Giving Radii a running start.

[[Me...]]

I was thinking about how small sections and sizes of radii, 90 deg. or less, are difficult to check. I was thinking of an idea that might help. So I'll run it by the Geniuses. What if we were able to, (or maybe we are able to and I just don't know this), give the probe a running start from a tangent feature, a line for instance. It starts back 2mm or so to eliminate the start up error.
Good - Bad?

[[To...]]

As long as the data points are masked, (first 1/4 second by default) I suppose that would work technically. Slower scanning speed would also create a smaller masked points range.

[[Aa...]]

I'm no genius, but one thing I've found helps with really small angle ranges is scanning both directions.

[[Me...]]

Aaron do you then recall both?

[[Me...]]

Tom I"ll screw around with it to see if I can "lie" to the software and do some masking. But I see this as something that Zeiss could do as an added feature. You would pick the lead in feature. The points from that lead in would be ignored and data from the radius is all that would be collected.
I'll also try Aaron's suggestion.

[[La...]]

My go-to for small rads is to use curve. Cut the curve around adjacent features and that's how I get my running start. Then I use recall feature points and the click-and-drag box around the rad specifically.

[[To...]]

Please sign in to view this quote.

What do you do if there are no adjacent surfaces?

[[La...]]

Extremely slow scan, super high point density, appropriate sized stylus. I've heard that constraints help in this situation, although I've never had to use them personally. There is a document about it in the e-learning.

[[Me...]]

Nice Laura. Unfortunately I don't have that option.

[[Ri...]]

Good/Bad? It depends. Are small circle segment callouts on your prints typically just size or location or both? What are tolerances, typically? What is the material and manufacturing method? Are run time and throughput issues that you have to worry about? How do you determine acceptable repeatability? Do you have a second inspection method to confirm the answer you're getting from the CMM? The answers to these questions are critical to determining a dependable strategy for these kinds of features. This problem is as old as CMMs. The software takes the measured data from your circle segment and constructs a perfect 360 degree circle. Any size, location, and form error on your part is magnified and can result in a wide range of results. Constraints on size or location are almost always required to reduce that range. I like following up a circle measurement of a small arc with a group of Plane Points and then using 2D Polar Distance from a theoretical point at nominal location if I have time and if tolerances are tight enough to warrant the extra effort. The bottom line is testing. Measure a lot of parts, many times under strictly controlled conditions with one method. Change the method and rinse and repeat. The answer to your question will be in the numbers.

[[To...]]

Please sign in to view this quote.

Richard, is this process any different than measuring the circle with n single points and using Radius Measurement?

[[Ma...]]

My goto, like Laura, is a curve. Then recall points into a circle using the select points method. Or better yet I use the curve and do an CAD evaluation. If you use this method be sure and go to the Curve's Evaluation and click on Nominal Vector Direction and click Parameters. Then right click in the CAD window and select Show Point Numbers. Then in your list select, by holding shift or CTRL, and pick the numbers that fall outside of the circle you are trying to get. This will not calculate the points you have selected.

Mark

[[Me...]]

The thing that kills me is that I have an old Mitutoyo manual cmm that will eat these up for breakfast. We added CMM Manager for the software but damn......

[[Ma...]]

Kevin

You can do the same thing in CMM Manager, but not the place to discuss that.

Mark

[[Ri...]]

Tom, to answer your question, it is fundamentally the same but by using a group of Plane Points you will bypass the software creating a 360 degree circle from the limited data set. Also, this allows you to scan the first measurement and collect 500 points. If I were to follow that up with Radius Measurement then I would have 500 results to look at or the extreme values or some other choice there. With the group of Plane Points I have greater control and can tailor my methods more specifically for the part.

[[Ow...]]

On small arcs/radius, I will avoid scanning and use single points every time and only scan when the radius is fairly large.
However, inspection time and form isn't that critical in most of our applications.
Even using single points for form, lots of them, produces more accurate results in my opinion.
I'll use single points even when evaluating small radius with curve.

Like Richard said, there are a lot of variables measuring arc's and constraints are almost always needed when measuring less than 90% of a circle and even then repeatability needs to be proven.

I would question the repeatability and accuracy of any CMM that eats radius for lunch.

[[Me...]]

Like Owen, I use single points as well. This opinion is free of charge.

[[Ro...]]

+1 Owen
+1 Dave
I have tried & tried & tried many different methods, but when dealing with small sections or partial features, i always use points.
I know the purists like scanning everything, but points for small features is where it's at!

and you don't really need very many points, 5 or 6 should do you fine.

[[Me...]]

Please sign in to view this quote.

It was breakfast. But to your point and Roberto's it is using points and I'll work towards incorporating points more often.
Plus I'd like to try the idea of scanning in both directions and recall the points.

[[Da...]]

Please sign in to view this quote.

Laura, please provide the link to this e-learning.

[[Ri...]]

Dane, I'm not sure but I believe the old E Learning has been scrapped and we are waiting on Zeiss to provide a new format with content. If you need help with this in the short term, send me an email and I will answer any questions you might have about this.

[[Da...]]

Please sign in to view this quote.

Thank you. Nothing critical, but I've send you pm anyway.

[[Aa...]]

Please sign in to view this quote.

Sorry, I missed that you asked this question. Actually, I scan both directions within the strategy for one feature. No recalling necessary. I guess you could set up two different features for point collection and recall their points (not the features themselves) into a third feature, but that seems like extra unnecessary steps.

[[Br...]]

Please sign in to view this quote.

I believe what Laura is talking about here is typically what I do. What a lot of people don't understand is what exactly the tolerance zone on a radius call-out looks like. Most CMMs just take the data given, and attempt to project it into a full 360 circle, then divide by 2 to give the radius, and get a center point based on the circle. The problem is the less arc distance you have and the the more form error you have, the more unstable your result will be. In Y14.5, the tolerance zone is actually closer to a profile tolerance than a size tolerance. The minimum and maximum radii create a tolerance zone that the actual surface must lie within. If its a fillet/corner radius, it must lie tangent, creating sort of a banana shaped tolerance zone.
If there are actual dimensions to the centerline of the radius, then the tolerance zone would be more open ended(Just two concentric radii.) This is what calypso can pretty much do.

Ok so once we understand that, now we can transfer this concept to Calypso. So lets say we have a radius we measured on an X/Y plane. The location of our radius will be in the X and Y coordinates of our base alignment. When we build the characteristic tolerance for this, click the constraints button, and check the boxes next to X and Y(You could also do this on the feature side). What this does is make the X and Y location of the Radius fixed instead of attempting to figure out the center point by projecting a full circle. You can use either the regular radius callout to do this, or you can use Calypso's "Radius Measurement" Characteristic. "Radius Measurement" will give you the polar radius of each individual point back to its center point, and if you have the constraints on, you are looking at where those points fall within a tolerance zone of two concentric circles. You could probably create the banana shaped tolerance zone with the curve form characteristic, but it might be alot of work.

[[Aa...]]

What you're describing is precisely what Jon taught us in the basic course. Constrain the size to find the location, constrain the location to find size. It's basically the old CMM operator's hack to get the software to give the numbers you want, and in most cases it's an easy to do just that. (Of course, when the size or distance tolerances are one-sided, additional tweaking might be needed.)

But I've always found this approach to be troubling. When you lock the center point of a radius and ask Calypso for a (Least-squares) size, what it's giving you is the average distance from that fixed point for each measured point. Ensuring that the average distance from the fixed point is not the same as ensuring that all the points fall within the banana-shaped tolerance zone, or any tolerance zone, for that matter.