Xact and legacy dimensioning

[chrisryn]

I know this has been discussed at length on this forum and I have read most of the post. I just wanted to share what I have learned about the two, whether I am right or wrong I wanted the opinion of others.

Just a little history I work for a global company that have many sites while I am the most experienced and knowledgeable programmer at my site I often find my self in a debate with another site that has many programmers. They use legacy exclusively where I tend to use Xact.

What I have found is that in most cases the issue in which scheme is used is understanding what Xact does and further understanding the intent of the design. The main issue I hear from other programmers is they don't trust Xact especially if the FCF doesn't constrain all 6 DOF's and the datum shift shows up on the reports. After extensive testing and back and for between Hexagon and GD&T experts I have come to two conclusions.

1. Either the designer didn't properly constrain all 6 DOF's due to lack of understanding of GD&T or that was the original intent.
2. if you use legacy and the intent was to not constrain all 6 DOF's you are not reporting the correct information. Since legacy is based off of the alignment which locks all 6 DOF's.

So what happens is a lengthy back and forth about who is right and if the part is good or not. There is another underlying issue based on which version of the PC-Dmis we are "allowed" to use by company standard. I really don't think in this case version is an issue. I post this because while most of the post regarding legacy vs exact are many years old I would like to get the opinions of the forum on this. If needed I can go into more detail.

TIA

[Schrocknroll]

You are correct, this has been largely discussed to death.

I would like to say that there are ways of emulating the lack of constraint on all 6 DOF, depending on the intent. For instance, if you have only a primary and secondary datum in the position FCF for a bolt circle you can use best fit rotate only instead of rotating to one hole.

[Ego Murphy]

2. if you use legacy and the intent was to not constrain all 6 DOF's you are not reporting the correct information. Since legacy is based off of the alignment which locks all 6 DOF's.

I spy a fundamental misunderstanding of how to correctly use Legacy Mode.

In real life, there is not such thing as not constraining all 6 DOFs. The very concept of "an unconstrained degree of freedom" in the field of coordinate metrology is a human imagination exercise of things to eliminate before proceeding with inspection, i.e. "clamp the dang part in a vice to keep it from rolling off the table."

If a Feature Control Frame's datum scheme doesn't control all 6 DOF, then this implicitly decrees that the features indicated by the callout are firmly responsible for controlling the remaining degrees not controlled by the datums - and right here is a huge disconnection between Legacy Mode and Xactmeasure Mode.

In Legacy Mode, the programmer creates a custom alignment to replicate the FCF. If the FCF leaves some degrees of freedom unconstrained, then the programmer takes that into account by locking those down by using the inspected features. The best example is making a 2D Best Fit alignment to the members of a bolt hole pattern before Legacy Position dimensioning them to their planar primary datum only. Note that in modern versions of PC-DMIS, during the creation of this alignment the programmer must make an informed and deliberate choice between different Best Fit types: Least Squared, Vector Least Squared, Max/Min, etc, as this greatly affects the outcome of the results. Note that after the Legacy dimensions are completed, the programmer should recall the previous working alignment as the custom alignment was only used for the dimensioning to satisfy that print callout.

In Xactmeasure, PC-DMIS creates an internal hidden alignment that locks all 6 DOF. If the datum scheme leaves some degrees open, then Xactmeasure makes it's pre-programmed decisions on how to use the inspected features to lock them down - with absolutely zero choice from the programmer in the matter. The temporary alignment is not visible to the user at all, we can only know it's existence with two clues: first, a tiny grey trihedron fleetingly glimpsed when the cursor is over the Xact dimension, and second the unusual (and often illogical) coordinates shown in the axial results. We are simply told to trust it, yet this becomes difficult when the results are displayed in a coordinate system that doesn't match anything we programmed and declines to show us how they arrived at it.

For many years Hexagon's official party line was that Xactmeasure was the correct way to do things. How's that working out now, when for v2020 (edited to correct year) Xactmeasure has been completely deleted and removed from the software and replaced by a whole new GD&T dimensioning system? One doesn't spend the enormous effort of mind-bogglingly complex coding to make such a drastic change to a product that's working perfectly, it's obvious that Xact's flaws couldn't be hidden by marketing flair any longer.

Just a little history I work for a global company that have many sites while I am the most experienced and knowledgeable programmer at my site I often find my self in a debate with another site that has many programmers. They use legacy exclusively where I tend to use Xact.

You should call a virtual conference among your programming coworkers and discuss this, with the goal of hammering out a company standard "best practices" that is broad enough to apply to all the product range. At least y'all are using the same software, it could be worse.

[vpt.se]

AFAIK, the new GD&T system is coming in a later version. The 2018/2019 XactMeasure change is new dialogs.

