Melt variation on relative partial dispersion
Posted 10 December 2012 - 07:36 AM
I would like to know what is a typical melt-to-melt variation on partial dispersion (lets say for e,C lines) for a given glass.
I know that dispersion (nF-nC) could be controlled at level of +-0.00002 (and if one asks the company even on 6th digit level), and that melt-to-melt variations are even larger. But I can't find the number for relative partial dispersion. Since it is a ratio of two dispersions, the melt-to-melt variations in refractive indices probably mostly cancel out. Is it safe to assume, that the relative partial dispersion is controlled between various melts at the level of 0.0001 (the glass catalogues usually state relative partial dispersion on 4 digits)?
Posted 10 December 2012 - 08:55 AM
I did my first research on this problem by calculating the partials out of an old Schott catalog, using the one of first digital calculators, a Friden, which weighed 15 pounds and had 4 functions!
Posted 10 December 2012 - 09:23 AM
This is not the case in the ultraviolet (like for U-BK7), where partials can and do vary more.
Posted 11 December 2012 - 04:05 AM
Do you know if one can ask the glass companies to measure it with 0.0001 precision (for reasonable price)? I know, one can ask for example OHara company to measure dispersion at +-0.000003 level, but this would still give about +-0.0004 precision on relative partial dispersion for e,C lines. But may be there is some other clever way how to measure the relative partial dispersion.
Posted 11 December 2012 - 06:48 AM
What's not clear to me is your high sensitivity to index and partials. Normally one does a "melt re-comp" on a lens based on real glass data instead of catalog values. After real glass blanks are received along with their melt pedigree, a separate glass catalog is created in the design software that exactly models the measured glass pieces. The design is then re-optimized and re-toleranced using real data, which usually restores full performance. If you're just making one lens, this is no biggie. If you're making a small production run of, say, 100 lens assemblies, you'd better already have all the glass on hand, with melt data, otherwise you'll be tweaking tooling during production, which eats up profit.
Posted 11 December 2012 - 07:55 AM
Sorry, if I was not clear enough. My original hope was that I would not need to worry about melt-to-melt differences if the relative partial dispersion stays at about +-0.0002 level from the catalogue listed numbers. Of course, once I know the relative partial dispersions, I can optimize the performance accordingly.