CFL and lightbulb temperatures

[astropi]

A CFL can be classified as "5000 K", but (correct me if I'm wrong) the gases inside the tube do NOT get anywhere near that hot. Which brings up my first question:

How hot does a CFL get? Since the spectrum is not a blackbody, it's not easy to tell just by looking at the emission spectra. I was unable to find an answer online, so I'm curious if anyone knows?

Second question: I took the spectra of an incandescent bulb, and found the peak to be around 660 nm. Wien's law then gives us a temperature of over 4000 K. However, the melting point of tungsten is 3400 K. Therefore, how does one resolve this discrepancy?

Thanks!

 

[the_emi_guy]

First, CFLs do not generate light using thermal radiation. They use an entirely different mechanism (luminescence of phosphors).

Second, 660nm? This is in the visible spectrum, close to the spectral peak of the sun.
Incandescent bulbs have spectral peak in the infared, above 1000nm, precisely due to the fact that tungsten will melt at higher temperatures.

I have attached a good paper on the physics of the lightbulb for your reading pleasure.

 

[astropi]

First, CFLs do not generate light using thermal radiation. They use an entirely different mechanism (luminescence of phosphors).

Second, 660nm? This is in the visible spectrum, close to the spectral peak of the sun.
Incandescent bulbs have spectral peak in the infared, above 1000nm, precisely due to the fact that tungsten will melt at higher temperatures.

I have attached a good paper on the physics of the lightbulb for your reading pleasure.

Thanks for the reply, I do know that CFLs use UV radiation and phosphors to generate light, however that still does not answer my question, which is how hot do they get?
:)

As for the incandescent bulb, you're absolutely correct. I think the reason I got a peak around 660 nm is because the machine was not properly calibrated, and that certainly solves that problem! If anyone can tell me how hot the gases inside CFLs get, I would very much like to know (entirely out of curiosity by the way).

By the way, thanks emi for the paper. It is very well written. I'm still stumped about the "real" temperature of CFLs however...

 

[phyzguy]

The plasma in the fluorescent tube is actually quite hot - in the 10,000-20,000K range, if I believe "alexandria.tue.nl/extra2/200311683.pdf" . However, the density is low enough that the walls of the tube don't get very hot. The 5000K actually refers to the brightness temperature, which is basically how hot a blackbody would need to be to come close to reproducing the spectrum.

 

[AlephZero]

Thanks for the reply, I do know that CFLs use UV radiation and phosphors to generate light, however that still does not answer my question, which is how hot do they get?

They don't get hot. The 20W CFL bulb in my desk lamp (which has been on for a couple of hours) isn't even warm.

The "5000K" is a measure of the color spectrum of the light output, i.e. it the light iis roughly the same as a black body radiator at that temperature (compared with about 6000K for sunlight, and significantly lower temperatures for filament lamps).

 

[astropi]

They don't get hot. The 20W CFL bulb in my desk lamp (which has been on for a couple of hours) isn't even warm.

The "5000K" is a measure of the color spectrum of the light output, i.e. it the light iis roughly the same as a black body radiator at that temperature (compared with about 6000K for sunlight, and significantly lower temperatures for filament lamps).

Thanks, but I would like to get a quantitative answer. I have read (online) that some people claim CFLs do get "hot", but it depends on the wattage. That said, I would like to know what "hot" means. I realize this may vary between brands/wattage, but it would be nice to get some idea.

 

[phyzguy]

Thanks, but I would like to get a quantitative answer. I have read (online) that some people claim CFLs do get "hot", but it depends on the wattage. That said, I would like to know what "hot" means. I realize this may vary between brands/wattage, but it would be nice to get some idea.

What do you want to know? You asked earlier about the temperature of the gas inside the tube. Are you now asking about the temperature of the glass envelope?

 

[astropi]

What do you want to know? You asked earlier about the temperature of the gas inside the tube. Are you now asking about the temperature of the glass envelope?