Also, I am fairly certain that if you want to follow the standards be it ASME or ISO, you will have to use XactMeasure. The current state of legacy is not performing according to standard, but can of course be used for whatever application you need it for.

[chrisryn]

In real life, there is not such thing as not constraining all 6 DOFs. The very concept of "an unconstrained degree of freedom" in the field of coordinate metrology is a human imagination exercise of things to eliminate before proceeding with inspection, i.e. "clamp the dang part in a vice to keep it from rolling off the table."

I agree with you only to the point of CMM programming you have to have all 6 DOF locked to properly inspect the part. The issue comes with following ANSI 14.5y standard as it allows datum deviation and is briefly explained in the 1994 revision as mimicking a hard gauge. They even elaborated more in the 2018 revision and actually have a symbol that allows for this translation to occur. Further for those who don't follow the standard they removed concentricity and symmetry as legal callouts in the 2018 revision just fyi. So that is why I spoke of design intent did the engineer/designer mean to allow this datum deviation to occur? Or did they not understand what they were doing not locking all 6 DOF's.

[Kp61dude!]

I spy a fundamental misunderstanding of how to correctly use Legacy Mode.

In real life, there is not such thing as not constraining all 6 DOFs. The very concept of "an unconstrained degree of freedom" in the field of coordinate metrology is a human imagination exercise of things to eliminate before proceeding with inspection, i.e. "clamp the dang part in a vice to keep it from rolling off the table."

If a Feature Control Frame's datum scheme doesn't control all 6 DOF, then this implicitly decrees that the features indicated by the callout are firmly responsible for controlling the remaining degrees not controlled by the datums - and right here is a huge disconnection between Legacy Mode and Xactmeasure Mode.

In Legacy Mode, the programmer creates a custom alignment to replicate the FCF. If the FCF leaves some degrees of freedom unconstrained, then the programmer takes that into account by locking those down by using the inspected features. The best example is making a 2D Best Fit alignment to the members of a bolt hole pattern before Legacy Position dimensioning them to their planar primary datum only. Note that in modern versions of PC-DMIS, during the creation of this alignment the programmer must make an informed and deliberate choice between different Best Fit types: Least Squared, Vector Least Squared, Max/Min, etc, as this greatly affects the outcome of the results. Note that after the Legacy dimensions are completed, the programmer should recall the previous working alignment as the custom alignment was only used for the dimensioning to satisfy that print callout.

In Xactmeasure, PC-DMIS creates an internal hidden alignment that locks all 6 DOF. If the datum scheme leaves some degrees open, then Xactmeasure makes it's pre-programmed decisions on how to use the inspected features to lock them down - with absolutely zero choice from the programmer in the matter. The temporary alignment is not visible to the user at all, we can only know it's existence with two clues: first, a tiny grey trihedron fleetingly glimpsed when the cursor is over the Xact dimension, and second the unusual (and often illogical) coordinates shown in the axial results. We are simply told to trust it, yet this becomes difficult when the results are displayed in a coordinate system that doesn't match anything we programmed and declines to show us how they arrived at it.

For many years Hexagon's official party line was that Xactmeasure was the correct way to do things. How's that working out now, when for v2019 Xactmeasure has been completely deleted and removed from the software and replaced by a whole new GD&T dimensioning system? One doesn't spend the enormous effort of mind-bogglingly complex coding to make such a drastic change to a product that's working perfectly, it's obvious that Xact's flaws couldn't be hidden by marketing flair any longer.




You should call a virtual conference among your programming coworkers and discuss this, with the goal of hammering out a company standard "best practices" that is broad enough to apply to all the product range. At least y'all are using the same software, it could be worse.

Mic DROPPED!

I simply didn't like the output of the report therefore I've stuck with legacy...

[NinjaBadger]

Nice explanation Ego.


The wider problem is there is a massive lack of knowledge and understanding around GD&T. Which is ironic, considering that when you DO properly understand it, it removes the ambiguity from traditional 'limit' style drawings.

And this lack of knowledge isn't specific to the shop floor, we've all seen drawings littered with nonsensical GD&T, and I've met dozens of inspectors who haven't got the first idea how to decode a control frame.

I think this is almost a self perpetuating problem, engineers (and those in manufacturing in general) are arrogant SOB's where any perceived lack of understanding about anything technical is so unpalatable to their ego's (sorry Ego!) that it verges on the unfathomable.

I have designers, engineers, inspectors and shop floor personnel look at me like I'm crazy (and flat out tell me I'm wrong) when I insist a Plane - Bore datum system is valid for concentricity (before everyone starts I'm in ISO land and I can assure you it is). Even when I show them an full example in the standard!

I butted heads with a contract programmer (whilst I was contracting also) who loved telling everyone that 'It should be called Best Fiddle, not Best Fit' simply because he didn't have a clue about GD&T.

