Engine coolant sensor trouble

Maybe someone can clear this up for me being new to arduino and electronics in general. I’ve created a circuit so my arduino-uno can read an automobile engine coolant sensor. The problem is the sensor itself. I’ve bench tested two Glowshift sensors and now a Dakota digital sensor, and they all display the same behavior. They are significantly inconsistent. While bench testing the Dakota digital sensor, when the temperature is on the rise I get 222 ohms at 176F, but when the coolant temp is on the fall and hit’s 176F again it reads 134 ohms. The temp on the rise hits 200F I get 145 ohms and on the fall at 200F I get 89 ohms. This results in significantly different values from the Analog to digital converter on my Uno. Am I missing something, or are these automobile coolant thermistors in these things that inaccurate to try to read temps with a microcontroller with any accuracy? Thanks you for the help.

Post your circuit and code.

Have you tried using a fixed resistors of various values to see if your measurement technique is plausible, before worrying with the sensor?


Can you post the datasheets for the sensors, I'm not wasting time hunting them down.

Thanks for the reply. As i stated, I bench tested the ohm values at different liquid temperatures with my fluke multimeter and got those inconsistent results. No need to involve the measuring circuit when the sensor thermistor ohms values are inconsistent. That will assuredly result in inconsistent values on the measuring circuit.

missdrew, I will when i get a chance here. Thanks.

Sry, missed the reference to the Fluke meter.

Does time (waiting) let the sensor reach the same value for a given temperature?


I was emailed these sensor values by the manufacture, but not the actual data sheet. I’ve requested the actual data sheet now as well. As you can see, not only do my measurements differ significantly from the manufactures stated values, but the values of my own measurements differ significantly depending on if the temperature is rising or falling. This has been the case with 3 different sensors I’ve tested now!

Temp / Ohms per manufacture’s values / My Ohms with temps on the rise / My Ohms with tempts on the fall

*248F / *36.5 / 66 / 48
*212F / *61.9 / 118 / 74
200F / N/A / 145 / 89
180F / N/A / 207 / 125
*176F / *112.1 / 222 / 134
*140F / *221.2 / 415 /

  • Manufactured supplied temp and ohm values

I can only suggest your method of measuring is flawed!

I'd ASSUME it's just thermal mass and thermal conductivity between then housing and thermistor, slowing-down the resistance the resistance change.

Are the readings consistent & repeatable if the temperature is stable for a few minutes?

It is difficult to maintain a specific temp for a few minutes. I don't have that kind of control im afraid. I know what your getting at. Perhaps the values they arrive at are at temps that have been stable for at least a few minutes. Thing is, that seems not ideal as the coolant warms up, you want to catch it as soon as it hits a temp you are concerned with and display or even alert you. Which is what i want it to do.

The coolant temperature measurement has nothing to do with the actual water cooling system. It is controlled by the thermostat valve that controls the water circulation. The temperature gauge is just something to keep the operator happy. My VW gets along just fine with no temperature gauge.
For many years, there was only a warning light for over temp.
If you want the REAL temperature, then you will have to get away from the automotive type devices.

Thanks for the reply Paul. Yeah I do know that the temperature sensor measurement is just alerting the user what temp coolant is at in theory. My aim is, if i can pull it off, is having a reasonably accurate temp measurement so i can display it and have an audible alert when approaching an unsafe threshold. Sort of an idiot display/light with sound alert. Lol. I am trying to set up one of those couplers inline with the radiator hose that a sensor can screw into and monitor the coolant temp. You are saying the regular engine coolant sensors wont cut it then. Bummer. Judging from my tests so far it seems to confirm your point. Any suggestions for a source that has a more accurate sensor for this kind of application?

Somewhere in the past I have removed thermistors from equipment that were enclosed with a glass envelope, kind of like the old NE2 neon light bulbs. Look at scientific equipment/components distributors.

Maybe try one of those "waterproof" DS18B20 sensors. Put it in a well attached to the radiator intake hose perhaps.

After doing some more testing I think I am getting a little better grasp on how these things want to behave. When the temperature is on the rise, especially from a cold start, the ohm values are wildly inaccurate, but stabilize once a temperature has been maintained long enough. So the thermistor lags behind real bad in rapid temperature swings. I was able to test the ohm values at desired temps and they stayed pretty consistent when I was able to a maintain that temperature for long enough time. I could work with that maybe, but what concerns me is the stabilization time. How rapid can an engines coolant temp rise and reach a critical threshold temp before the sensor stabilizes to the proper target ohm reading so the micro controller circuit can trigger it’s alerts? If it does not stabilize fast enough, I experience a critical overheating event before I’m alerted by my circuit. Seems to be a lot of variables to try to account for in this. The reaction time of the thermistor in these sensors appears to be the deal-breaker.

The "reaction time" of a typical thermistor in good thermal contact with the object to be measured is a few seconds at most.

You have not clearly described your experimental setup, nor is it very clear how you are making your measurements, but the thermal conductivity of whatever is between the thermistor and the object of interest will be by far the most important issue.

Thanks for the reply jremington. The test setup of the sensor is this; I have the tip of the sensor and a digital thermometer submerged in a pot of vegetable oil. I heat the oil up and note the temp to ohm relationships with my Fluke multimeter connected to the two leads if the sensor. For better or worse that's it.

How do you measure the oil temperature?

Is the oil constantly being stirred? Heat flow through still oil is very slow. That is true for water, too.

I have a digital thermometer poked into the oil right where the sensor is to read the temp of the oil. No im not stirring it.

Stir constantly and well, and you should see a dramatic difference in the outcome.