Diagnosing Intermittent Electrical Faults: A Systematic UK Workshop Method
Every workshop has one car that keeps coming back. The customer is convinced the warning light came on, you cannot get it to come on, the scan tool shows nothing live, and the previous garage already changed a sensor that did not need changing. Intermittent faults pay for a lot of unnecessary parts because the trade habit is to guess, fit, and hope. There is a better way. It is not glamorous, it takes longer the first time, and it almost always ends at a connector.
Start with the customer, not the car
The single most underused diagnostic tool in any workshop is a notepad and ten minutes with the customer. By the time the car is on the ramp the fault is gone, but the customer was there when it happened. They know things the scan tool cannot tell you. Most of us are too embarrassed to ask properly because it feels unprofessional, but the questions you do not ask are the questions you will pay for in chasing your tail later.
The script I use looks like this, and I write the answers down rather than trying to remember them. When does it happen: cold, hot, dry, wet, after motorway, after stop-start traffic, only on the M1, only on the school run? What does it feel like: warning light only, drivability change, audible noise, smell? How often: every drive, once a week, never twice the same day? What were you doing in the thirty seconds before: indicating, braking, going over a speed bump, turning the heater on? Has anything been done to the car recently: service, tyres, MOT, jump-start, accident repair, aftermarket fit?
The "anything done recently" question alone solves about one in five intermittent faults outright. A new battery fitted last month with a slightly slack negative terminal will produce a glorious selection of random codes once a week. A bumper repair after a kerb knock will have disturbed parking sensor wiring. A new audio install will have tapped a permanent live with a Scotchlok. None of this shows up on a scan tool. All of it shows up in conversation.
Read freeze-frame properly
Freeze-frame data is the snapshot the ECU took at the moment the fault was confirmed. People glance at it and move on. They should be reading it like a witness statement. Engine speed, road speed, coolant temp, intake temp, throttle position, fuel trims, load and time since start at the moment of the fault. That tells you whether the fault happened cold or hot, idling or under load, before the engine had warmed up or after.
If freeze-frame shows ECT at 22°C and road speed at 0 km/h, the fault happened just after start on a cold morning. That points at warm-up enrichment circuits, cold solenoids, things that draw current heavily for a few seconds. If it shows ECT at 90°C and a brisk load value at 80 km/h, the fault happened on a hot, working engine. Different connector, different harness section, different fault tree.
On vehicles with multi-frame logging (a lot of newer Fords, the VAG group post-2015) you can get a small handful of frames either side of the event. Read them all. The trend across frames tells you whether a parameter was drifting or whether it dropped instantly. Instant drops are open-circuit faults. Drifts are usually thermal.
Voltage drop, not continuity
This is the bit most jobs get wrong. A continuity test with the circuit dead tells you the wire is intact. It does not tell you the wire can carry current without losing voltage. A connector terminal that has shrunk back to three strands of copper will buzz on a continuity test and drop two volts under load.
Voltage drop testing is simple and almost no-one does it. The circuit must be live and under its normal load. Put one probe at one end of the section you want to test, the other probe at the other end, on the same wire. Whatever the meter reads is the voltage lost across that section. For a power feed under load you want less than 0.3V; for a signal wire under its working current you want less than 0.1V; for a ground return under load you want less than 0.1V. Anything above those numbers and the joint is degrading.
The trick is to break the circuit down into segments. Battery post to fusebox stud, fusebox stud to relay input, relay output to component, component to chassis ground point, chassis ground point to battery negative. Each segment gets its own test. The bad one stands out instantly because it carries most of the lost voltage. People who skip this step end up replacing parts because the part "looked worst", which is a coin flip dressed up as diagnosis.
The wiggle test, done properly
Hold the harness still and probe it with the meter. Then wiggle it. Then run a finger along it pressing each connector in turn. Then flex the wire entry into each connector. Watch the meter the whole time. A connector that has lost retention will show its hand within five seconds.
Do the wiggle with the circuit live and under load, not dead. A loose terminal that passes 100 mA of static current may fall over completely under 5 A of working current. If the system is a CAN bus, scope the differential signal while you wiggle. If it is a sensor, watch live data on the scan tool. If it is a power feed, monitor voltage at the load.
For really uncooperative faults a cheap accelerometer or a phone strapped to the dashboard logging vibration will sometimes correlate with a glitch. I have caught a broken solder joint on an instrument cluster PCB by logging cluster blackouts against speed bumps. The customer thought it was random. It was not random. It was every speed bump.
