I’ve lost count of how many ECUs have landed on my bench with a note taped to the lid saying “module dead, needs replacing.”

Maybe a third of them were actually dead.

The rest? Water in a connector. A corroded pin. A dry solder joint that only misbehaved once the engine bay got warm. One memorable one had been cooked by a bloke who’d welded a tow bar onto the chassis without disconnecting the battery. That ECU didn’t stand a chance.

So before you order a replacement engine control unit, clone it, code it, and cross your fingers — read this. Because if you don’t find why the first one died, the new one usually joins it.

Here are the five causes of ECU failure I see most often, in roughly the order they show up.

1. Water and Corrosion — The Number One Killer

Water is the single most common reason an ECU ends up on a repair bench, and it’s the one people misdiagnose most often.

The frustrating part is that water damage rarely announces itself. You don’t get a nice clean “module not responding” fault and nothing else. You get weirdness:

• No communication with the scan tool — the ECU simply isn’t there on the bus
• Intermittent operation — runs fine for a week, then stalls on a wet morning
• CAN faults across multiple modules, because a corroded ECU can drag the whole network down
• Complete module failure once corrosion finally bridges two tracks it shouldn’t

Where does the water come from? Blocked scuttle drains. A perished bulkhead grommet. A leaking windscreen seal. Pressure-washing the engine bay a bit too enthusiastically. On some models the ECU lives in the plenum chamber under the wipers, which is a genuinely daft place to put sensitive electronics — and those ones drown regularly.

Do this first: unplug the connectors and look inside them. Torch, magnifier, take your time. You’re hunting for green or white powder on the pins, water droplets, a tide mark on the plastic, or that dull grey oxidised look on what should be shiny gold plating. Check the harness where it enters the connector too — water wicks along wire strands under the insulation and travels surprising distances.

Inspect connectors and wiring before you open the ECU. If the corrosion is external, you’ve found your fault, and opening the case just adds contamination risk.

If water has got inside the housing, act quickly. Corrosion doesn’t stop when the puddle dries — chloride residue from road salt keeps eating copper for months. An ECU that got wet in January will often fail properly in April.

The fix: proper ultrasonic cleaning of the board, neutralising the residue, replacing corroded components, repairing lifted tracks, then conformal coating the board again. And — critically — fixing the leak. Otherwise you’re doing this again next winter.

2. Electrical Overload — Fast, Brutal, Completely Avoidable

If water is the slow killer, voltage spikes are the assassin. One bad moment and the ECU is gone.

Here’s what causes it:

Incorrect jump-starting. Leads on the wrong terminals, or a big truck’s 24V system hooked to a 12V car. The spike goes straight down the power feed into the ECU’s regulator and output drivers. This is easily the most common overload cause I see.

Reverse battery polarity. Fitting a battery backwards, even for a couple of seconds. Some ECUs have reverse-polarity protection; plenty don’t, and the ones that do often sacrifice a protection diode saving themselves. Either way, that’s a bench repair.

Welding without proper electrical protection. Welding current looks for the easiest route back to the earth clamp. If that route runs through a sensor wire and out via the ECU, the module is toast. Disconnect both battery terminals, and clamp the earth as close to the weld as you physically can.

Failed alternator. A regulator that’s lost the plot can push 18V+ into the system, or throw savage AC ripple down the line. Ripple is nasty — it doesn’t kill instantly, it slowly degrades capacitors and regulators until the ECU starts doing odd things months later. Always scope your alternator output when investigating a “random” ECU failure. If you fit a repaired ECU to a car with a duff alternator, you’ll be seeing that customer again very soon.

Shorted solenoid. Purge valves, VVT solenoids, boost control solenoids, transmission solenoids. When the winding shorts, it demands current the ECU’s driver was never designed to supply. The driver dies.

Shorted injector. Same story, higher stakes. A shorted injector coil takes out its driver stage — and on ECUs with paired or shared drivers, it can take out neighbouring channels too. Always measure injector resistance before fitting a repaired ECU.

Damaged wiring. A chafed loom rubbing on a bracket, a rodent-chewed harness, a botched aftermarket alarm or stereo install. A signal wire touching 12V or chassis earth will kill the input or output stage it’s connected to.

The rule: never fit an ECU to a car until you’ve found the short. Measure resistance to earth and to 12V on every driver circuit that failed. It takes twenty minutes and saves an expensive repeat.

3. Heat and Vibration — The Slow Grind

ECUs live a hard life. An under-bonnet module can swing from -15°C on a winter morning to over 100°C after a long motorway run, then cool down again. Every single day.

Metals expand and contract at different rates. Solder, copper, fibreglass PCB, component leads — all moving against each other, thousands of cycles a year. That’s thermal cycling, and it’s relentless.

What it produces:

• Cracked solder joints, especially on heavy parts: relays, large capacitors, connector pins, power transistors. Connector pin joints are the classic — they carry the mechanical strain of the loom and the thermal stress.
• Dry joints that behave perfectly cold and open up when hot. The dreaded “only faults when warm” complaint.
• Lifted or cracked PCB traces, particularly around heat-generating power sections.
• Degraded component connections where a lead has fatigued at the body.

