An A3 8P that shows the wrong fuel level and the wrong coolant temperature has two faults on the dash or one fault in the wiring that feeds both, and the cluster on this car also holds the gateway for the whole network. That last detail matters, because a cluster that is failing shows up as communication faults in half the modules on the car. The checks below start at the senders and work back to the cluster.
The fuel gauge sits at empty with a full tank, or it swings across the dial as the car goes round a bend. The temperature gauge climbs to the middle in two minutes from cold or it never moves at all. Some owners report both faults at once, and others say the fuel reading drifted over a month while the temperature stayed honest. Write down which gauge is wrong and when it started.
Ask whether the car has had a radio fitted or a tow bar wired recently, since both put work into the loom behind the dash. Ask whether the cluster has been out for any reason, because a plug that is not fully latched gives readings that come and go with the road. Note the mileage and the engine code from the label, since the resistance tables differ between builds.
Fill the tank to the brim and watch the gauge settle. A gauge that reads empty on a full tank has a sender that is open or a wire that is broken, and one that reads full on an empty tank has a sender shorted to earth. A gauge that moves about on its own as the car is driven points at a sender float that is sticking in the tank.
Watch the temperature gauge from a cold start with the cabin heater off. A gauge that reaches the middle in two minutes has a sensor reading high, and one that takes fifteen minutes on a warm day has a sensor reading low or a thermostat that is open. Compare the gauge with the heater output and with the temperature of the top hose by hand before you trust either.
Reach the sender plug under the rear seat and measure the resistance across its pins with the tank at the level shown on the gauge. Then move the float by hand through its travel and watch the reading change in a smooth sweep. A value that jumps or that drops out at one point on the travel is a worn track in the sender, and that is the fault the gauge has been reporting.
Compare the readings at empty and at full with the table in the repair data for this car. A sender that reads within the table at both ends but jumps in the middle is still faulty, so write all the figures down as you sweep. Check the earth at the sender body as well, since a poor earth adds resistance to every reading you take.
Measure the coolant sensor at its plug with the engine cold, then again at working temperature, then compare each reading with the table for that sensor type. The resistance falls as the temperature rises, and a sensor that reads open when warm gives a gauge that stays low. Take the engine temperature from the tester at the same moment, since the two readings should agree.
Check the sensor wiring from the plug back to the cluster with a meter and a load on the line, rather than with an ohmmeter alone. A wire that reads a few ohms on the meter can still drop volts under load, and the cluster will read that drop as a colder engine. Wiggle the loom at the bulkhead while the gauge is watched, since a gauge that moves with the loom has a wire fault.
Open the cluster address and read the fuel level and the coolant temperature in live data before you look at the needles. The cluster converts the sender resistance into a value, and that value is what the gauge displays. When the live value is correct and the needle is wrong, the gauge drive inside the cluster is at fault and the sender is innocent.
When the live value is wrong, compare it with the resistance you measured at the plug. A cluster that reports a level well away from the sender reading has a wiring fault between the two, or a cluster input that has drifted. Write both figures side by side on the sheet, since that comparison is the measurement that decides the repair.
Read the gateway address and the fault memories of the modules on the network. A cluster that is failing gives no communication codes for several modules at once, and those codes can look like separate faults across the car. Clear nothing until you have written the full list, because the pattern of modules that cannot be reached is the evidence.
Check the bus wires at the cluster plug for continuity and for a short to earth, then scope them while the ignition is on. A quiet pair with a cluster that has a good supply points at the cluster. A pair with activity and a cluster that does not answer puts the fault in the plug or in the cluster input stage, and the unit comes out for bench work. Read the gateway coding as well, since a cluster coded for equipment the car does not have reports modules that are not fitted and fills the list with faults that mean nothing.
Power the cluster on the bench and feed the fuel input with a resistance box through its range while you watch the gauge. A needle that follows the box in a smooth sweep has a healthy drive, and one that steps or sticks has a worn stepper motor. Repeat the test on the temperature input with a resistance box set to the cold and hot figures from the table.
Look at the board around the stepper motors for a dark patch or a dry joint, since these gauges fail at the motor winding more often than at the input circuit. Check the plug pins for corrosion, and check the board for liquid damage from a spilled drink through the dash vents. Photograph the board before any cleaning, because the marks are the evidence.
Refit the cluster with every plug latched, then switch the ignition on and watch both gauges sweep through their self test. Fill the tank and let the gauge settle for a minute before you read it, since the cluster damps the signal to stop the needle moving with the fuel. Run the engine to working temperature with the tester attached and compare the gauge with the live value.
Read every module on the network once more and confirm the communication codes have gone. Then drive the car over a rough surface with the gauges watched, because a plug that is not seated shows up on a bad road rather than on a smooth one. Write the sender readings and the cluster number on the sheet so the next visit starts with figures.
| Symptom | Likely cause | What is measured |
|---|---|---|
| Fuel gauge at empty with a full tank | Open sender or broken wire | Sender resistance at the plug, swept by hand |
| Gauge moves on its own on bends | Sticking float in the tank | Sender resistance through the full travel, earth at the body |
| Temperature gauge up in two minutes | Sensor reading high | Sensor resistance cold and hot against the table |
| Several modules cannot be reached | Cluster with the gateway inside | Bus wires at the plug, live values in address 17 |
It drives, but you lose the reserve warning and you are guessing at the level. Fill by the mileage until the sender is tested, and keep the tank above half.
The sender is reached under the rear seat and the cluster stays in the dash. The cluster only comes out when the live value is right and the needle is still wrong.
The gateway for the network sits inside the unit, so a cluster that is failing takes the path down with it. That is why the whole list of codes is read before anything is cleared.
A resistance that jumps or drops out as the float is swept by hand, seen with a good earth at the body and a clean plug at the tank.