Celsius vs. Kelvin: When it does and doesn't matter


While the Kelvin and Celsius (or centigrade, if one prefers that name) scales are indeed different, they differ only by an offset; the interval or change in a physical quantity represented by 1 degree in either scale is identical. If temperature scales were rulers, the distance between marks on a Kelvin ruler and a Celsius ruler would be identical.

Because the increment represented by one degree in either scale is exactly the same, in the context of temperature changes or differences it’s perfectly fine to use °K or °C interchangeably, because there is absolutely no difference. If the weather forecast says tomorrow’s high temperature will be 2°C or 2°K warmer than today, it means exactly the same thing because a temperature change is being referenced. On the other hand, an outdoor temperature of 30°C is pleasant or maybe a bit warm, while an outdoor temperature of 30°K would be uncomfortably cold (Oxygen freezes into a solid @ about 54°K). The distinction between scales matters in the latter case, because what’s being referenced is an absolute quantity; a measurement made relative to some physical reference point where “zero” actually means “zero” in a sense of nothingness, vacuity, or a complete and total absence.

When temperatures are mentioned in the electronics field, there are some cases when a temperature change or difference is being talked about, and some other cases where an absolute quantity is being referenced.

Resistor temperature coefficients are an example of the first case. When a resistor datasheet specifies a temperature coefficient of +/- 100ppm/°K for example, what that means is that for every 1°K change in device temperature, the measured resistance of a device may change by 100 parts per million in either direction from the initial measured value. Because the quantity in question is describing the effects of a change in temperature and a change of 1°C is exactly equivalent as a change of 1°K, the units of °C and °K can be used interchangeably with no difference in meaning in this case, and use of either one is simply a matter of preference.

Different suppliers have different preferences; some resistor manufacturers specify temperature coefficients in terms of °C, others in °K. One of our goals at Digi-Key is to present the products we sell in a way that helps our customers make apples-to-apples comparisons between similar products, and find similar products to compare with a minimum amount of mouse-clicking. Part of that process is eliminating distinctions without a difference, and because there’s no practical difference between a resistor temperature coefficient of (x) per °C and (x) per °K, we’ve chosen to eliminate that distinction.

An example of where the use of °C or °K often does matter is in the context of temperature sensors. Semiconductor temperature sensors can be calibrated to deliver outputs that are proportional to either °C or °K (or °F, or °R, for that matter) so if such a device has an output equation of 10mV/°(x), the difference between (x) being °C or °K would mean a difference in sensor output of 2.73V at a given temperature. Other devices such as thermistors have non-linear temperature characteristics, making it numerically important to use the right temperature scale in one’s calculations.