The gas inside the tube. AlephZero said his 20-W bulb does not get warm, which implies that the gas inside the lamp is "cool", however I'm not sure that is the case. So, if anyone has read an article (a link would be great), or has studied this, I would like to have some quantitative idea of the temperature of the gas. Again, this is for my own curiosity. If no one knows, I may just contact GE or some company and inquire with them :)

 

[AlephZero]

"Not even warm" means the glass envelope is about 5 degrees C above room temperature, for me.

If you want to know what somebody on another forum means by "hot", you need to ask them, not us.

 

[astropi]

"Not even warm" means the glass envelope is about 5 degrees C above room temperature, for me.

If you want to know what somebody on another forum means by "hot", you need to ask them, not us.

Understood, but nonetheless, my original question was about the temperature of the gas :)

 

[krd]

Second question: I took the spectra of an incandescent bulb, and found the peak to be around 660 nm. Wien's law then gives us a temperature of over 4000 K. However, the melting point of tungsten is 3400 K. Therefore, how does one resolve this discrepancy?

The gas would also have a blackbody continuous spectrum. All gases and all solids do. It will just be at a much lower intensity. An ordinary light bulb, with tungsten is a continuous black body spectrum.

What you have there are two spectras - one will be the line spectra, that will not give you the temperature. The other will be the blackbody spectra.

Have you got a pic of the spectra you took?

The black body spectra is a smooth curve with no spikes. What you want to do is remove the spikes. Then you look for the peak on the smooth curve, and by Wien's Law you'll have your temperature.

 

[krd]

The "5000K" is a measure of the color spectrum of the light output, i.e. it the light iis roughly the same as a black body radiator at that temperature (compared with about 6000K for sunlight, and significantly lower temperatures for filament lamps).

No, that's a complete miuse of Wien's law.

With CFL, the energy is mostly limited to discrete line spectra. It's completely misleading, and completely wrong, to take the peak wavelength (a line) and then draw equivalence between temperature of a blackbody and the amount of light you're getting from your 20 watt bulb. You a never getting more or less energy out of your light by using CFL or tungsten.

If you had a 20 watt tungsten bulb, you would get exactly the same amount of light from it as you're CFL - much of that light wouldn't be in the visible spectrum. But you'd have the same energy output.

If you had a blackbody radiator, and Wien's law was giving you 5000k, you would be getting a lot of light, a lot of heat - but you probably wouldn't get that with a 20 watt bulb.


But using the peak wavelength from a CFL to give a temperature is completely wrong.

 

[phyzguy]

The gas inside the tube. AlephZero said his 20-W bulb does not get warm, which implies that the gas inside the lamp is "cool", however I'm not sure that is the case. So, if anyone has read an article (a link would be great), or has studied this, I would like to have some quantitative idea of the temperature of the gas. Again, this is for my own curiosity. If no one knows, I may just contact GE or some company and inquire with them :)

I already linked you to an article that modeled the plasma inside the tube and said that the plasma temperature was around 20,000K. Because the density is low, the heat transfer to the glass envelope is small, so the glass envelope stays cool.

 

[CWatters]

Thanks, but I would like to get a quantitative answer. I have read (online) that some people claim CFLs do get "hot", but it depends on the wattage. That said, I would like to know what "hot" means. I realize this may vary between brands/wattage, but it would be nice to get some idea.

Try looking at it from the point of view of efficiency. An incandescent bulb is only 2-3% efficient so 97-98% of the power consumed is lost as heat.

A typical CFL bulb is 9-11% efficient which is five times better, but even so 90% is still lost as heat. HOWEVER the improved efficiency means that for the same light output only 1/5th of the power is required. So roughly speaking a 10W CFL is equivalent to a 50W Incandescent (argue about that elsewhere!).

The 50W incandescent emits 50 x 0.975 = 48.75 W of heat
The 10W CFL emits 10 x 0.9 = 9W of heat.

9W is still quite a lot in a small package the size of a CFL light bulb.

 

[CWatters]

You can do the same numbers for LEDs.