We all know from the language people use when talking about it (Datum/Datum Feature/Basic/Boxed/Datum Reference Frame etc) whether they get it or they are a chancer.

[Matthew D. Hoedeman]

Nice explanation Ego.


The wider problem is there is a massive lack of knowledge and understanding around GD&T. Which is ironic, considering that when you DO properly understand it, it removes the ambiguity from traditional 'limit' style drawings.

And this lack of knowledge isn't specific to the shop floor, we've all seen drawings littered with nonsensical GD&T, and I've met dozens of inspectors who haven't got the first idea how to decode a control frame.

I think this is almost a self perpetuating problem, engineers (and those in manufacturing in general) are arrogant SOB's where any perceived lack of understanding about anything technical is so unpalatable to their ego's (sorry Ego!) that it verges on the unfathomable.

I have designers, engineers, inspectors and shop floor personnel look at me like I'm crazy (and flat out tell me I'm wrong) when I insist a Plane - Bore datum system is valid for concentricity (before everyone starts I'm in ISO land and I can assure you it is). Even when I show them an full example in the standard!

I butted heads with a contract programmer (whilst I was contracting also) who loved telling everyone that 'It should be called Best Fiddle, not Best Fit' simply because he didn't have a clue about GD&T.

We all know from the language people use when talking about it (Datum/Datum Feature/Basic/Boxed/Datum Reference Frame etc) whether they get it or they are a chancer.

"Well, that tolerance for that is 3-in-a-box"

I just want to biotch-slap people when they say that.

[dwade]

I'll put in my .02 on this and I don't know what type of business your company is involved with. Most of my programming experience has been in machine shops in one flavor or another. I have tinkered around with Xact enough to be very dangerous with it and I don't think any of the companies that I have worked for would pay for the training one needs to even be able to comprehend its advantages over Legacy dimensioning. With that said the more complicated I make my results for the average Tom, Dick and Jane machine operators they are going to be charging me with down time as they try and decipher MMC shift from datum features as the report will often show ZERO deviation from the basics, they will have no way understanding which direction they need to shift or offset tooling to make corrections.

[William Johnson]

I'll put in my .02 on this and I don't know what type of business your company is involved with. Most of my programming experience has been in machine shops in one flavor or another. I have tinkered around with Xact enough to be very dangerous with it and I don't think any of the companies that I have worked for would pay for the training one needs to even be able to comprehend its advantages over Legacy dimensioning. With that said the more complicated I make my results for the average Tom, Dick and Jane machine operators they are going to be charging me with down time as they try and decipher MMC shift from datum features as the report will often show ZERO deviation from the basics, they will have no way understanding which direction they need to shift or offset tooling to make corrections.

I agree with you on this. I had a job where I would output in Legacy for adjustments as Reference only then output in Xact for the accept/reject results. It was a pain in the Butt but it worked once they understood how datum shift worked. It also helped them to see how Legacy could/would reject good product.

[Paul Sarrach]

I spy a fundamental misunderstanding of how to correctly use Legacy Mode.

In real life, there is not such thing as not constraining all 6 DOFs. The very concept of "an unconstrained degree of freedom" in the field of coordinate metrology is a human imagination exercise of things to eliminate before proceeding with inspection, i.e. "clamp the dang part in a vice to keep it from rolling off the table."

If a Feature Control Frame's datum scheme doesn't control all 6 DOF, then this implicitly decrees that the features indicated by the callout are firmly responsible for controlling the remaining degrees not controlled by the datums - and right here is a huge disconnection between Legacy Mode and Xactmeasure Mode.

In Legacy Mode, the programmer creates a custom alignment to replicate the FCF. If the FCF leaves some degrees of freedom unconstrained, then the programmer takes that into account by locking those down by using the inspected features. The best example is making a 2D Best Fit alignment to the members of a bolt hole pattern before Legacy Position dimensioning them to their planar primary datum only. Note that in modern versions of PC-DMIS, during the creation of this alignment the programmer must make an informed and deliberate choice between different Best Fit types: Least Squared, Vector Least Squared, Max/Min, etc, as this greatly affects the outcome of the results. Note that after the Legacy dimensions are completed, the programmer should recall the previous working alignment as the custom alignment was only used for the dimensioning to satisfy that print callout.

In Xactmeasure, PC-DMIS creates an internal hidden alignment that locks all 6 DOF. If the datum scheme leaves some degrees open, then Xactmeasure makes it's pre-programmed decisions on how to use the inspected features to lock them down - with absolutely zero choice from the programmer in the matter. The temporary alignment is not visible to the user at all, we can only know it's existence with two clues: first, a tiny grey trihedron fleetingly glimpsed when the cursor is over the Xact dimension, and second the unusual (and often illogical) coordinates shown in the axial results. We are simply told to trust it, yet this becomes difficult when the results are displayed in a coordinate system that doesn't match anything we programmed and declines to show us how they arrived at it.

