Is it possible to calculate an incandescent bulb's temperature from V?

[Eric Bretschneider]

Of course rather than arguing who has the best theoretical model/knowledge, you could take advantage of the fact that Illuminating Engineering as a field is over 100 years old and take advantage of the information base of that field.

The following is from "Lighting Handbook: Reference and Application" (8th ed.) from the IES (Illuminating Engineering Society) p.186

Life1/Life0 = (Volt0/Volt1)^13
Lumen1/Lumen0 = (Volt0/Volt1)^1.9
Efficacy0/Efficacy1 = (Volt0/Volt1)^3.4
Watt1/Watt0 = (Volt1/Volt0)^1.6
CCT1/CCT0 = (Volt1/Volt0)^0.42

Subscript 0 is rated and subscript 1 = adjusted value

There is of course some variation between manufacturers. The exponents are not intended to be applicable to huge differences in input voltage. In the absence of experimental data this should suffice.

 

[Eric Bretschneider]

We are on topic, but the question has now become “how can you best estimate colour temp based on filament voltage.

I contend that you need to measure both filament current and voltage. From that you can compute resistance to maybe three decimal places. Then using the polynomial relationship between temperature and the resistivity of tungsten, you can accurately calculate the filament temperature.

An old method used to measure the temperature of a furnace was to view the furnace through a filament globe. That showed the outline of the filament against the hot background. The filament voltage was then adjusted until the filament became invisible against the background, neither darker nor lighter. At that point the filament resistance was computed from V and I, so the temperature could be known. That method cancels most of the spectral band variation in the temperature comparison.

But the method does not actually measure the colour temperature of a black body, which is a broad weighted spectrum, much of which is unseen by the eye.
We need to define why we need to measure or know, and what we mean by colour temperature.
Maybe with narrow band LEDs and three colour cameras and displays colour temperature has become an irrelevant concept.

Color Temperature is only valid for light sources with chromaticity coordinates that lie on the Planckian Locus (the locus of points for all emitters whose spectra follows Planck's Law).

For light sources with chromaticity coordinates that are near the Planckian locus, a different metric: Correlated Color Temperature (CCT) is used. Correlated Color Temperature is defined as the minimum distance between the chromaticity coordinates of a given light source and the Planckian locus as measured using CIE 1960 uv chromaticity coordinates.

CCT is still used and is not an irrelevant concept. However many people incorrectly use color temperature when they actually mean CCT. (a nit picking detail)

Color Temperature and CCT are relevant because they are used to define "shades of white". White light is most often characterized by appearance. Different spectral power distributions can have the same appearance due to the way the LMS cones our visual system process light. CCT is important for displays as it relates to the white point of the display.

While outside of the scope of this thread, color rendering (how objects appear under a light) is a critical issue in the lighting industry.

 

[hutchphd]

CCT1/CCT0 = (Volt1/Volt0)^0.42

Again (now for the third time) this exponent follows because the filament is a very good black body emitter...please see #9 and #25 above.
I have designed (and redesigned) several commercial colorimeters, the first of which used a tungsten light bulb as a source (and yes I am that old!)

 

[Eric Bretschneider]

Again (now for the third time) this exponent follows because the filament is a very good black body emitter...please see #9 and #25 above.
I have designed (and redesigned) several commercial colorimeters, the first of which used a tungsten light bulb as a source (and yes I am that old!)

Your model is a good approximation. However my point was to utilize something based on decades of DATA.

Theoretical derivations are nice, but the OP didn't really require that.

Although for extra credit you should factor in the change in spectral power distribution to shift your exponent from 0.40 to 0.42.

 

[hutchphd]

I seem to be having some trouble making my primary point.
My "model"(your word) can be written down in three lines if you know the physics. It requires no spreadsheet or even knowing the numerical constants.
So with one pencil and a small scrap of paper the original question

Hi!

I have a question:
Is it possible to calculate incandescent light color temperature based on it''s input voltage?

can be answered simply with an emphatic "yes, if you know a little physics"....

 

[Eric Bretschneider]

I seem to be having some trouble making my primary point.
My "model"(your word) can be written down in three lines if you know the physics. It requires no spreadsheet or even knowing the numerical constants.
So with one pencil and a small scrap of paper the original questioncan be answered simply with an emphatic "yes, if you know a little physics"....

The answer is also an emphatic "yes, if you do a little research"

You can always try first principles calculations, but Mother Nature makes no assumptions.

