What's new
What's new

What's the most accurate way to find the center of a sphere?

Stephan Spears

Aluminum
Joined
Apr 12, 2014
Location
AB,Canada
Hello,

I have a sphere clamp on a vertical 3-axis mill. And I would like to know the coordinate of the sphere in machine coordinate system. What is the most accurate way to do it? By the way, the sphericity of the sphere is very good.

11.jpg


So far, as for its position in XY plane, I can use a dial indicator loaded onto the spindle to touch the sphere and rotate the spindle. Meanwhile, I keep adjusting the spindle position until the runout is minimum.

However, what can I do to find its center in Z direction?

Thanks in advance.

Steph
 
Assuming no probe, dial indicator to .0001 (as you've done), then tool touch off top of sphere (using for instance .002 stainless steel shim as intermediary), with tool offset by radius to avoid influence of bottom relief. Then add ball radius plus shim value to Z offset.

As always, your process accuracy determines final results.
 
Use your indicator to find the high spot of the sphere. Zero the indicator. Then measure to a known Z value. Subtract the radius of the sphere and you have your Z.
 
The tool I would suggest is a Haimer 3d Taster. You can touch three points around the diameter, then come down on the top for a fourth point.
I plug in the coordinates for all four points into an online calculator. It shows me the calculated diameter of the sphere which is usually .001-.002 off from what it should be. Maybe some will laugh at that, but works for me and its the best I can do without a probe, (and I know I have runout issues in the spindle too).

I think I use this one. There is another one but it rounds off to 2 decimals. No good. Sphere Solver - Calculator finds the center and radius of a sphere given four points in Cartesian coordinates
 
or, you can use a block of known size, like a 1-2-3 block. You can find your Z to the top of the block, then use a depth mic to measure down to the top of the sphere. I just find measuring down with a depth mic easier thang going back and forth with an indicator trying to find the high spot
 
Most accurate ?
Use real proper sensors/switches from metrol japan.
Accurate to 1 micron.
Probe.

About 100$ each.

More accurate is probably interferometric stuff via a special light and a glass substrate.

Reasonably accurate and dirt cheap is a good digital 1 micron dti.
Accurate to about 2 microns, and repeatable to about the same.

All 3 options are likely more accurate than your machine is repeatable.
 








 
Back
Top