Diameter of a hole, AME, UAME, theory vs reality and other nonsense 1

[sendithard]

Been digging in the weeds of the Y14.5 standard a bit. The more I read, the more I learn, but it feels like I get dumber sometimes.

My most profound question is what is the diameter of a thru hole and a blind hole? Is it the largest pin that will slide thru and largest pin that will go to depth?
---I believe my above statement is correct and that is the UAME(unrelated actual mating envelope). When I was reading that nonsense it just doesn't make sense. The actual mating envelope is the pin, and the unrelated envelope is the theoretical. Seems backwards to me.

Then Y14.5 hits us with 'The actual local size of an individual feature at each cross section shall be within the specified tolerance of size." (pg 18 2009 standard)
---So in theory, you could have a pin that fits within the tolerance of size, but due to form issues(pitting, tool marks) the part could fail if you sectioned the part and measured it. Not to mention you have to measure it perpendicular if the hole had a curve to it show in Fig 2-6 2009.

 

[Burunduk]

sendirhard,
A hole on a real part doesn't have a perfect form, but we need to do some stuff with it for which we wish it was nice and straight and uniform. Stuff such as measuring the diameter, extracting an axis, etc. But the reality is what it is, so what do we do? The standard gives us things like the "unrelated actual mating envelope" and "actual local size".

So how large/small an imperfect hole is?
A hole is just as small as the diameter of its unrelated AME, which is as you mentioned the largest gage pin that will go through, and it's just as large as its maximum distance across opposed points - which is what you see mentioned as the "actual local size", and what you get from a measurement by a caliper or micrometer.

Does this help you to make sense of these concepts?

 

[Belanger]

I know that you're dealing with the ASME standard, but it's worth knowing that the ISO system allows specific modifiers for designating the algorithm to determine a feature's size. For example, you might see an oval with the letters GG inside -- that means least-squares. Or there is GX, which is the maximum inscribed, or GN is the minimum inscribed. Those are all detailed in ISO 14405-1.

John-Paul Belanger
Certified Sr. GD&T Professional
Geometric Learning Systems

 

[sendithard]

Yep all that helps out, thanks. I've been learning this software GOM inspect, to analyze the CT scans and I've had to decipher some of the ISO language, due to it being a euro company. I'm slowly grasping some of the terms. One of the areas that bothers me at the moment, is it measures planes either by least sqr or by ISO 5459 which is external min/max. All the pictures I see are all 2d images showing one or two touch points then it is a min/max for the other points. This makes no sense to me b/c they are leaving out the 3d other areas of the part that the part regardless how rocker it is it will find a 3rd point. From there, I don't understand how ISO 5459 default min/max plane is any different than the new default ASME L2 constrained least sq.

So technically is sounds like per Y14.5 both sizes must technially meet the limit of size of the hole. Both the largest pin size that can go thru and then each cross section point must fall within this size. Seems strange to me to go to that extend b/c without a ct or cmm you almost need to cut the part in half to measure the hole cross section.

Thanks.

 

[Burunduk]

sendithard,
I'm pretty sure GOM Inspect has ASME compatibility option. When doing alignments according to the ASME datum concepts, the software should be simulating tangent planes, not making least square approximations of planes. Did you try adjusting the default options?

Seems strange to me to go to that extend b/c without a ct or cmm you almost need to cut the part in half to measure the hole cross section.

Why? What is wrong with using gage pins for the minimum limit of size and rule #1 evaluation, and performing a series of caliper or internal micrometer measurements for evaluation of conformance to the Max limit of size? I agree that a CT or a CMM scan may give you reduced measurement error, but that still doesn't mean that the manual methods are no good...

 

[sendithard]

Burunduk,

GOM will have ASME plane choices in the next version(it does have max inscrib/min circum tho). As of now you have two options, you can align via what they call 'geometry'(Guassian best fit or Chebyshev) or if you choose 'gdt' it uses only iso 5459. I tore thru PCDMIS manual for images on both(5459 and the new l2 constrained lst sqr) and honestly they only give 2d images and I don't see how they are much different, both are external(surface plate/tangent as you said I believe) and if there is rocker they both equalize in some way. ISO seems to only require a single high point then equalizes the lows. I just don't understand how you can create a plane that will in fact touch 3 points on a surface plate, but then say no thanks I'd rather just touch 1 or 2 highs and equalize.

