I want to find the rotational center of an elliptical bowl, using the internal and external surfaces (one at a time). The thought crossed my mind that I could use a reflectorless robotic total station to observe three or more spots on the surface, essentially using 3D resection to locate the instrument relative to the bowl's surface. The bowls range from 10" to 20" in diameter.
Then, use a laser pointer or some other method to "stake out" a point at the center of the surface, relative to the other points measured. Distance is not really a factor here; just angular location.
If this were to be possible, it raises several questions:
1. What is the minimum operating distance of most modern reflectors total stations?
2. How far off normal to the surface can a reflectorless total station record precise distances?
3. How do you "stake out" without a prism?
Alternatively, could this be done with a laser imaging scanner, that would scan the surface and compute the center?
Any other suggestions on how to do this?
Send me a bowl and I can scan it. I met a fellow using the same scanner who is scanning parts in power plants. He uses ping pong balls for registration. I just happened to pick up a package of ping pong balls just in case...
If you do mesh, try ThinkBox. They had an interesting mesher in beta.
> Am partially kidding, but I always like to think of how such a problem might have been solved in antiquity... put a drop of mercury in the bowl and see where it goes 😉 (I know that does not completely answer the questions and bowl orientation relative to gravitational plumb and /or the observing instrument would be key).
I like the idea. Might work for the ID. Now, how about that OD, lol.
The problem is that it relies on the edge of the bowl, which can't be relied on. We need the mathematical "best fit" for the actual surface.
>
> Scanning and polymesh modelling; then run contours at a fine interval (of course that depends on the orientation of the bowl being scanned - another rough method, but depends on what the goal-for-the-bowl is. Are you looking for all axis of rotation?
There should be only one "closest fit" axis about which to rotate the ellipsoid, in order to turn it to re-establish it as a sphere.
> Send me a bowl and I can scan it. I met a fellow using the same scanner who is scanning parts in power plants. He uses ping pong balls for registration. I just happened to pick up a package of ping pong balls just in case...
> If you do mesh, try ThinkBox. They had an interesting mesher in beta.
I have no scanning or meshing capabilities. I can send one, but beyond the scanning, do you have the mathematical (or CAD) tools to determine the center?
> The scan sounds best, then you could work the inside and outside. Still some speculation here as am still a little unclear as to this "bowl......
Imagine, if you will, that the bowl is the planet earth, cut in half at the equator. The outside layer is the geoid model and the inside is the ellipsoid, or some variation thereof; representing, more or less, the line between the crust and magma.
Earth
Centered
Earth
Fixed
Dougie
It's a bowl, sure enough...
>- a photo would have helped).
Here's the 14" model, the biggest of the lot.
The process needs to be automated, and precision probably only needs to be +/-.005'
It's a bowl, sure enough...
I would tend to want to use a mechanical jig/story stick type method over some fancy laser scanner.
Wood workers solve these problems with low tech solutions.
You may have to build your self some sort of adjustable caliper.
It's a bowl, sure enough...
Looks like it is already contoured, just interpolate the contours.:-P
It's a bowl, sure enough...
> Cool.
> Automating might be a challenge. A CMM table (metrology) might be able to record a minimum number of observed points and some routine for least square fit for non-linear objects.
> There are metrology service providers that maybe could run some sample ones, and a jig could be designed?
I was thinking something like a "jack" (from the old days of playing "jacks").
The point would float, and the two cross pieces would have spring loaded dial gauges, but that each pair would be linked so that the major and minor axes could be gauged...kind of like a tire balancing machine...back and forth back and forth until the cross point of the two axes is determined. Then mark the point or something. Still working on it...Thanks for the ideas.
