DRF Interpretation

[[Br...]]

Please sign in to view this quote.

First of all, Why? You could just use the proper algorithm and skip this extra step. Second, It really depends on what algorithm you used to create the offset plane. Neither LSQ or Minimum Zone just simply splits the form error in half and sticks the plane there either. They are much more complicated algorithm's than that. for instance LSQ takes all the data points and iterates a perfect plane between them so that the distance of each data point squared and added together equals the smallest number possible. So not only does it not locate it smack dab in the middle of the form error number, it wont be oriented in the same way as a datum simulator contacts the highest points.

The most accurate way Calypso provides to best represent the Datum Simulators within their respective datum reference frames is to use outer tangential and the appropriate constraints for each datum feature and their respective precedence within the Datum Reference Frame.

[[Me...]]

Actually, the publication, https://shop.metrology.zeiss.com/INTERSHOP/web/WFS/IMT-US-Site/en_US/-/USD/ViewProduct-Start?SKU=600060-7005-002&CategoryName=570000&CatalogID=500000/, available from Zeiss, describes the ‘Outer Tangential Plane’ as fitting the measured points using Chebyshev association followed by an offset of one half the form error along the computed surface normal.

While not commonly used, this could be considered a reasonable approach when a datum feature contains multiple candidate datum planes (consider a nominally flat surface produced as V-shaped) and one seeks to “balance” the surface with the currently available tools. There are a number of published papers available on this topic, but until the standards committees agree upon a new mathematical definition for establishing datums, there will be differences of opinions on how to best approximate a DRF to the current standard.

New Mathematical Definition of Datums.pdf

[[Br...]]

Please sign in to view this quote.

I'm assuming this is an older version of the current measurement strategies cookbook. I don't see anywhere in the more recent cookbook that describes how the different fitting algorithms work. Though I do know at least that with circles and cylinders OT is exactly the same as Max Inscribed, and Min Circumbscribed, which are considered Chebyshev algorithms. Maybe you can help point me to where you read this.

One way you can actually test this though, is to simply take a plane measured with some real data and look at its A1 and A2 values. Now switch between the different fitting methods. I think what you will find is that those values change between Outer Tangential and Minimum Zone. If Outer tangential was simply Minimum zone with location along its vector adjusted by half the form error, the A1, and A2 values should remain constant. We only should witness changes in the X Y and Z values.

Please sign in to view this quote.

I completely understand your concern with regard to the issue of unstable or "rocking" datum features. In this case I would agree with you that a more stable algorithm should be used. Dane's suggestion might work pretty well in this specific case. I think the latest version of Y14.5.1 actually endorses the use of the Constrained Least Squares algorithm from the publication you put on here. It would be great to see this particular algorithm get implemented in Calypso. But as far as most surfaces go since you don't have this algorithm at your disposal, I would stick to OT for the majority of my evaluations. According to the Calypso manual you have to use Outer Tangential in order for the ISO 5459 to be useful.

[[Aa...]]

This perplexes me. I'm at a loss as to why Y14.5.1 needed to deviate from the "candidate datum" prescriptions in the 1994 edition. Datum instability can be a royal headache, but usually this kind of instability points to a deficiency in the controls applied to the datum features themselves. (E.g., a "rocking" datum plane that causes significant instability in the results of FCFs referencing said datum plane could be avoided with a restrictive enough flatness control.)

[[Me...]]

Please sign in to view this quote.

Yes, Calypso determines the tangential plane different from what is described in their other material, and I would assume Zeiss had reasons for doing so. Here is a better link to the books offered by Zeiss.>>>https://world.probes.zeiss.com/en/Training-Material/Books/category-1012.html/

Please sign in to view this quote.

The statement from the user manual leads one to believe that if “ISO 5459” setting is applied, then all datum features are automatically set to outer tangential fitting and appropriately constrained (if setup to do so). I will contend that this is simply not the case. Often you will find that the entered datum references are ignored and the base alignment is used without any indication. Other times, the datum references are used but revert to either unconstrained LSQ fit or unconstrained OT fit.

If you have time, I would encourage you to establish the DRF shown in Fig. 4-15 of the Y14.5-2009, and simulate inspecting the hole pattern. Be sure to use a "Symmetry Plane" for the tertiary datum since the constraint portion of the ISO 5459 setting only applies to features that contain measured points.