I welcome the gage pin method, in fact prefer it b/c it is real. What I was saying I thought was a little over the top was the requirement that each cross section need to be withing limit of size. Just like run-out I'd say you check enough cross sections to be satisfied so you don't need to spend all day, but it just adds to the complexity of a simple positional fcf. After seeing the hit points from the CMM vs the massive amount of data points on the CT any bad artifact can throw out a tolerance when more and more layers of requirements are placed on an callout. If the pin slides in within tolerance but there is a few micron groove/gouge inside the hole throwing the local size out, I think it should pass a simple locating pin, but for a high pressure valve different story. The local size deal just seems overkill to me, but I'm still quite green to mechanical and these tight tolerances.

 

[Burunduk]

sendithard,
I wouldn't say the requirement that size limits apply at all cross sections is over the top. In fact, it is like any other tolerance - you mentioned runout. Circular runout applies to each cross section separately and to all cross sections along the shaft, but as you noted it is never checked at all cross sections - surely not when the classic FIM method is used. I doubt that even GOM gets close to analyzing the entire surface when it measures circular elements. But that shouldn't be a big issue, since ideally you should only check just enough samples to be confident the feature is good, knowing your manufacturing process capability. Then if you're good with that, you shouldn't expect an inspection method that collects a great amount of data, such as the Gom image processing measurement method, to increase scrap relative to more traditional metrology such as CMM with its limited sampling or even manual inspection when the measurement error is reasonably small. This all applies to limits of size, circular runout, and many other requirements...

Regarding the GOM Inspect simulation of datum planes - you may be right. I'm not an inspection guy but I have watched some of this company's tutorial videos on alignment and it seems they have multiple alignment methods between the nominal CAD geometry and the actual scanned geometry, including pre-alignment, various best-fits, some method called 3-2-1 when you randomly click on points on the surfaces, an axis system based thing, and among all this may be one method which could actually allow to properly align to datums and establish a datum reference frame - and I wasn't able to figure from the explanation whether it may conform to the ASME conventions.
But I do recall someone showing me that at some initial stage of working on a scan you have the option to choose between ASME and ISO... on the other hand I also recall being showed those "Geometry" & "GDT" alignment choices too which I don't know what to think of... I really don't know about this measurement application as much as I would like to.

 

[sendithard]

Appreciate the discussion as always. It's great to be able to discuss this with people with so much experience.

GOM has all of what you stated. Basically for planes right now you can only create them with the Chebyshev(internal, middle, or external) or Guassian(least squares). Seems this is all ISO stuff. Then when you create a datum structure for a profile of a surface, say A|B|C being all planes, you can select Geometry or GDT. If you choose Geometry the datums retain the math method you choose above like Guassian. But if you choose GDT it recalculates the planes surface using ISO 5459, which is quite similar to current ASME methodology now. It is external surface plate, tangent type method. So it is a bit confusing b/c you can't create the feature using GDT external/surface plate/tangent type plane, but when you go to align the CAD and Scan or just do a GDT measurement with datums you CAN force the features to re-calculate to iso 5459 plane.

There is a Y14.5 vs ISO 1101 dropdown you can use with GDT measurements. Toggling b/w both does not change Profile of a surface measured values as Y14.5 now doubles the worst when it added min/max a while back. But, going from iso to Y14.5 dropdown on position can yield crazy diff results b/c ISO uses guassian/best fit on the hole when Y14.5 uses max inscribed(gage pin) method. Took me a while of making my own cad models with bad geometry to figure that out.

321 just lets you orient a miscellaneous mesh file without cad to XYZ. You hit 3 points to create Z plane, then two points creates the 90 deg Y Plane, and one point creates the 90 deg X plane. Just like surface plate, slide to an angle block, slide the the next angle block for simulated datums.

 

[Burunduk]

I appreciate the discussion and the exchange of knowledge as well. You seem to have investigated Gom Inspect deeply and that's great.

So do I understand correctly that the program has both ASME Y14.5 and ISO 1101 supported measurement options for tolerance requirements but the alignment/datum simulation is always done the same - either "Geometry" which is Gaussian or "GDT" which conforms to both ASME and ISO?

Do you say that the 3-2-1 alignment method acts similarly to using a surface plate and angle blocks? What I understood from the explanations I've seen, is when using the 3-2-1 alignment method you randomly click 3 points on a scanned surface, then 2 points on another one, and then one point on the third one. But what I'm thinking is - since these points are selected randomly and are arbitrary they are not necessarily the high points on which the part would be supported by the surface plate and angle blocks, so the alignment doesn't quite mimic physical datum simulation. Am I wrong?

