Scan until probe has 3 contact points

[Wes77]

Hello,

I've been using this forum for years and normally it helps solving all my questions.
But today I've met my match, and am hoping some of you could help.

The machine is a Leitz PMM-G with LSP-S2.2 probehead.

De sphere with the blue arrow is the tip. It's diameter is Ø29.368 mm.
With this probe it should scan the 2x radius R15 at the green arrows. The centerpoints of these radii are not the same, they are 0.2 mm apart.
After that it should scan to the plane with the red arrow.
So basicly, when the tip touches all 3 surfaces the hitpoint should be taken..

Scan_.jpg
I have another part with the same procedure, but with the plane (red arrow) not perpendiculair to the radii.
A bit harder to achieve maybe.

I hope someone has an idea on how to get this working.

Thanks.

[AndersI]

A 29.368 tip can't touch all three surfaces at the same time (30.2 betwen green surfaces), so what are you actually asking?

[Wes77]

It can touch all 3 surfaces. The R15 surfaces are about 5mm wide. The pocket above the radii is 31 mm wide.

Hexagon NL said I had to use scan => center.
He tested with a v-groove without problems. But now when he tries on a non flat surfaces, he can't get it to work.
He is confident that our machine, head and software are capable, but he doesn't know how.

[AndersI]

But not at the same time! You have to be much more clear, for us to understand what you want to do.

[BIGWIG7]

why cant you use a smaller probe tip?

[JEFMAN]

A center scan cannot work accurately here.
A center scan has to search the surface (for example right radius), the goes through the plane, and stop when all movement in the vector direction is impossible.
In your case, the center scan will slip on the plane, and will be lost on the left radius.
You should use a linear scan, if possible with a smaller ball.

[Wes77]

The actual product that sits in these 3 pockets, has 3 spheres 29.368 mm mounted.
Our customer has made this procedure and also delivered us the probes.
They already did testing on a Zeiss machine, and that's how they know it's possible.
But the part is large, and they don't have a large Zeiss.
And we, a regular supplier, have a large Leitz.

[JEFMAN]

What is the dimension that you're looking for ?

[Wes77]

Per pocket there is a centerpoint of the probe.
Three pockets makes 3 points of which a plane and a circle will be constructed.
The parallelism of the plane and the position of the circle are important. Expectations are within 0.005 mm, which is normal for the size of our CMM.

[NinjaBadger]

From your description I'm still having difficulty understanding the geometry you're describing and what you're actually after in terms of measurements.

[UKCMM]

To make life simple place the 29.368 sphere in the part locating as you need then probe the sphere this will give the centre point required.

[NinjaBadger]

great suggestion!

[AndersI]

Can anyone explain what I don't understand - how can a

Ø29.368 sphere touch all three surfaces at the same time when the curved surfaces are 30.2 mm distant?

[JEFMAN]

Not me !
If Zeiss can do it, it's magic !

[UKCMM]

The 2 partial radius are at the bottom locate the Z and X, the plane back face locates the Y looking at the picture.

[AndersI]

Ahh, I thought the R15 were the *sides* of the slot, didn't realize there were radii at the bottom. I would definitely suggest putting a sphere there and measure on that...

[Wes77]

UKCMM is correct

Our customer investigated this before they ever came to us.
They thought a partial sphere would give too much measuring error. That's why they came up with ideas like drilling a hole in and make a plane on the sphere, Then they could measure the hole and plane, and could calculate the spheres centerpoint. Less accurate than measuring a partial sphere in my opinion.
But then Zeiss did testing and told our customer that it was possible. So our customer had a plan B, measure the spheres, but they put it aside because said it was possible.
When they came to us, we placed the problem at Hexagon, and they did testing on a v-groove with flat surfaces with Scan -> Center. But that;s not the same a sa v-groove with curved surface.
I and our customer re now starting to think that Zeiss did the exact same testing as Hexagon. So it's highly probable that it cannot be done.

Well, Hexagon Germany dis come up with this idea (below part is not the same as the above photo, but the same idea):

Naamloos.jpg

Scan both radii with the adjecent part in between.
Construct a circle using Scan Minimum with a fixed radius. Repeat a few millimeter further.
Construct a cylinder from the 2 circles.
Measure the plane and construct an ofset plane with sphere radius.
Construct a pierce point from the circle and offset plane. And this is the centerpoint of the sphere we need.
This is plan C.

We are going to test both plan B (measure on the spheres) and plan C and compare them.
But if anyone has an answer to plan A, please inform me.

Thanks to everyone who has taken the tim to look at this.
Very much appreciated

[JEFMAN]

What about measuring some vector points (or scan), whatever the tip diameter, and then construct a max inscribed sphere, or a fixed radius sphere on them ?
As AndersI , I didn't realized where were the radii...
Thx UKCMM for the explanation !!!!!