# Why is a temperature rise within an electrical component or assembly measured in kelvin?

I came across an IEC standard which uses Celcius as a temperature unit for maximum temperature limit E.g. IEC 60085 states class Y insulator maximum temperature limit to be 90°C, average ambient temperature to be 35°C.

For measuring the temperature rise, it is always expressed in kelvin. The maximum temperature rise limits are always expressed in kelvin and not in Celcius or Fahrenheit (likewise). As far as I know, Kelvin is a scientific scale which has its lowest limit at which molecular activities cease. Why is a difference or a rise (ΔT) expressed in kelvin? Is there any relation between the molecular activities and a rise?

• Do you know what the relationship between Kelvin and Celsius is? Commented Apr 15, 2023 at 15:54
• @DKNguyen This is what I learnt throughout my academics - byjus.com/physics/relation-between-celsius-and-kelvin Commented Apr 16, 2023 at 4:07
• Okay, so you're asking more about the convention than anything else. Commented Apr 16, 2023 at 6:49
• But the temperature rise is not always measured/specified in kelvin. It's usually deg C, at least in many industries. Commented Apr 16, 2023 at 11:54
• @SteveSh it seems common for the thermal resistance of a package, heatsink, etc. to be expressed in °C/W, while the thermal conductivity of the material it's made from is W/m·K - the former is more from engineering and the latter from physics, though I don;t know whether that's the reason Commented Apr 17, 2023 at 9:51

A change in Kelvin is exactly the same as a change in degrees Celsius, so there is no difference in meaning.

Somebody was just trying to be a stickler for SI units. Measuring thermodynamic ("absolute") temperatures in degrees Celsius is there not to confuse people and prevent arithmetic errors with 273.15, but the difference with relative temperatures is simply labeling.

The Celsius/centigrade unit is a major flaw in the traditional Metric system. One problem is that it was invented before the understanding of thermodynamics and absolute zero, so it has an offset of 273.15 from where it "should be". (This is called a difference-scale unit.) This makes multiplying and dividing degrees Celsius meaningless. In fact, there was issues of whether SI prefixes could be used with Celsius.

The second problem is its development before understanding of nuclear physics, a major complication in measuring properties of water is the properties are affected by isotopic composition, which varies by source.

• The issues of absolute zero and "difference" unit can be applied to Fahrenheit. And also to the Reaumur temperature scale surely. Commented Apr 15, 2023 at 8:02
• Yes. But OP was specifically asking about degrees kelvin vs Celsius/centigrade. Commented Apr 15, 2023 at 14:01
• If you are going to change the metric system, please also make gram into the base unit of mass.
– pipe
Commented Apr 15, 2023 at 15:42
• The thing that makes °C so hard to get rid of is that the melting and boiling point of water are much more relevant to everyday life than absolute zero. But you don't even need nuclear physics, you just have to move a little bit higher for the boiling point of water to not be 100 °C anymore - for example in Quito, Ecuador, it's just 90 °C. Commented Apr 16, 2023 at 19:49
• @pipe: The CGS version of the metric system (centimetre / gram / second) is apparently still used in some fields of science or engineering. The density of water is a convenient 1 in that system. (And CGS has different ways of defining some electrical units). But I suppose you mean keeping the metre and second to go with the gram, so none of the base units have a prefix. Yeah, always seemed a bit odd to me to call kg a base unit. Commented Apr 17, 2023 at 10:00

For many applications kelvin and degrees Celsius are interchangeable.

Where use of the Kelvin1 scale is handy is in measurement or calculations of thermal energy. For example, 20°C isn't twice as hot as 10°C because the zero point is arbitrarily chosen as the freezing point of water instead of absolute zero. However, 20 K is twice as hot as 10 K and a temperature rise from 300 K (27°C) to 400 K (127°C) in a block of metal, for example, raises the energy by $$\ \frac {400 - 300}{300} = 0.33 \$$.