What about datum targets? Can you align to them? And how does that relate to using the nominal CAD model to compare the actual scan data to?

 

[sendithard]

The best way to explain is a real example…..

Regarding your first issue with 321…I previously stated that 321 is only for aligning a scanned mesh to a XYZ datum system. I only do this for quick and dirty checks….imagine a part so rotated in space you have no reference planes. I quickly hit 3 planer points on the scan, 2 line points preferably sharing a planer area, and 1 touch point and BOOM my scanned part is squared up. GOM’s 321 does NOT find the highs as you rightfully questioned. You can also align CAD and scan this way, but that would be a poor choice.

I can answer your other questions, but give me a day b/c they way GOM works is hard to explain.

 

[Burunduk]

Thanks sendithard, now I understand the idea of 3-2-1 better.

 

[sendithard]

For all the asme/iso non-sense in GOM it's best to mentally separate the idea of alignments, plane creation, and hole(cylinder) creation for GDT measurements. So if you want the min/max height measurement from a surface plate type/tangent setup you create a plane for the bottom surface. In this scenario, you can only make that plane guassian or chebyshev. I'll argue a chebyshev outside selection is rather similar to ASME tho, so technically you can't make a quick plane ASME style, but in reality you may only be in the tenths of difference. Now if you had a profile of a surface callout to ABC, while you previously made plane ABC all guassian(least squares) you can retain that guassian by means of selecting 'geometry' in the datum creation setup. If you want to go full tilt GDT you just select 'GDT' during datum creation. As of today, GDT creates planes based on ISO 5459 which I think is chebyshev outside(almost same as ASME now). SO here is where it gets interesting....say you have a hole(cylinder) as a datum and you created it using max inscribed based on Y14.5. Now you need to make a GDT datum structure from said hole....welp, if you choose GDT it uses the ISO standard which is guassian. It will not create a hole datum using the ASME datum hole standard which is maximum inscribed(gage pin method which is ASME). In this situation, you are at a fork in the road....because you can only use 'geometry' or 'GDT' as your datum creation....so if you want to retain that max inscribed hole datum, you have to use 'geometry', but then your planes are forced into 'geometry' which if you made them guassian you have an issue. So, right now if I were to make the best ASME setup I could, I would use chebyshev outside as my planes and max inscribed as my holes then select 'geometry' as my datum setup method. As we proceed away from datum creation you get the only true Y14.5 dropdown...this is when you create a feature to be measured in any GDT measurement. If when you made the hole feature you initially used guassian, b/c it is your go to baseline strategy. Now when you go to measure the hole for position in GDT it allows you to measure the hole based on ISO or Y14.5. It's basically saying, I don't care how you creates this hole before, but I'm allowing you to measure it based on ISO(guassian) or Y14.5(gage pin method). I'm probably the only person that is depraved enough to actually make a out of tolerance CAD models to simulate these differences. Hope that makes sense. I'm told next version Y14.5 dropdowns are more abundant.

 

[Burunduk]

sendithard, thanks for the detailed explanation. It seems like working according to ASME Y14.5 in the Gom Inspect software is not as trivial as just using a drop-down menu to choose the applicable standard. I am a mechanical designer and sometimes the parts I'm working on get inspected by Gom. When I need to address issues regarding the measurements and questions arise whether the measurements match the design intent, it is always troublesome to address those issues because of the lack of easy one-to-one connection between the theoretical basis of dimensioning and tolerancing (ASME Y14.5 in my case) and the inspection software operation procedures. Your explanation helps, but I still have some gaps in understanding. Here are the points what I'm still not clear about:

1. You mentioned "creating" planes and holes by Chebyshev or Gaussian and later using them in a "datum structure". I assume that those planes and holes are created as simulated datums (in case of the planes) or as "theoretical datum feature simulators" / "true geometric counterparts" in case of the hole - is that right? In that case aren't they created from the design CAD model? If they are created from the design CAD, what are Chebyshev and Gaussian for at this step? These are conversions from imperfect geometry to perfect elements, but the model is already perfect. Or do you mean that these are used when the nominal CAD model is not available and the elements are created from the measurement data?

