Hallo Max,
Thank you for these interesting test results.
I assume that we are talking about the whiffletree design with support points at 14° and 45° from the vertical.
First of all, it's good to underline that according to the FE analyses, the ideal positions for a whiffletree are 22.5° and 67.5° - in other words 4 points separated by 45° on the bottom of the mirror. Deviating from these values removes all the advantages of using 4 contact points instead of 2. The results computed for the 14°/45° whiffletree are worse than for two simple 90° edge rollers.
I have run the FE analyses for the test configurations on the 350x33 mm, f=1582 mm mirror with the 14/45° whiffletree.
Here are the analysis results for the 4 offsets (nm on the surface, waves on the wavefront):
- Abweichung -2 mm: 11.1 nm or 0.042 wave (test result: 0.063 wave)
- Abweichung 0 mm: 8.2 nm or 0.031 wave (test result: 0.048 wave)
- Abweichung +2 mm: 22.5 nm or 0.085 wave (test result: 0.218 wave)
- Abweichung +4 mm: 37.5 nm or 0.151 wave (test result: 0.337 wave)
Qualitatively the test results correspond reasonably well with the FE analysis. For example, the asymmetry of the error with respect to the COG: the case below the COG (Abweichung -2 mm) is a lot better than above (Abweichung +2 mm).
Quantitatively, there remains a factor of about 2 between the FEA results and the tests. This might come from an incorrect model for the back support which is important for the case with an offset from the COG.
Anyway, I would never claim that these FE analyses are perfect, but I hope that at least in a qualitative way they can be used to show which configurations will work well and which will not.
Grüsse,
Robert Houdart