1 The SI unit 'kelvin' is lowercase when spelt out and its symbol is capitalised, 'K'. In "Kelvin scale" the name is a proper noun so it is capitalised.

• That "raises the energy" part seems a bit suspicious to me. For example, water at atmospheric pressure would undergo a phase transition in that interval and the resulting state would have more energy per mole than the initial state than if there was no phase transition. However, upvoted for the important point that "20°C isn't twice as hot as 10°C." Commented Apr 15, 2023 at 16:40
• @AndrewMorton, that thought had occurred to me so I didn't use water and ignored phase-change to keep the answer simple. Thanks for the edits. Commented Apr 15, 2023 at 16:50
• In EE field, there is one more pretty much useful property of kelvin - most pure-ish metals (including copper, tungsten or aluminum) have their resistance directly proportional to their temperature in kelvin. This gets somewhat inaccurate below e.g. 10K but 10K is rarely seen on Earth these days. Commented Apr 15, 2023 at 20:42

The SI unit of temperature is kelvin, so it becomes natural to talk about K/W for a heatsink, ppm/K for temperature dependence and ΔT for temperature rise.

The absolute zero temperature have no impact on the examples you give.

As engineers, and it started at university, we were told to always work in Kelvin for temperature.

With time and some understanding, we would often just take the difference when working in deg C as the difference is not affected. However in certain calculations Kelvin has to be used otherwise there is an error.

Many of the units, like the specific heat capacity of water (J/kg/K) are specified with Kelvin.

• That would be "kelvin". SI units named after a person are lowercase when spelled out. Commented Apr 15, 2023 at 9:25
• On second thoughts, it depends on whether you are referring to "the Kelvin scale" (proper noun) or the SI unit (degrees) "kelvin". Life get's complicated. Hmm! Commented Apr 15, 2023 at 10:42
• The lines between university and industry are blurred. But seems to me that the only classes that used Kelvin were physics. In my engineering classes and 53 years of work experience, we worked exclusively in centigrade. Room temperature, 23 C. Thermal uncertainty, 11 deg C. Max junction temperature, 125 deg C. One of the few times Kelvin came up was when using Arrhenius equations to predict failure rates of ICs. Commented Apr 15, 2023 at 16:35
• @SteveSh IEC 61439-1, Table 6, specifies the maximum temperature rise on an aluminium conductor (in absence of manufacturer declaration) to be 55K. And the document initially mentions that all the temperature are mentioned with daily average ambient of 35C and peak of 40C. Thus, IEC could have mentioned 55+35 = 90C as an absolute value. But it refrains and instead mentions 55K as max. temperature rise. Commented Apr 16, 2023 at 4:02
• I guess I don't see the significance of that. I can find many specifications that only reference deg C. Kelvin is nowhere to be found. Commented Apr 16, 2023 at 16:52

Kelvin is the SI Unit. Ideally you’d have everything in kelvin.

Unfortunately for absolute temperature it never really caught on. I don’t think there is any data sheet which expresses a maximum temperature as e.g. 370k.

Kelvin and celsius use the same step size, so for temperature differences it doesn’t make any difference. By convention we usually use Kelvin for them.

Measuring in Kelvins also helps avoid the mistake I sometimes see of converting to Fahrenheit by multiplying by 9/5 and adding 32, when (for temperature differences) you should just multiply by 9/5 = 1.8.

If you use an online converter to convert, say, 20K to degrees F you’ll still get the wrong answer (-423.67 degrees F) but it is more clearly recognized as nonsense than 68 degrees F.

• I agree. The underlying phenomenon is that temperatures and temperature-differences should be considered as inhabiting different mathematical spaces: for most practical purposes, temperature is an affine space which doesn't have unique zero point (or, well, the zero point is irrelevant), thus Celsius and Fahrenheit choosing different arbitrary zero points. Whereas temperature differences are clearly in a vector space where zero has an obvious and important meaning. Commented Apr 17, 2023 at 8:31