Tungsten Filament Emissivity Assistance requested

[seycyrus]

Hello,

I have a Tungsten filament radiance source that has been calibrated in the UV region and would like to extrapolate the radiance to longer wavelengths. By calibrated, I mean, a NIST plot of spectral radiance v. wavelength at a certain set of operating conditions.

I'm new to this area, so i will start off with a big "I think" and apply it to everything I write from now on :)

I have an old lamp with a more complete (larger range of wavelengths) calibration. It is fairly obvious that I cannot treat it as a simple grey body as the old report contains words like "brightness temperature" at different wavelengths. In other words, one single T won't fit all the data.

That lamp HAD spectral radiance values in my region of interest, but that lamp is now broken :) Now I must make do with the source that has a much smaller range of calibration and would like to figure out what the spectral radiance would be in my range of interest.

I came upon a paper that indicated that the emissivity had a certain wavelength dependence and then used ORIGIN to do a multi-variable nlsf. I was pretty proud of myself when I could fit the broken bulbs data set "perfectly".

I though that I might be able to do the same sort of nlsf on the bulb with the smaller set of calibration data and then use the fit to get values for the wavelengths I wanted.

But when I tested this with a subset of the broken bulbs data, it didn't work. The peak position was way off.

Anyone have any idea on how to proceed? Is there a standard way to calculate the spectral radiance of a given metal at an arbitrary wavelength (besides paying NIST to do it).

Thank you for your time

 

[GT1]

I found on my heat transfer book graphs of Spectral normal emissivity vs. Wavelength and Total normal emissivity vs. Temperature for Tungsten - you can start from there. In which temperature and wavelength are you interesting ?

 

[seycyrus]

I would like to be able to predict the spectral radiance from the near UV to the near IR for a specific filament.

The literature I have found suggests that emissivity plots are very sample dependent, which is why I was thinking about fitting the data I already had, and using the results to predict the spectral values at higher wavelength.

 

[GT1]

by looking at the graphs it's seems that the normal emissivity doesn't change much for wavelengths shorter then about 0.8 micron and from 0.8 micron to 6 micron it changes almost linearly. if the trend is the same for every sample and only the values are different you can use your data and build a graph based on your data and the textbooks slopes.

 

[seycyrus]

GT1

Thanks for your help. My problem is that I have a few experimental plots of tungsten emissivity and they DO seem to differ in many ways. Surface conditions and such seem to be a big factor.

I was hoping to find an expression with suitable constants that would "absorb" such sample-dependent parameters.

I've found reference to a Hayens-Rubens relation that says that the emissivity should go some thing like, E = a/(lambda)^1/2 + b/(lambda) + c/(lambda)^3/2

But I'm not sure where that is derived from.