And the tristimulus response curves can't be derived from first principles . . .

 

[hutchphd]

The tristimulus response curves have nothing to do with the Planck temperature.
They do tell us what the surface temperature of the sun is (and has likely been for some time!)
But the sun would shine the same whether we can name its colors or not (falling tree in the forest notwithstanding!)

 

[Eric Bretschneider]

The tristimulus response curves have nothing to do with the Planck temperature.
They do tell us what the surface temperature of the sun is (and has likely been for some time!)
But the sun would shine the same whether we can name its colors or not (falling tree in the forest notwithstanding!)

"Planck temperature" is a very different concept - it is the highest possible temperature (about 1.4e+32 K). I think you were referring to the temperature of a Planckian source which is different.

Tristimulus response curves are used to calculate chromaticity. Without those you can't calculate color temperature. The concept of color temperature includes the human visual system which in turn brings in the tristimulus response curves.

They have nothing to do with the surface temperature of the sun. They allow us to calculate the chromaticity coordinates of sunlight (which does not lie on the Planckian locus).

 

[hutchphd]

Yes the term "Planck Temperature" was absolutely incorrect. I was unaware of that definition and was using it as a shorthand for the temperature of the blackbody radiator. My bad.

Tristimulus response curves are used to calculate chromaticity. Without those you can't calculate color temperature. The concept of color temperature includes the human visual system which in turn brings in the tristimulus response curves.

They have nothing to do with the surface temperature of the sun. They allow us to calculate the chromaticity coordinates of sunlight (which does not lie on the Planckian locus).

The entire response of the human eye has evolved to maximize the use of solar radiation. Compare the response curve of the eye to a 6000K blackbody. Serendipity? Do you really think that if we lived near a red dwarf we would have the same color response?

No Mas...

 

[weirdoguy]

"Planck temperature" is a very different concept - it is the highest possible temperature (about 1.4e+32 K).

No it is not. The same goes with Plancks lenght, time and other units. They don't have any physical meaning, except that they can define some sort of usable scale. But not always, e.g. Plancks resistivity is about 30 ohms and it's not the highest nor the lowest resistivity possible.

 

[pinball1970]

I did some similar work on this with LEDs What changed increasing wattage on measurement was the intensity not the SPD or CCT. More photons but at the same wavelengths.

 

[Baluncore]

I did some similar work on this with LEDs
What changed increasing wattage on measurement was the intensity not the SPD or CCT.
More photons but at the same wavelengths.

Unlike a filament lamp, LEDs are fixed voltage devices, with light output dependent on forward current, and forward voltage dependent on material = fixed colour.
The blank lines in your post make it look like you used an MS windows based editor, then pasted your text ?

 

[pinball1970]

Unlike a filament lamp, LEDs are fixed voltage devices, with light output dependent on forward current, and forward voltage dependent on material = fixed colour.
The blank lines in your post make it look like you used an MS windows based editor, then pasted your text ?

Yes I am lazy, I am always missing words out or something then noticing after,so I type in a blank email then paste. I am usually on pf at lunch at work while I am answering mails.
You can get dimmable LEDs, does the SPD change on those?

 

[Baluncore]

You can get dimmable LEDs, does the SPD change on those?

We see Spectral Power Distribution as colour, which is set by quantum chemistry of LED materials.
A dimmable LED will probably be a violet LED, with three or more fluorescent chemicals to change the colour balance. It should not change colour with power.

 

[martin314159]

Regarding the 220 volt incandescent 2800 cct running at 110 volts. Omcheeto had It correct. The cct goes to the 0.38 power of voltage. The new cct is 2125 K. The power drops from 100 watts to 33.3 ; the lumen output from 1550 to 95; the operating resistance from 482 ohms to 363.
best regards,
Martin314159

 

[martin314159]

I would also like to say Eric Bretschneider reply below in his August 19, 2019 #31 is correct too. The power of the voltage can vary from .38 to .42 depending on the temperature of the filament whether and whether the filament is in a vacuum or inert gas.

The following is from "Lighting Handbook: Reference and Application" (8th ed.) from the IES (Illuminating Engineering Society) p.186

Life1/Life0 = (Volt0/Volt1)^13
Lumen1/Lumen0 = (Volt0/Volt1)^1.9
Efficacy0/Efficacy1 = (Volt0/Volt1)^3.4
Watt1/Watt0 = (Volt1/Volt0)^1.6
CCT1/CCT0 = (Volt1/Volt0)^0.42