# Do linear thermistors exist on the market?

I'm working on a project right now that requires me to use a thermistor to sense the tempertaure change with an op amp comparitor with hysteresis. Because of the non-linearity of a normal NTC or PTC thermistor, I wanted to know if there existed any thermistors whos resistance changes linearly to the amount of heat applied. A quick google search shows me that a Silistor is exactly that, but there isn't a product that I found that has that name. So do these linear thermistors exist on the market, and do they exist in a through-hole format? Side note, I'm not making this a product recommendation. I just want to know if they are sold and if they have a specific name.

• how linear do you need? Imho, resistors are relatively linear, so if PTC isn't linear enough for you, it's hard to tell what would be. Commented Mar 11, 2020 at 12:09
• Where it has to be placed, wire distances? Are you planning to use a MCU and ADC, or just analog circuits? Commented Mar 11, 2020 at 12:46
• What do you mean by linear? How linear? In physics almost anything can be linear (approximately) or nonlinear (if you look close enough).
– Curd
Commented Mar 11, 2020 at 13:29

## 7 Answers

All thermistors are non-linear, some more so then others.

RTD's (Resistance Temperature Device?) are pretty linear, but they are not anywhere as sensitive a thermistors.

There's a whole range of temperature sensing options available if you want to consider an active device.

No, thermistors are always nonlinear. But RTD do exist and they are linear. They are widely used in industry as Pt100 100 ohm resistance at 0 degrees Celsius, since their low resistance are very immune to the environment noise, but difficult (expensive) to measure. On the other hand HVAC use Pt1000 i.e. 1000 ohm at 0 deg. They are easier to measure as Pt100.

The common "Pt100" sensor has a positive temperature coefficient and is linear within a certain temperature range, see this example datasheet.

Note that the Pt100 is a resistance temperature detector or RTD, which although it has a positive temperature coefficient, isn't a PTC.

• What do you want to show with that graph? On a log-log-scaled graph almost anything looks "linear".
– Curd
Commented Mar 11, 2020 at 13:25
• @Curd I disagree, can you prove me wrong by give an example of a (mathematical) function that looks linear on a lin-lin scale but does not on a log-log scale? Commented Mar 11, 2020 at 13:32
• The other case is relevant: a function that is not linear (and of course also doesn't look linear on a lin-lin-scale) but looks linear on a log-log-scale. There are plenty: e.g. $y=x^a$ where $a$ is any constant (!=1).
– Curd
Commented Mar 11, 2020 at 14:06
• $y=x^2$, $y=\sqrt{x}$, $y=\frac{1}{x}$ ... all of them look perfectly "linear" on a log-log-scale but are not.
– Curd
Commented Mar 11, 2020 at 20:01

It depends on what range of temperatures you want to measure. If it's restricted to about 10 °C, then form a potential divider with a resistor equal to the resistance of your NTC at the midpoint temperature in series with it, and measure the voltage at the junction.

If you are happy to switch resistors, or use several dividers and multiplex their outputs, then you can extend the linear range by stitching several small linear ranges together.

Adding to what the other members already wrote. If you are willing to go a bit crazy and need a higher accuracy, you could theoretically "increase" the linearity of a thermistor. This can be accomplished by using a bunch of them and averaging their values (Much like the jar of beans experiment). In practice, you could use a non-inverting amplifier to average the values read out by each thermistor.

Consider the following simulation: It consists of three resistors which are modeled as thermistor (In this case the time represents the temperature). Due to the exponential factor and the offset constant, they produce different non-linear voltages. At the right side, you have an average amplifier which also acts as a buffer.

As you can seen in the waveform, the output voltage is nothing but an average of all thermistors' voltages.

If you are patient enough you can find thermistors which would produce a "differential" behavior, therefore creating a linear temperature behavior if averaged out. Once again, I am not sure whether you really need such a linearity in your project.

I hope it helps.

You could search for them under "PTC temperature sensors" or "PTC thermistors".
Here is what Digi-Key has in that category:
https://www.digikey.com/products/en/sensors-transducers/temperature-sensors-ptc-thermistors/550?k=ptc+thermistor&k=&pkeyword=ptc+thermistor&sv=0&pv1989=0&sf=0&FV=-8%7C550&quantity=&ColumnSort=0&page=1&stock=1&pageSize=500

Pt 100, pt 500 and pt 1000 are pretty linear device, there are tables and algoritme for Conversion, but if you need high accurancy at Wide range there are a variety of thermo controlers and 4 to 20 ma transducer. Choice should be made by the purpose.