Building feature Strategies and Theory

[[Ti...]]

I put a lot of hours into the software over the last week and have some general questions on strategies.

First, it appears you need to do a prealignment so all the features are 'near' each other. At this point, you can make an Auto Cylinder and then just right click it and do a fitting strategy to create the real feature.

My question is when you do an Auto feature and you click the CAD model it is perfect geometry so the selection boundaries are perfect. Then you do a fitting element and even if you push the boundaries back by 2mm you still have a slightly mis-aligned mesh that is being told by the software to build out a feature. I'm wondering if the pre-alignment is sufficient to built the actual element correctly or would it be better to do a local alignment to just that feature so the CAD and mesh are as close to perfect as they can be before you create the auto and actual feature?  Then after you are done you build out your datums and measure GDT correctly.

So one strategy would be to just prealign and then build all your features. The other strategy would be to prealign, local best fit to each individual feature, then build each individual feature.

Thanks for any input.

[[Na...]]

Hello Tim,

in the properties of an actual element -> general, there is a checkbox with drop down called "alignment required". If the prealignment is not sufficient, because of high local deviations on your part e.g. a local best-fit for this element may be created. This alignment is the "alignment required" for the element. The checkbox is ticked in the background for elements of the measuring principle fitting element with the active alignment you are creating it in. This topic is also part of the inspection training, which is highly recommended if you are using the software.

Regards,
Nanno

[[Ti...]]

Please sign in to view this quote.

Are you talking about the free starter training or is there additional training somewhere. I finished the starter training.

[[Ja...]]

Tim , there are distributors of GOM products worldwide who offer face to face training.. I would recommend this is the best way to grasp the base mechanics of the software.    To your original point , there is no fixed ' best' strategy , it might work out global best fit ( prealingment can contain this) , or needing multiple local alignments to help find the features.  I would have in mind that if you need the local fit strategies the part is likely heavily distorted.  So for example if the part is warped such that a plane datum  curved then things will likely be out of tolerance related to that datum   Localised fits and colourmaps can help understand the 'why' in my experience. 

[[Ju...]]

Please sign in to view this quote.

Hello Tim,

I believe that Nanno is referencing the starter training for GOM Inspect Pro that you can follow on the GOM Training Center.

You can purchase this eLearning from your local distributor to increase your knowledge of the software.

This is an 18 hour training.

But as James is saying, if you are struggling with elemental concepts and with the basic usage of the software, some personal time with an experienced AE would be helpful.

[[Ti...]]

Juan,

I think I see that here:

https://www.gom.com/index.php?id=1474&L=16&tx_gomtraining_pi1[training]=210&tx_gomtraining_pi1[action]=show

But when I go into the training center and login I cannot find that class specifically. I'll poke around some more. I feel I have some solid fundamentals down now, but I'm slow at building strategies for simple tasks. I have deep knowledge in several major CAD and CAM programs, but some measurements in CAD that would take me seconds seem much longer on GOM. I know this is due to it being real data and not perfect data, but some things seem cumbersome.

For instance here a fellow user asked a question I had trouble with...how to get a measurement from the closest point of a cylinder to a plane:

https://forum.gom.com/topic/364-how-to-measure-to-edge-of-cylindercircle/

In any CAD program you would highlight the 2 features and there is a min/max distance from both(along axis as well).  The solution in that thread seemed much longer than it needs to be. I'm not sure I could have figured that out without reading that thread to be honest. I went on to successfully do this using a center of a sphere as well using the sphere pt and nominal direction as the plane...then the rest was the same. It just adds 6 elements in the nominal tree and then 6 corresponding elements in the actual tree. The tree organization gets gets crazy large for such a simple task. It took me a while to understand why we are creating fitting elements, I still feel this should be done at the moment you create the nominal element. I feel right clicking and creating the fitting element is cumbersome. Regardless, I'm enjoying the software and its capabilities, very powerful watching it analyze these CT scan meshes. The parts are going under much more scrutiny than on a CMM no doubt.

Thanks again as always.

 

[[Na...]]

Hello Tim,

the training should be available under this link. It is also possible to attend a classroom training at gom or your local distributor:
https://training.gom.com/home/LearningPath/7199?r=False&ts=637932128340599373

Please sign in to view this quote.

There might be some other solutions to that problem with less clicks, but it won't get much shorter. The main difference to a CAD program is, that we are dealing with actual data. With a measuring principle you have to decide how to derive a geometry from your actual point cloud e.g. a max. circumscribed element vs. a Gaussian element. Sometimes a measuring principle is more complex than that, like the project offset section. A simpler (= less clicks) approach to the thread you linked would be the GD&T quick creation. Here you could create a position check, click on the cylinder and then on the datum system, which is the plane. The measuring principles will be defined automatically according to the ISO or ASME norm. But in every other inspection you have to decide on how to derive the values you are interested in. This is covered by a chapter in the training by the way 😉 

Regards,
Nanno

[[Ti...]]

Please sign in to view this username.

Thanks for the link, I can see that class....it is 620 the one I took is 780. Regardless, I'm happy you shared the link with me b/c I will take the class tonight.

I'll have to investigate your idea of using a positional check. On the surface, I don't think a positional check works for getting the minimum distance b/w two surfaces. I understand you have to create a 'real' measurement principle.  But, it just seems like you should be able to say 1) make a cylinder 2) apply guassian 3) make a plane 4) apply guassian 5) GET ME the smallest distance b/w the two features.

Instead, you have to create the strategy in the above link I provides of making lines and points. I will have a look at your suggestion, but below is the dimension I'm looking for exactly to be clear. I appreciate your help with my early questions, thanks so much.

 

 

 

 

[[Ja...]]

Tim, it is a bit more advanced i guess but one opportunity you have in the case in your image is a touch edge point to the rearside of the mesh.  If you create a surface point, on the apex of the depression and choose an inverse surface normal you can have a touch edge loint touch the lowest point on the depression.   You can then have a projected point distance from this tothe plane.    You can have multiple of these perhaps and group them to understand the feature behaviour.

The general issue you face  issue is best summarised for me by thinking about measuring between two planes...there are infinite ways to do this if you consider that  planes in real life will never be parrallel. 

 

[[Be...]]

Another simple possibility would be the Deviation to geometry and a minimum picker. Create the fitting plane and calculate the deviation of the cylindrical area of the mesh to the plane. Or you can calculate the Material thickness on this area which yields the deviations to the nearest triangle on the other side. 
But as already mentioned: The different strategies will give you different results and it depends strongly which value you like to evaluate. This depends which functionality you like to check (and therefore you must make this decision and the software can only offer different strategies).

Regards

[[Ti...]]

Please sign in to view this quote.

Nanno,

I think that training is gone. If you click the link and then try to take the class there is an error message. Maybe it was replaced with the class I complete called 780 starter:

https://training.gom.com/home/LearningPath/7265?r=False&ts=637935281697585734

[[Ti...]]

Please sign in to view this quote.

Where is the minimum picker?

[[Be...]]

Inspection -> Min/Max deviation label

or in the I-Inspect if your deviation object is selected in the explorer