Parallelism and the new GD&T enigne

[[Ry...]]

Hello, I am looking for some clarification on what is different between the new GD&T engine parallelism and the old method. I have a part that is checking parallel of two flanges on a round part that need to be 0.002” parallel. When I check it using the old parallelism characteristic, I am getting .0039” ( Using LSQ for the evaluation methods and the standard filter and outlier settings from the cookbook for planes of that size.)

 

I wanted to compare with the new gd&t characteristic I am getting .0041” with the evaluation methods set to suggestion. If I set the evaluations to Gaussian for both the datum and the feature checked the error drops down to .0007”

 

My first question is what is calypso suggesting for the evaluation method? Secondly why is Gaussian showing so much better than LSQ I was under the impression that do to how Gaussian worked it was much more susceptible to outliers meaning it normally would read a larger value not a better one? Also why does the new GD&T engine not list LSQ as an option for evaluation?

 

Thank you for all the input

 

I am using Calypso 2024 with service pack 1 on and ACCURA 2 with a VAST XTR

[[Jo...]]

Hi,

i got no clue about Calypso, but least square method is by Gauss.

 

[[To...]]

In the characteristic window, near the bottom, there is a light bulb icon. Click it and it will tell you how things are being evaluated.  Unlike how it's done in the Classic GD&T, in the evaluation window.

Normally, parallelism requires a datum feature which is traditionally Outer Tangential.  In my way of thinking, I would either create 2 characteristics, swapping the roles of the considered features and datum features.  But the more I think about it, maybe there is a way to evaluate both planes as LSQ by using the "loose" option, which I am not able to explain how. 

Please sign in to view this username.

is a big fan of this option.

Edited 9 hours ago

[[Ri...]]

The Suggestion will follow whatever standard you have set. 

In 2025+, the ASME standard for a Plane is the Guassian Plane - Shifted Outside. I believe for ISO it should be a Minimum Zone Plane - Shift Outside. 

In the Loose, it will follow whatever settings you it set to in the Feature side. 

You can always change the evaluation, regardless if it is abiding by a standard or not. 

[[Ry...]]

Please sign in to view this quote.

Richard, Thank you for input do you know of any documentation in asme that says we should default to a Gaussian Plane.

[[Ri...]]

It can be found on Page 18 of ASME Y14.5.1-2019.

4.7.11 Alternate Stabilization Procedures

   In accordance with ASME Y14.5-2009, if irregularities on a datum feature are such that the part is unstable when brought into contact with the corresponding datum feature simulator, the default stabilization procedure is per the candidate datum set as outlined in this Standard. ASME Y14-.5-2009 does allow for different stabilization procedures to be specified. When a single solution that minimizes the separation between the feature and the simulator is specified the default procedure is a constrained L2 for datum features of size and the constrained L2 applied to the external envelope for planar datum features. 

 

While it does not explicitly say to use the L2 (which is just the Gaussian shifted outside), it gives you the option to for stability, so I believe most softwares are just defaulting to that. I'm onboard with it either way because the OTE just isn't robust enough. 

[[Cl...]]

Please sign in to view this quote.

Gaussian and least squares in metrology

In the context of metrology, Gaussian refers to the mathematical representation of the Gaussian function, which is used to model the behavior of many natural phenomena and industrial processes. Least squares is a statistical method used to estimate the parameters of a model, such as the Gaussian function, from observed data points. While they are not the same, they are often used together in metrology to fit observed data and model the behavior of processes. The least squares method is particularly useful for estimating the parameters of a Gaussian function, which can be applied to various metrological applications, including fitting observed data points and modeling the behavior of processes. 

 

ZEISS Group