Originally Posted by jhaywood
Absolutely, I was not talking about a traceable source, I was thinking about something that might be good enough to provide a degree of confidence in a sensor vs assuming it was OK.
Short of doing sampling, at a minimum, of the inbound components, you ARE making this assumption. Even with sampling, you are making an assumption, as well, that your sample is, in fact, representative of the products you are receiving. Ideally, a DIYer would test each of the filters he or she planned to use, since you aren't heading for large-scale manufacturing, or you would use equipment from a "known good" source where someone else has done enough testing to verify the quality of the products. Averages are entertaining here, but standard deviations (and standard error) are where the rubber meets the road.
The questions are valid - Would a regulated halogen source have a stable enough spectral output to be considered constant? Are the filters stable?
Filter stability is a different issue. When people talk about the stability of a filter, it is about a decay/degradation over time for an individual unit, not variance across units. As for a halogen lamp, it is probably "good enough" for DIY use if you had the ability to measure it at that given instant with a high quality meter. You absolutely cannot trust one over time (e.g., read it today, come back a month later, then use it as a reference without re-measuring).
I believe that dichroic filters are stable - they produce the band pass by using an interference layer that is sandwiched/sealed between two layer of clear glass. I've used these type of filters in signal detection for plasma processes without seeing degradation over multiple years of use.
Dichroic filters are good because of a number of factors. However, good ones tend to be expensive, thus violating the "cheap" ideal.
If luminosity is maintained over the bulb lifetime,[..]
Which it isn't, as most any projector owner can attest.
does this mean that we don't see significant filament temp changes & hence the spectral output of the remains acceptably constant?
No. The spectral output of a lamp will change over time.
Do you have any data on the magnitude of the change in the emission spectra that might be expected & how that would impact the CIE coordinates, all other things being equal?
For lamps in my house? No. For the lamps in some of the projectors I have owned? I did at one point, but I did not keep the data when the projectors went to new homes. My current projector, Sony VW60, I've been meaning to some longitudinal studies on, but have not had time to do the basic, background measurements. I've been looking into adding a stable, NIST-traceable light source to my lab, but we will probably put one up in Seattle, first, rather than with me.