Following replacement of the temperature sender unit I decided I wanted to check the calibration of the two items, the old Veglia gauge and a new aftermarket sender unit. The Veglia water temperature gauge once you look at it closely is a very clever simple design. Basically you have three connections (apart from the bulb) numbered 12V, GN, BN. Now I thought 12V would be the 12v input, it isn’t, it’s the feed to the sender unit, (still cannot figure that one out) BN is the supply (12 volts) and GN is ground. Across each connection are two coils of different resistances. One is 344 ohms (GN to BN) the other is 122 ohms (GN to 12V) These two coils are positioned 90 degrees to each other and the current passing thru the coils creates a magnetic field The resistance of the temperature sender unit decreases on temperature rise so this resistance decrease causes more current to flow through the first coil and less current to flow to the second coil. This change in coil current causes a magnetic field across the two coils to move the needle. The new sender unit was placed on a test rig at 90C (195F) (optimum engine temp) the gauge read about 30% low which surprised me, I was expecting it to read higher, so then connected a small potentiometer across 12V(sender wire) to Ground. This effectively worked in conjunction with the sender unit which allowed the gauge to be adjusted to match the actual temp. Incidentally we also varied the voltage input from 8 volts to 14.2 volts under the test. Any variance in voltage is almost all cancelled out due to the 2 coil design. At 8v there was possibly a 10% error. To make a correction to the gauge the potentiometer measured 550 ohms and resistor was placed on the gauge. The test re-run and at 90 C (195F) the needle was in the exact position on the gauge. The temp was raised and lowered as required and it worked perfectly although it was difficult to assess as the gauge scale is rather vague at best. So the aftermarket sender units are close to being accurate but do need adjustment to give a true reading. Now this is the bit I don’t get, putting it all back in the car, running the engine to 90C (thermocouple inserted down the main flow pipe from the cylinder heads), the gauge now read higher. Perhaps there are some local interferences once refitted back into the instrument binnacle, to move the gauge back down to match what was going on with the actual water temp, a resistor was placed across GN to ground instead of 12V to grd. This corrected the gauge once again and am now confident that at when the engine as a 90C my gauge is telling me exactly that. The resistance values are not linear so a 90C it is 100% accurate, above & below that I would say a small error but insignificant. So long as my gauge is accurate at operating temperature a slight error either side of that is immaterial.