Intensity Calibration Idea

[Robert Dyer]

OK, I'm stepping into optical waters way over my head, but sometimes ideas, or their better derivatives, come out of right field...

 

I've been intrigued and very interested in all the discussion about turning an inexpensive spectrometer into a more "advanced" device. I'm learning tons, so I really appreciate all that's been shared. What I've distilled it down to in my brain is the following:

1) The resolution of the DVD as a refraction grating is good enough for most purposes.

2) Wavelength distortion on the ends of the spectrum may be a problem, but new calibration software could help.

3) Intensity distortions due to the imaging device's Bayer filter may be a problem, but again, new calibration software could help.

4) New software won't fix that we currently get qualitative (wavelength) data, but not quantitative (intensity) data.

5) A solution might be some sort of standardized light source with known wavelength AND intensity.

 

I was wondering if a different approach to the last problem might work, but again I'm over my head optically, so this may be missing something elementary. It seems like we already have a cheap, readily available standardized source for calibration in the CF bulb. And actually for most people this is probably enough. It does give us wavelength info that's sufficient for many purposes. And with newer, better calibration software it might be even better. (And my current setup is resolving the two green lines.)

 

Instead of a true standardized source, would it be possible to add a second step to the existing calibration procedure for those people who want more from their device? Could we use an ND filter, or possibly two different ND filters? We could take our first reading of the CF bulb with no filter as usual, and then take a second (or third) with a filter over the input. While neither reading would be absolute, their relative measurements should be. Could then a true base-line for the entire device be computed?

 

ND filters seem to be readily available from camera stores along with their transmission spectra. They're not prohibitively expensive for the people who want this kind of accuracy. And maybe someone knows of a really cheap alternative - like whichever genius figured out that a chunk of a cheap DVD could replace an expensive diffraction grating!! :))

 

Anyway, those are my thoughts. I wouldn't be surprised is there is a really good reason this won't work, but who knows....

 

Robert

[Dave Stoft]

Good; it's the only way to get your feet wet. You now grasp what a thorny problem this is to "solve" and that any "solution" is actually just (hopefully) significantly measurable improvements...

- Yes, the DVD lines have close enough spacing that it is not the primary source of error
- Yes CFLs are a sufficient source for a wavelength cal
- Yes, with a single 45-deg incident configuration and camera ortogonal to grating, the math is direct for linear correction s/w
- But yes, s/w add-ons can provide at least a measure of improvement toward gain correction over the bandwidth -- like with most measurement details, is a matter of degree and at what cost. We're trying to keep the cost low by becoming more clever.

Yes, using broadband attenuation (i.e. your ND filter idea) is useful and can observe amplitude linearity and even be used to expand the dynamic range (i.e. HDR techniques like those attempted). However, ND filters are, by nature, broadband 'flat' devices which affect all wavelengths equally so, alone, they do not provide direct data for gain correction. In theory, however, if the source were well characterized (know spectra) but had excess roll-off, taking tow sets of data, one with and one w/o an ND filter for attenuation, and then carefully combining the data the "less than flat" source could be made more useful for gain correction by removing some of that source's dynamic range. However, one would also need to add at lease some uncertainty if the spectral response of the filter were not known. So, useful, but more complex.