Can I digitize a 2D curve during measuring?

[[No...]]

First of all: My knowledge about the curve option is close to zero.
Second: I want to do this on the O-Inspect.

The following problem:
I have a part with a shape similar to a gear with a few missing teeth. The rotation of the part on the CMM is unknown (but the position is) and there's no fixture.
I want to measure the shape of the part as a 2D curve and generate an alignment based on it (XY translation and Z rotation). But because I don't know how the shape is rotated, I'd like to scan the curve during each measurement in the same way as it is done when digitizing an unknown curve. Then I want to fit the curve to the nominal shape and get a coordinate system from it.

Is this possible? Or can it be accomplished by a different approach?

[[Ow...]]

I've never digitized in CNC mode so, I don't even know if it's possible.

However,
If the curve shape and size is fairly close, you should be able to digitize it to get its nominal's and then scan the feature and use the curve best fit to create a secondary alignment. I’m not sure it could be best fit for the original base alignment.

You'd have to have a way to locate each subsequent part to where the curve was close to the same location and orientation.

I'm not on a computer with Calypso on it or I'd give you some screen shots of how to do it but,…… basically, digitize the curve, change the amount of points to be 3 x times the amount of nominals, turn on the filters and then scan it. If it’s a curve with a lot of small bends or acute angles in it, I’ve always got better more actual nominal results digitizing with single points. Click the curve features evaluation tab and select the coordinates you want it best fit to and there’s also an option to create an alignment off of that best fit. It will place the best alignment in the characteristics and you would assign that alignment to all the features needed to align to its new position.
Clear as mud? Probably not but, hopefully somebody else can add to it if needed. I'll try and check in tomorrow to see if you get it figured out.

[[Ri...]]

You can do Unknown Contour in CNC mode. If you have already made a nominal and go into the segment, check mark Unknown Contour at the top left. You will need to set it up the same way with a start/stop point, space axis, direction, etc.

I'm unsure if you can do an Unknown Contour with optics though. I've never tried.

[[No...]]

We will find out today 😃

This is the outer shape of my part:
The outer diameter is 60 mm. I need to have either the X or Y axis run through one group of three "teeth", but it has to work without a fixture and should not need any user interaction. I know (or can find) the center of the part, but don't know its rotation.
Using a curve was just an idea because I could easily derive an alignment from it. But I'm open for alternative ideas.

[[No...]]

Let me announce that unknown contour does indeed work on the O-Inspect.
But I can't find the usual optics settings in the strategy, like the width of the AOI etc.
Since the "teeth" in my part's shape are quite big and I can't be sure whether measurement starts at a tooth or not, I think I need a wide search area, but where can I set it? There's no optics tab in unknown contour. 😕

1 hour later......

Aaaarrrrgh, this is so weird!! I can't even set proper optic parameters on a normal curve segment. I extracted the curve from the CAD model as a segment for now, but I can set NONE of the important parameters! No rising or falling edge detection, no N-th number of edge to be detected. I can't even set the light profile to be used in the strategy. It always shows "OFF". I have to go to the measurement plan editor to do this! In a segment I can at least set the AOI width and a few other parameters, but in an unknown contour I can edit nothing. Zeiss, what have you done there??? Or am I just dumb? 😡

[[Ow...]]

Somehow, I read right past the fact you were doing it with an O-Inspect....smh.
Best of luck.

[[No...]]

No problem, Owen, most of what you said also applies to the O-Inspect.
But I'm really (negatively) amazed how Zeiss always manages to include such bloopers in their software. So now I am where I've been so many times on the O-Inspect: 1 inch before the finish line, and then Calypso trips me up.

I just pray that now that Zeiss has bought GOM they won't start messing up their exquisite software too. 😡

[[No...]]

Plan B:
Instead of scanning an unknown contour I'm now trying to use points recalled from a circle measurement. I think I made some progress this way, but somehow I can't get the curve best fit right (I told you I have little knowledge about the curve option).
The rotation of the fitted curve is always a bit off, like this:
My questions are:

• What should be bigger: The number of nominal curve points or the number of measured points? And what is the optimal ratio?
• How should I set the curve fitting parameters to get a proper alignment from it? I tried Gauss and restricted it to XY translation and Z rotation, but the result is like in the screenshot (at best). What am I doing wrong?

[[No...]]

I just don't get it.
No matter which fitting method I choose, I get a fitted curve like in the picture above (the green dotted line - or is that something else??) but the alignment created doesn't reflect that. I get Z rotation angles smaller than 1 degree while the part is rotated a good 45 degrees against the curve extracted from the CAD. Why?

[[Ow...]]

3 measurement points to 1 nominal point is the preferred ratio.
The Z rotation being shown might be in radians, not degrees?
You might be able to scan two of the small planes on each side of the middle tooth and create a symmetry from them to use as your planar axis?
I've never used an O-inspect but, I would think there would have to be a simple way to do what you're trying to do with the optical.
We used to have a system at former place of employment that used an optical system to align/best fit 3-piece profile cutters being sharpened , if it could do it, the o-inspect should be able to do the same.
Not a lot of o-inspect people on here so, your best bet might just be to call tech support.
Before I spent a lot of time trying to do it with probing, I'd see if the optical could do it quicker.

[[No...]]

Please sign in to view this quote.

Ok, noted. Thanks!

Please sign in to view this quote.

I don't know, but the axes just pointed in the wrong directions. After a million of tries I think that kind of best fit just fails because there's too much symmetry in it. I know that behavior from the GOM software. Sometimes the best fit is just wrong. In GOM you can then set an anchor point on the CAD and the point set to fit. The software then keeps the two points as close together as possible. But even that wouldn't have been an option here, as it would need user interaction. I don't know what that dotted green line means, but it's obviously not the fitted curve. Maybe it's a preview of what the resulting spline will look like.

Please sign in to view this quote.

In the end I did something like that. I placed the part against a 90 degree angle with the smaller diameter and turned it until the outer teeth touched the angle. Repeatable enough to find the middle tooth for an alignment.

Please sign in to view this quote.

HAR HAR [sarcastic], I was thinking the same when we bought that thing years ago, but NO! 😡
This is not the first time I fail on this machine after investing hours in a special problem that looked easy to solve at first.
As long as you do boring everyday stuff with it you're mostly fine. But as soon as you're trying to be creative and want to achieve something unusual, you're doomed.

Please sign in to view this quote.

guess why 🙄

Please sign in to view this quote.

HAR HAR again. 🙁 I wrote an email to support, asking why I can't set all optical parameters in curves.
Answer: This has not yet been implemented in the software.
(Note: The O-Inspect is being sold with the curve option already included - but Calypso can't handle it ! )

And the answer went on: If I would like to place that as a software wish, I should fill out an attached form, a two-page document very similar to the software wish section on the old Zeiss portal. It asks things like "why is this feature useful?", "what are the advantages?", "Can you solve your problem without it?"

I guess this is an example of the famous German bureaucracy - to get (or not) a very basic missing feature, I first have to fill out a form. Am I too harsh if I call this ridiculous? 🙄

[[Ow...]]

Quote "Am I too harsh if I call this ridiculous?".
No, not a all.
In fact, I've seen much more distasteful terminology used here but, I have PCDemon friends that are even worse on their forums.
I'm jealous you got an e-mail reply.... 😮