Correcting automobile temperature gauge reading; V to I converter

I have a temperature gauge that reads a little low and would like to make it more accurate. I believe a voltage to current converter circuit would work but I need one that increases the current as the voltage decreases. The sender is a negative temperature coefficient thermistor that decreases in resistance as temperature increases, allowing more current to flow into the gauge. I propose to connect the sender to a resistor matching the gauge resistance, about 79 ohms, which would be connected to 12 Volts; then the voltage across the sender would be the input to the converter, which would have an adjustable gain to compensate for the inaccuracy of the sender. The sender resistance is a nominal 240 to 30 ohms. the midscale of 150 degrees on the gauge is about 173 ohms. The current though the gauge is 21 ma at 100 degrees, 60ma at 150 deg, and 110 ma at 240 deg. The gauge currents were measured with a variable pot on the bench. Voltages at the sender range from approx 9.5 volts at 100 deg, 7 volts at 150, and 3.5 volts at 240 deg. All the converter circuits I have been able to find increase current with increases in voltage and so I am asking if the op amp circuits typically used can be re configured to increase current out with decreasing voltage in. Or would I need to place an inverting voltage amp before the converter? The attachment is the general circuit I have in mind but needs to be modified as above. Your advice is appreciated.

get a more accurate sensor...? Far simpler

get a more accurate sensor...?

Click to expand...

Where's the fun in that? I have two senders already, both track pretty much the same; and new senders have no specs listed. I've read that they are as much as 20 per cent off. The meter itself may not be that accurate either.

If you just want more current through the gauge you could simply add a fixed resistor in parallel with the NTC one. That way you could make the gauge accurate at one particular temperature of interest. (Of course, it would do nothing to overcome the inherent non-linearity of the NTC resistor).

180° is the temperature of interest and a fixed resistor could correct that and wouldn't affect the 240° full scale that much , but it distorts the lower readings too much for me. Still looking for the possibillity of an inverting V to I.

Hi Bill

Not sure exactly what you have but I am curious. Attached is an image, the upper shows two configurations for a thermistor like used in automotive applications. In that upper image the thermistor is configured on the top and bottom of the voltage divider it forms. The bottom would be normal and as T increases the Eout decreases. With the thermistor on top the Eout would increase as temperature increases.

The lower drawing shows what should be a typical automotive application. My read for these things is they apply 5 volts and generally use a 350 ohm resistor to form the divider. The junction of the Thermistor Sensor and the internal series resistor provides the ECU with the temperature signal. My guess is the ECU then provides an analog signal to any on board vehicle temperature gauges. This is where your voltages tend to confuse me? I am guessing a voltage at the sensor between 0 and 5 volts. I also wonder if placing anything in parallel for example with the sensor will screw with what the ECU is reading?

Ron

Well here is what I have; an old Stewart Warner temp gauge and sender. The sender screws into the engine and the single terminal connects to the "S" terminal on the gauge. So the sender is always grounded. As the temperature increases the sender resistance decreases and the voltage at the "S" terminal decreases. There's no ECU involved here, just basic Ohms law. I measured the voltages and resistances needed to correct the readings so I would like to amplify the voltage across the sender about 1.3 X and convert it to a current to suit the gauge. The gauge has about 79 ohms from the "S" terminal to the Ign terminal; thus the voltages are the result of this divider made from the sender and the gauge resistances.

What you have is a crossfield type instrument. To recalibrate you need to dismantle the gauge & udjust the value of a resistor. Heres some pictures of a typical crossfield Temp gauge.

From this Stewart Warner document https://www.stewartwarnercanada.com/media/pdf/install_instructions/Service%20instructions%20for%20electrical%20gauges.PDF
this paragraph describes the type of gauge I have. Your "crossfield " gauge is something different, probably newer.

Have a good look at the crossfield gauge, its 2 windings at 90deg, same as a bi torque gauge. Just two different terms for the same type of gauge.

Well its all academic anyway, to disassemble mine would probably ruin it as the bezel/trim ring is crimped to the case and it takes a special tool to crimp it back on; so its not coming apart.

Just for interest sake here is the circuit of one of them & the NTC resistor is 800ohm @ 25degC. I have pulled one of those types of gauge apart to repair & re crimped the ring back on. And yes there will be some damage.