For many years Hexagon's official party line was that Xactmeasure was the correct way to do things. How's that working out now, when for v2019 Xactmeasure has been completely deleted and removed from the software and replaced by a whole new GD&T dimensioning system? One doesn't spend the enormous effort of mind-bogglingly complex coding to make such a drastic change to a product that's working perfectly, it's obvious that Xact's flaws couldn't be hidden by marketing flair any longer.




You should call a virtual conference among your programming coworkers and discuss this, with the goal of hammering out a company standard "best practices" that is broad enough to apply to all the product range. At least y'all are using the same software, it could be worse.

Ego, I upgraded to 2019. What do you mean it has been deleted, removed and replaced with a new system? Where is this info?

[vpt.se]

I think he refers to the new "look" of the XactMeasure dialogs. The new GD&T library won't probably be released for another year (see Wes' summary from Goonapalooza). AFAIK nothing has been removed from XactMeasure for 2019.

[Wes Cisco]

Hold my [-]beer[/-] coffee. . .

Firstly I wanna say IMNSHFO GD&T is about conveying the form, fit, and function requirements of a part. Parts that are functionally rotating shafts do not require GD&T with all 6 DOF to be constrained. While CMMs are the primary method of evaluating GD&T, especially in this neck of the woods, they are not the only method. It is entirely possible to fully define, manufacture, and inspect a rotary shaft without ever constraining the third rotational degree of freedom or using software much less a CMM machine machine. I have done it, more than once.

In v2019 R1 there is a new GD&T library - fresh AlGoreythms. The UI may have been gussied up a bit too, but it is still Xactmeasure. You are getting new math because the Standards have instituted new default math for datum definitions. See my summary elsewhere or a copy of ASME Y14.5-2018.

With v2020 R1 they plan to introduce the new GD&T command. It will replace Xactmeasure. It is similar. You will define datums features. You will build FCFs and select features to be evaluated by those FCFs from a list. The UI is different. You will have new and different options for choosing which math is used where. There will be much better feedback from the software when you try to do things outside of the methods prescribed by the standard you have chosen, (e.g. ASME or ISO). Rob remarked that he was trying to do metrology better and found he had to do drawing checking too. To the extent they are able, the error messages, pop-up warnings, and help files will explain why the incorrect GD&T attempted is incorrect, at times possibly even directing you to see certain portions of the standard(s).

Legacy was better than no GD&T command. It is capable of evaluating most of the less complex FCFs correctly if the person creating the routine knows their busyness. It is also quite easy to use it in ways not condoned by the standards and in most cases there are no warnings or errors triggered. And yes, much of the more complex GD&T is beyond legacy's capabilities.

Xactmeasure, while far from perfect, evolved into a better than legacy tool for most GD&T, especially the more complex stuff. Again, it requires a trained and conscientious individual to create the routines. Intuitive it is not.

I have had only one brief encounter with the new GD&T command. Based on that and the presentations I have summarized elsewhere, I believe the new command will be much better.

HTH and Humpy Hapday!

[AndersI]

You said it much better than I could have done! +1 on everything in your post.

[Ego Murphy]

Ego, I upgraded to 2019. What do you mean it has been deleted, removed and replaced with a new system? Where is this info?

As Wes posted, it's v2020 - I will edit my original post to reflect that.

[sealevel]

For me? Since 2010 or so since the launch of Xact, it has come a long way. My biggest puzzles were using the later 'width' construction and watching it go jasper on me (had to read deeper about the Xact trihedron and what IT demanded) and planar niceties.... as well as midlines/midplanes taught me to check upstream in the program...those components had some VERY small vector anomalies. Xact was watching.
I'd get the graphic for the TP for planar callouts getting squishy at times, but was activeX/label being pissed, not the blame of Xact.
IMHO? Can't wait for the 2020 GD&T.
New PC-Demon models on the show room floor usually gets bloated with new bells n' whistles that do nuthin' mostly, but have upmost belief the Product Owner of the GD&T will not release just 5 new shades of pig lipstick, but a real improved product.

[AndersI]

Xact was introduced in 4.0, I believe...

[sealevel]

AndersI, I will take your comment as gospel. I was thrown in the pool @ 4.2MR1 @ boot camp. After 1 week? I wasn't feeling frisky to take on Xact as a Demon rookie.....

[vpt.se]

New PC-Demon models on the show room floor usually gets bloated with new bells n' whistles that do nuthin' mostly, but have upmost belief the Product Owner of the GD&T will not release just 5 new shades of pig lipstick, but a real improved product.

+1