an open supply path.
The BSD line is a single wire that carries data between the engine unit and several sensors around the engine, and a short on it can drag the unit into a state where it stops answering the tester. Test the line before any unit is ordered.
Put the scope on the BSD wire with the ignition on and compare the trace with a car that runs. A line that sits flat, or one pulled low and held there, names a shorted sensor rather than a dead unit. Unplug the sensors one at a time.
Once the supply and the line are cleared, the unit goes to the bench on a stabilised supply with the pin assignment it sees in the car. Confirm communication there before any repair, because a unit that answers was never the fault. Save the file first.
Inspect the pin row under magnification and look at the solder where the connector meets the board, because heat and movement open those joints early. Flex the board gently while the unit is communicating. If the address drops as you flex, the repair stays local to that joint.
Write down the KL15 value you measured and the addresses that answered before you started, then put the list from after the repair on the same sheet. Keep the code list as the tester printed it, with a note on which entries came back.
Tell the owner what was found even when the answer is a wire and not a unit, because the same symptom returns when a ground is left loose. Pack the unit in antistatic foam if it travels, and send the connector with it.
Confirm the engine control unit is awake before you condemn it, since a unit with no supply behaves exactly like a dead unit on the scan tool. Measure the permanent feed and the ignition feed at the connector with the key on, then check the ground pins under load. If both are present and the unit still does not answer on the bus, the diagnosis is closed by measurement and the board work is justified by evidence.
| Symptom | Likely cause | What is measured |
|---|---|---|
| No engine address, other units answer | KL15 missing at the unit | Voltage on the KL15 pin with the plug fitted |
| Address drops during cranking | Supply loaded on the same circuit | KL15 trace on the scope across the crank |
| Engine address returns when a sensor is unplugged | Short on the BSD line | BSD waveform with the ignition on |
| Silent in the car, answers on the bench | Connector joint or car wiring | Current at switch on and the pin row under magnification |
Because the diagnostic line is shared and only one address is missing, so the line still works. That points at the feed of that unit or at the unit, and the KL15 pin is measured before the housing is opened.
Yes, a supply below the threshold keeps the processor down and the address silent, and the reading returns once the battery is supported. That is why the battery is measured under load first.
The line goes on the scope with the ignition on and the trace is compared with a running car, then the sensors on it are unplugged one at a time. When the trace returns as one comes off, the short sits in that sensor.
No, a unit that answers on a stable supply was not the fault, and the car still has a supply or wiring problem to find. The measured values travel with the case so the next step starts from the same numbers.