Since I have not fitted the speedo sensor to mine car yet I got it on the bench and did some careful investigation with a power supply and a scope. With the Blue/red lead grounded and the green lead taken to a 5v current limited supply via a 1k resistor I found I got 8 pulses for every rev of the sensor (the pulses were about 3v high). I guess from this that the sensor is a hall probe and some magnets. I think the method could be quite susceptible to electrical noise. The wires are not screened and run in the loom along with lots of other wiring so its quite possible that the odometer is adding up any spikes from the ECU, the indicators, your mobile phone etc. I can't see anyone going to the trouble to code in a rounding function to the unit just so it can cause this problem!