2. If the "theoretical datum feature simulators" / "true geometric counterparts" are created from the measurement data (the actual part scan) then aren't "creation" and "alignment" a simultaneous, inseparable process? For example, suppose that datum references A,B,C are in a feature control frame. The primary datum feature A is a planar surface, the secondary datum feature B is a hole, and the tertiary C is a slot. Then if Chebyshev is similar to ASME planes, the primary theoretical datum feature simulator / simulated datum is a plane that contacts the 3 highest points on datum feature A, so it is already "aligned" to the scan per its definition. The secondary datum feature simulator is a cylinder which is perpendicular to simulated datum A and is inscribed by the scanned hole - which means it has to contact some points on the measured hole, and then the simulator for the slot datum feature is at some basic orientation to datum feature simulators A and B and at some basic location from B, but it also should contact the high points of the measured slot. So aren't those elements already aligned to the measured part? Why a subsequent alignment step, at which "GDT" and "Geometry" may drive different outcomes, is needed?

 

[sendithard]

I'll try to clarify your issues, but give me a day to consider them, I didn't know you were dealing with gom in reality. I'll try to help. I will say this now tho.....

Long story short, get a slt file back from your supplier that has already been 'pre-aligned' to the cad. This way you can put it up against your model and see if the
scan has areas around the datums that are much shorter than the sharped edge corners of those perfect CAD files. Also you will see issues with features scanned.

I'm tring to convince management to fillet all datum or ctitical features on the STEP file b/c a CT scan simply cannot produce a sharp edge, nor can a tool do this.

I know if you sent a cam guy a part with 10 fillets he would shit himself saying hey buddy just chamfer these edges and that is the norm...so just send them a sharp cornered
step file and let them break the edge....but if you get a scan on the same part I think the STEP file needs a lot of filleted edgs and most importantly on Datum edges.

 

[sendithard]

Brun,

You can also make a bad CAD file that deviates from the nominals and throw it into GOM them go into Operations > CAD > CAD to ACTUAL MESH.

Then you export that mesh, which is an stl file....this is how I stressed tested the software for position in ASME vs ISO. I wanted to see
why each delivered a difference answer. Takes a lot of time, but help understanding the software.

 

[Burunduk]

I'm not working on the Gom software myself but as the design guy that supplies the drawing and the nominal CAD I sometimes get to address issues raised by the inspection guys and that's when I try to understand whether the measurement was done correctly - with the right alignment, etc.

The sharp corners issue - the way I have seen it handled is by shortening the planes that are later used as datums so that they are a bit short of reaching the corners where the CAD surface ends. Could this resolve the need for fillets?

Thanks!

 

[sendithard]

Burunduk,

You can 'decrease selection at borders' but as you know from CAD irregular shapes don't collapse at the same rate. It's best to get a scan sent to you and just toss in into gom and simple use preslign...this will put them together as best as can be and you can see any weak areas where the CAD and scan don't play nice.

To simplify the ASME deal...For datum construction you can only use a 'GDT' dropdown which forces ISO. So datum construction really only deviates from ASME when using planes, b/c ASME vs ISO 5459 plane construction is just slightly different. They are both high points. To use ASME datum structure for a hole, you would make the hole at inception as max inscribed and use 'Geometry' instead of 'GDT'.

You only get a Y14.5 dropdown when you choose a GDT measurement like position. So if you previously created a hole using Guassian, but not wanted to used the max inscribed/gage pin measure for position you can by selecting Y14.5.

Hope this helps...FYI there is also a SIGMA selection the programmer must decide on when capturing all the data points on a feature. SIGMA 3 is 99.7% of all points, Sigma 2 is like 95%, and Sigma 1 is like 65%. As you can imagine for a plane, if you select Sigman 3 your flatness will be larger than if you choose Sigma 1. It throws out the outliers as you choose less points. I'm not experienced enough with larger parts that are more easily manually measured to understand when sigma 1 may be viable. I can say that Sigma 1 is another shortcut to remove goofy points near sharp edges. So is the max angle....You can get strange triangles in your scanned data(stl file) So far I found a decent setting to get rid of bad areas without compromising is at 10deg. This setting just says hey....at what deviation in vector direction would we remove this point of the scan.

Good luck, hit me up if you got any more questions, I'm still learning myself. Measuring distances in CAD is 10x more easy than on this software, I'm not sure why they made it so tough.

 

[Burunduk]

Thanks a lot, sendithard, for another very helpful explanation and the additional info.
Could you by chance provide a link to any written material or video that deals with the particular step of "datum creation" - the one with the dropdown to choose between "gdt" and "geometry"?
And I'm still a bit confused on whether that "gdt/geometry" thing takes place during alignment, or when you start inspection of geometrical tolerances, or in between?
I would like to learn more about this particular selection and its implications...

 

[sendithard]

I'll do a video for you...give me a couple days...

 

[Burunduk]

Great. Thanks! no hurry.