Heat and cold, on demand
If the customer says the fault appears after twenty minutes on the motorway, you do not have twenty minutes on the motorway in the workshop. You have a heat gun, and that is enough for most thermal faults. Aim it at the suspect connector from about 15 cm away on low setting and watch for the fault to appear. Do not melt the loom. The point is to mimic underbonnet soak, not to weld things.
The opposite trick works for cold-only faults. Freeze spray (the proper electronics kind, not aerosol degreaser) on a single connector will drop its temperature thirty or forty degrees in a second. If the fault clears as the connector chills and returns as it warms back up, you have isolated the failing joint to a single connector body. Re-warm with the heat gun to confirm. Two minutes of work for a fault the previous garage chased for three weeks.
Heat-soak faults on hot starts are particularly amenable to this. CKP and CMP connectors, fuel injector connectors, coil-on-plug connectors, anything mounted close to a hot exhaust or directly on the engine. The plastic relaxes, the terminal lifts, and a perfectly good sensor reads open circuit. Cool the connector, signal returns. We covered the worst offenders in our DTC codes that lie article.
Scope traces tell stories that meters cannot
A multimeter averages. A scope shows you what is actually happening. For intermittent signal faults a scope is the only sensible tool, and you do not need to spend four figures on one. A Pico kit or one of the cheaper Foxwell or Topdon scopes will do for most diagnostic work, and even an old Snap-On scope picked up second-hand is a step up from guessing.
The two patterns every workshop should know cold are the CAN bus eye pattern and the ABS sensor square wave. CAN High and CAN Low sit at 2.5V at rest, swing apart to 3.5V/1.5V dominant, and should look like clean mirror images of each other. Any asymmetry, ringing, or slow rise time on one line points at a wiring fault between the scope point and the next termination resistor.
ABS sensor square waves should be clean, equal mark-to-space, and consistent amplitude wheel to wheel. A ragged top edge or amplitude that drops with speed says the connector or the toothed ring is suspect. Compare side to side on the same axle. A scope makes a five-minute job of what a meter cannot do at all.
Staging the fault: min/max recording
If you cannot make the fault happen on the ramp, set the car up to catch itself. Most decent scan tools have min/max recording on live data. Pick the parameter most likely to glitch (sensor voltage, fuel trim, supply voltage to a specific module) and let the tool run while the customer takes the car for a normal day's driving.
When they bring it back, the min and max values either side of the operating range will tell you whether the parameter swung wildly during the day. A 5V reference that dropped to 3.1V at some point is a wiring fault. A coolant temperature that briefly read -40°C is an open circuit somewhere. A supply voltage that touched 9V on a working engine is a high-resistance joint in the power feed.
Some scan tools will trigger and store a snapshot on a parameter threshold. Set the threshold tight, let the customer drive, and the tool will capture the moment the fault occurred. This is how you turn a customer-reported gremlin into a hard piece of evidence in your hands.
When the loom has had enough
Sometimes the right answer is to stop fixing and replace the whole section. A harness that has been chafed through against a chassis edge, soaked in coolant from a heater core leak, cooked by an exhaust touch, or pulled by an accident impact is not going to be repaired pin by pin in any economic way. After the third or fourth connector on the same harness branch, you are losing money chasing the next one.
The honest conversation with the customer is that beyond a certain point of loom damage the labour for piecemeal repair exceeds the labour for a section replacement. This is particularly true on Land Rover Discovery 3/4 A-pillar looms, BMW E60 boot lid harnesses, Vauxhall Astra H column looms, and any vehicle where a previous repairer has been at it with Scotchloks and Sellotape. Strip back to the nearest clean connector either side and put a known-good section in.
OEM repair sections are available for many high-failure looms, but they are not cheap. Quality aftermarket repair pigtails with OE-spec connectors are usually the practical answer, and that is the bulk of what comes through our doors at Auto-Connectors. The principle is the same: matching pin count, matching seal type, matching terminal gauge.
What this means in practice
Systematic diagnosis of intermittent faults is not a single trick. It is a sequence: customer interview, freeze-frame read, voltage drop test, wiggle test, thermal provocation, scope confirmation, min/max staging if needed, harness section replacement if all else fails. Each step is cheap. Each step rules out a category of fault. By the time you have done all of them on a stubborn case, you know exactly where the problem is, and the part you fit is the part that needed fitting.
The pattern we see, talking to UK workshops every day, is that the modern car gets blamed for being "too complicated" when in fact the diagnosis discipline has not kept up with the diagnosis tools. The car is fine. The fault is the same kind of fault it has been for fifty years: a poor joint, an open terminal, a chafed wire. The connector is almost always the suspect. The method above gets you to it without guessing.