Once the simulated run is completed, look at the default printout. You’ll see that the symmetry plane reverted to an unconstrained LSQ fit and oriented the datum planes as if a translation modifier had been applied. All of this is happening even though the datum references in the position characteristic indicate otherwise.

[[Br...]]

I apologize for the late response. I was on vacation with my family for the past week and didn't get on this forum.

Please sign in to view this quote.

Ok I see, It's something completely separate from the Cookbook. I'm quite skeptical to your claim though that Zeiss would would completely contradict themselves on these topics. If this is true that zeiss makes two contradictory claims, one of the claims have to be false. Though I cannot obtain a copy of this book right not to see where you have read this, I guess I will have to take your word for it.

Please sign in to view this quote.

It simply doesn't say that. It says Outer Tangential is what was intended to be used with the ISO 5459 setting. They are informing the user what evaluation method should be set for the ISO 5459 setting to be useful since the user has the freedom to do whatever they want with the feature evaluations. There's a setting the user can set to make all datum features automatically Outer Tangential as a default even. The user has to set this though.

Please sign in to view this quote.

If the datum features are being ignored then you should use a secondary alignment. This is a bit of a red herring. We're getting off the topic now about best practices as far as the proper algorithms go into quirks in calypso. I've certainly noticed odd things happening in the Characteristic DRF myself. Especially when its not fully filled out. It's by no means flawless. This is why its imperative to have a good grasp on the standard you are working to in order to make sure the calculations being made best represent the requirements. You have a lot more control with the secondary alignments.

Please sign in to view this quote.

I'd have to see a specific example. Sometimes the constraints make sense because a line or a circle feature, due to their 2D nature, cannot be constrained to the higher precedented datum feature. If this is a concern, it's up to the programmer to make sure it is measured as a 3D feature instead. The LSQ thing could be because its a constructed feature, which means there's no physical measured data points within it to even run any algorithm on. So even though it says "LSQ" It does't actually do a Least Squared algorithm on the data since it can't.

Please sign in to view this quote.

Its a great point, and another example as to why a secondary alignment would be much more useful in this particular application. Calypso doesn't handle non cylindrical features of size very well (In my opinion). So you have to work with what you have. There's a lot of things in the standard that Calypso wont be able to perfectly satisfy. But as a general principle, in most cases, OT for datum features will be your best bet. Obviously things like this, or a complex datum feature measured as a 3D curve is going to be far more complicated than that and will require some extra work.

[[Me...]]

Must be the time of year for unplugging and taking a vacation!!

I don’t feel that Zeiss is contradicting themselves. They appear to be taking exception to what the ISO 5459 standard states regarding the default association criteria for planar features;

A.2.1 - “The associated features, used to establish the datums or datum systems, simulate contact with the real integral features in a way that ensures that the associated feature is outside the material of the non-ideal feature. When the result of this process is not unique, then the associated feature to be used is the one that minimizes the maximum distance normal to the associated feature between the associated feature and the filtered feature representing the real feature.”

Section A.2.3.2 further describes the use of the “Minmax” objective function with the additional requirement that the associated feature lie tangent outside of the material.

On the other hand, Calypso calculates the outer tangential plane / line by finding the outer most three or two points. Instances where multiple tangential planes exist (not unique), Calypso picks the tangential element that has the shortest distance to the focal point of all points. Both methods will produce a tangential plane / line, but with a different orientation.

The constraint portion within Calypso's ISO 5459 setting (when it works) is doing nothing more that projecting the measured points onto the higher precedent datums. For instance, fitting a cylinder to be constrained to a given plane is equivalent to a 2D circle fit with the points projected into the plane.

Calypso’s “quirks” are relevant to the discussion since best practices can only be determined once an understanding of what the software is doing has been realized. I’m not advocating against using tangential fitting within Calypso, but only trying to highlight potential differences. This must be an interesting topic based off the view count this thread has generated. 😃

[[Da...]]

Please sign in to view this quote.

Holly $hi7, I think we have a part where "something strange" is happening with clocking.
And I see some connections to your explanation/s Andreas. Thank you again! And others.
I'll try to do make some kind of testbed and diggest what is happening in next weeks and report back.