One major potential issue is the use of capacitors in the body of CFL & LED to provide the correct drive circuit (inverter/converter). Some of these capacitors have a lifetime that is very temperature dependant. It could well be that these fail well before the 30,000 hours typically claimed for the LED die themselves if used in fittings with poor ventilation.

There is potentially some advantage to using 12V LEDs with a remote and easily ventilated transformer.

 

[f95toli]

No, that's a complete miuse of Wien's law.

With CFL, the energy is mostly limited to discrete line spectra. It's completely misleading, and completely wrong, to take the peak wavelength (a line) and then draw equivalence between temperature of a blackbody and the amount of light you're getting from your 20 watt bulb. You a never getting more or less energy out of your light by using CFL or tungsten.

If you had a 20 watt tungsten bulb, you would get exactly the same amount of light from it as you're CFL - much of that light wouldn't be in the visible spectrum. But you'd have the same energy output.

If you had a blackbody radiator, and Wien's law was giving you 5000k, you would be getting a lot of light, a lot of heat - but you probably wouldn't get that with a 20 watt bulb.


But using the peak wavelength from a CFL to give a temperature is completely wrong.

You are missing the point: The 5000K refers to the colour of the light; it does not imply anything about how the light was generated. It is simply a measure of how we perceive the colour of light.
Colour temperature is usual way of specifying this and is used not only for light bulbs but also in cameras; when you are changing the while balance of the camera you are actually telling the camera the colour temperature of the light you are using (e.g. 5600K for Flash photoraphy or 3000K for incandescent light); DSLRs will often allow you to enter the colour temperature directly.

 

[astropi]

I already linked you to an article that modeled the plasma inside the tube and said that the plasma temperature was around 20,000K. Because the density is low, the heat transfer to the glass envelope is small, so the glass envelope stays cool.

Thanks, I did not go through the entire 133-page document :)
However, if you can recall where (page number) it says that the plasma temperature is around 20,000 K I would be interested in looking up the reference. What you said entirely makes sense, and I would like to read up on the physics behind it.

 

[sophiecentaur]

You are missing the point: The 5000K refers to the colour of the light; it does not imply anything about how the light was generated. It is simply a measure of how we perceive the colour of light.
Colour temperature is usual way of specifying this and is used not only for light bulbs but also in cameras; when you are changing the while balance of the camera you are actually telling the camera the colour temperature of the light you are using (e.g. 5600K for Flash photoraphy or 3000K for incandescent light); DSLRs will often allow you to enter the colour temperature directly.

The problem is that you can't really specify the subjective effect of a line spectrum illuminant in a single figure. You can project the CFL light onto a screen and 'match it' to a hot tungsten filament. BUT that doesn't tell you at all, what coloured objects will actually look like under that illumination. There are many different combinations of lines that can 'match' a tungsten filament but the gaps between the lines mean that there is no information about the reflectivity of a pigment in those gaps.
The use of Colour temperature in photography is a rather different matter. The analysis of the three colour sensors is essentially, broad band and is matched fairly well to the eye's analysis. Hence you can do a fair job of colour correcting for different lighting conditions when the illuminant is a basic black body spectrum. If you try it with fluorescent lighting it tends to fall to pieces and, at least on cameras I have used, doesn't commit itself to a 'temperature'. It just gives you a set of options for the least worst result. (Which sort of proves my original point, actually)

 

[phyzguy]

Thanks, I did not go through the entire 133-page document :)
However, if you can recall where (page number) it says that the plasma temperature is around 20,000 K I would be interested in looking up the reference. What you said entirely makes sense, and I would like to read up on the physics behind it.

There's a graph on page 97 showing the electron temperature as a function of radius. This shows it is more like 12,000 K. I got the 20,000K from the table on page 72, which is discussing the temperature of the tail of the electron distribution. Be aware that temperatures of discharge plasmas is usually a complex subject, since the distribution is usually not Maxwellian, and the electron and ion temperatures are usually different. But no matter how you look at it, it is quite hot.