Reading fault codes: what an OBD code tells you and what it does not

Cars from recent decades have a standardised diagnostic socket, usually in the driver's footwell or behind a nearby cover. It was made mandatory so that emissions-related faults could be read with a uniform device. Workshop equipment now talks to almost every control unit in the car through the same connector.

Only part of it is standardised, though. Engine and emissions faults can be retrieved with any simple reader; airbag, ABS, gearbox, comfort electronics and assistance systems often speak a manufacturer-specific language and need a device that understands it. A cheap adapter reporting “no faults” may therefore simply not have looked everywhere.

A standardised fault code consists of a letter and four characters. The letter names the area: P for powertrain, C for chassis, B for body and interior, U for communication between control units. The first digit after it separates generic codes, which mean the same for every manufacturer, from manufacturer-specific ones whose meaning only the manufacturer defines.

The plain-text description a reader shows is a translation of that number. With manufacturer-specific codes a generic device sometimes guesses — and gets it wrong. Anyone looking up a code should therefore search for it together with the make and engine.

A control unit reports that a reading is outside the expected range or that a signal is missing. It does not know why. A code describing the mixture as too lean can come from a leaking intake pipe, weak fuel pressure, a dirty air-mass meter or a faulty sensor. The sensor reports — it is to blame in only one of those cases.

Replacing parts according to the code is therefore guessing with the owner's money. The actual diagnosis starts after the code has been read: looking at live values, checking wiring and connectors, testing specifically. That work is exactly what a workshop charges as diagnosis.

For many faults the control unit stores a snapshot of the conditions: engine temperature, revs, speed and load at the moment of the fault. That data is often more valuable than the code itself, because it shows whether the fault occurred cold or warm, at idle or under load.

Then there is an entry's status: currently present, intermittent or merely stored, and on some devices a counter of how often it occurred. A one-off entry after a jump-start is something different from one that returns on every drive.

Besides stored faults, most readers show live data: coolant temperature, revs, intake air temperature, sensor voltages and, on petrol engines, the mixture corrections. That often lets you confirm or rule out a symptom — an engine that never gets properly warm shows it in the temperature curve before any fault code appears.

Interpreting them, however, needs target values, and those are in the manufacturer's documentation, not in the app. A value that looks odd can be normal for that engine. Live data is therefore good material for a conversation with the workshop, but rarely a diagnosis in itself.

Every reader can clear the fault memory, and the light then goes out. If the cause persists, it comes back, sometimes after minutes, sometimes after days. In the meantime the freeze frame and the counter are gone — exactly the information that would have got the workshop to the cause faster.

Clearing also leaves a trace. The emissions monitors' readiness tests then show “not complete” until the car has gone through a series of driving conditions. At a roadworthiness test and when buying a used car, that is a sign that somebody cleared the memory recently.

A simple adapter with an app is good for a first orientation: which area is reporting, is the fault current, what do a few live values look like? For viewing a used car it is a useful tool, if the seller agrees.

A workshop works with equipment that handles manufacturer-specific codes, guided test routines and adaptations, and with experience of which cause is common on which engine. Ask in advance how diagnosis is charged — by time or at a flat rate — and whether it is credited against a subsequent repair. The answer differs from business to business.

When buying a used car, reading the memory is one of the few checks that says something about the electronics within minutes. Three things are revealing: stored faults even when no light is on; readiness tests showing “not complete”; and warning lights that do not come on at all when the ignition is switched on.

A freshly cleared memory is no proof of a concealed fault — after work on the battery it is normal. But it deserves a question, and the answer should match what is in the service book and on the invoices.

If you have read the codes yourself, give the workshop the codes with the freeze frame — as a screenshot or list — and do not clear anything beforehand. Add the symptoms in your own words: when, how often, in what circumstances. Ask for the result of the diagnosis to be recorded in writing, with the readings the workshop is relying on.

Through Carvexia you can find businesses and their services in the garage directory. State in the booking request that it is about diagnosis, give the codes and the light, and ask how they intend to proceed. A workshop that explains how it plans to search before naming a first figure is, as a rule, also the one that finds the fault.