Now add vibration. Engine-bay-mounted ECUs get constant buzz plus the occasional pothole slam. Combine tired solder with vibration and you get a joint that makes and breaks contact under load — the source of countless intermittent, unrepeatable, “no fault found” complaints that get cars sent from garage to garage.

How to catch it: freeze spray and a hot air gun. Get the fault to appear, then localise it with controlled heating and cooling on the board. Inspect suspect joints at 20x magnification — a cracked joint has a fine dark ring around the lead. Reflow properly with fresh solder; don’t just melt the old joint and hope.

4. Failed Internal Components — What Actually Dies Inside

When the outside checks out clean, it’s time to look at the silicon. These are the parts that fail most:

Voltage regulators. The most common single point of failure inside an ECU. They step the car’s messy 12–14V down to clean 5V and 3.3V for the processor and sensors. When one fails you get no start, dead-scan-tool, or 5V sensor references collapsing. They run hot by design, so they age faster than almost anything else on the board.

CAN transceivers. The chip that connects the ECU to the vehicle network. Kill this and the ECU may be running perfectly internally while being completely invisible to every other module and your diagnostic tool. Bus short-circuits and jump-start spikes are the usual culprits. Good news: it’s often a cheap, straightforward chip-level repair.

MOSFETs. The workhorse switching transistors driving solenoids, fans, relays, pumps and heaters. They fail short (output permanently on) or open (output dead). Almost always caused by an overloaded or shorted external component — which is why you must find the root cause.

Injector drivers. High-current, high-heat, high-stress. Symptoms are a dead cylinder, a misfire code that follows the ECU rather than the injector, or repeated failure of the same channel. Peak-and-hold driver stages on diesel and GDI applications are especially vulnerable.

Ignition drivers. Same story on the spark side. A failed coil primary or a shorted coil pack takes out the driver. Result: cylinder-specific misfire that stays put when you swap coils and plugs around.

Capacitors. Electrolytics dry out with heat and age. As ESR rises, the power supply gets noisy and the ECU starts doing strange things — random resets, comms dropouts, phantom fault codes. Look for bulged tops, brown residue around the base, or feet lifting off the board.

EEPROM and flash memory. These hold immobiliser data, adaptations, VIN, mileage and calibration. Corruption here gives you no-start with a valid key, lost adaptations, or checksum errors. Failure often follows a low-voltage event during cranking or a botched flash. Always read and back up EEPROM before any repair work — that data is often irreplaceable.

Microcontrollers. The brain. Least common failure but the most serious, usually from a big overvoltage event or ESD. Symptoms are total unresponsiveness with good power and ground present. Replacement means transferring the software and security data across — specialist work, and not always economically sensible.

5. Connector and Harness Faults — The One That Fools Everyone

I’m putting this last so it sticks: a damaged ECU connector looks exactly like a dead ECU.

Same symptoms. No comms, intermittent faults, multiple unrelated codes, sensor signals dropping out. And a huge number of ECUs sent away for repair come back marked “no fault found” because the real problem was six inches away in the plug.

What to hunt for:

Loose or spread pins. Female terminals lose their spring tension after repeated unplugging, or a pin gets pushed back in the housing during a rushed reassembly. Result: a connection that’s fine at rest and drops out under vibration. Gently test the retention with a matching male pin — you should feel real resistance.

Corrosion on the terminals. Even a light film of oxide adds resistance. On a 5V sensor circuit, a few hundred milliohms shifts the reading enough to trigger a plausibility fault.

Water contamination in the connector body. Seals perish, latches don’t fully click home, and the connector fills up. Between neighbouring pins, moisture creates leakage paths that generate the most baffling fault combinations you’ll ever see.

Wiring strain. Booted looms pulled tight after engine work, chafing on a bracket, or a bodged repair with twisted wire and insulating tape. Broken strands inside intact-looking insulation are the sneakiest fault of all — the wire tests fine on a meter and fails under load.

Do this: wiggle-test the connector and loom with live data on screen. Voltage-drop test the ECU power feeds and grounds under load, not just with a static meter reading. And check the earth points — a corroded chassis earth strap causes more “ECU faults” than most people would believe.

Quick Diagnostic Order That Saves Real Money

Work through it in this order and you’ll avoid nearly every unnecessary ECU purchase:

1. Battery and charging system. Load-test the battery, scope the alternator for ripple and overvoltage.
2. Power and grounds at the ECU. Voltage-drop test under load.
3. Connectors. Unplug, inspect, look for water and corrosion.
4. Harness. Wiggle test, check for chafe points and previous repairs.
5. Outputs. Resistance-check injectors, coils and solenoids for shorts.
6. CAN bus. ~60Ω across CAN High and Low with the ignition off, and scope the waveform.
7. Only then consider the ECU itself.

Repair or Replace?

Repair usually wins when the fault is a known component failure — regulator, CAN transceiver, driver stage, capacitors, cracked joints. You keep the original immobiliser data and calibration, so there’s no coding headache, and it’s typically far cheaper than a new module.

Replacement makes more sense when the board is heavily corroded, the microcontroller is gone, or the ECU has already been repaired several times.

Whichever route you take, fix the root cause first. A repaired ECU fitted to a car with a shorted injector or a knackered alternator will fail again — and next time, it’ll be blamed on the repair.