I did try constraining just the edges of the hole. That will involve modeling multiple parts and probably using contact instead of a hard constraint. With conical mirrors (supported at a hole) they tend to use an O-ring of some elastic material at the CG as the lateral support and some other vertical support near the hole at the back. I was already unsure if the constraints were correct. And exactly this is why I was asking my question. Having been a professional FEA-analist, I can tell you that it is easy to get results. You have to treat a mirror radially exactly as you should treat a mirror axially: use well-defined constraints because on submicron scale and deformations that are so critical, a full contact is an impossibility. In reality the glue in this case is on one line in perfect tension, one line in perfect compression and in the rest of the hole it is somewhere in between with an increasing amount of shear, if they are in full contact at all (and no the glue isn't in full contact). What your model shows now, is setting all the displacements around the hole on zero. This way of modelling a mirror with such constraints is not correct. Mirrors are very difficult to properly set up and constrain at the horizon, but relatively easy at the zenith. It is REALLY easy to add artificial stiffness without proper constraint methods, and those methods can be complex. (The supports are at radius 0.87" or a bit outside the hole.)Įdited by ckh, 19 January 2019 - 07:38 PM.Īs I learned, you really need to know what your are doing constraining a model. However, it's not supported by the hole so maybe the three point support isn't as good. But plop used a 3 point support of course, instead of supporting it by the hole. The trick is to use a negative focal length to make the Plop front of the mirror convex (and input the 1/16" edge thickness, the back is always flat in Plop). I also analyzed the convex-back mirror in Plop using a trick. I can probably analyze meniscus mirrors accurately this way. There is obviously a lot to discover here. In fact it may be better supported because the deformation is uniform from center to edge which tends to keep the surface the nearly the same geometrical shape (parabolic).Īlso the thickest part is only 1/2" and the rest of the mirror is thinner than that. So the thin mirror is 1/7th the weight and is equally well supported. The thin mirror with the convex back has a volume of 3.8 in 3 while the 1" thick, flat-back mirror has a volume of 28.3 in 3. The thin mirror with the convex back has a surface PV of 5.3 nm while the 1" thick flat-back mirror has a PV of 4 nm, a difference of 1.3 nm (about 1/400th of a wave surface). Plop gives this result for that mirror with three optimal support points: It is 6" in diameter but with a flat back and 1" thick (standard thickness). The deformation is symmetrical and gradually increases from 0 nm at the hole to 5.3 nm at the edge. I can't say what the RMS is but the deformation is very smooth (maybe 1 nm RMS). The maximum deformation occurs at the edge and is 5.28 nm which is quite good. The mirror is supported on the inside surface of the 1" hole through the centerįusion 360 created mesh for it and then analyzed the deformation under the influence of gravity with the face pointing to the zenith. It's essentially like a conical mirror only the back is spherical instead. The bronze color is the edge 1/16" thick. This is what it looks like from the side: The blue part is the convex back with a 1" hole in the middle. For my first example, I choose to model something simple:įocal length: large (I modeled the front of the mirror as flat for simplicity)īack of mirror: Convex (sphere) with radius 9" I wanted to test the idea that a mirror with a convex back and thin edge would be much lighter and easier to support with low distortion than a flat back mirror.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |