Standard or guiding principle for how to validate tolerances (i.e. profile)

[DonkeyPhysics]

From what I understand, there is no official standard I can point to that would indicate specific methods or instrument types for validating geometric tolerances specified in accordance with ASME Y14.5 (whatever year).

What I'm wondering is... are there any guidelines I might be able to refer to? I ask because our machine shop here is very new to the formal part validation process. We have a Faro arm, but other than knowing how to pull the trigger and define surfaces, theoretical planes, etc from point clouds... the expertise is missing for deciding how many points are really necessary to validate any given surface/profile, etc.

Currently, it's up to the engineers here to essentially TELL our quality folks (who are doing the measurements) how many points/distances they have to measure to validate compliance. To be fair, even GD&T is a new thing here... just getting warmed up.

To give an example of what I'm looking for, let's say I have an annular disc, made out of stainless steel, basically shaped like a flat washer, but with a thickness of (let's say) 150mm, and an outside diameter of 1m. For the same of simplicity, let's say I apply a (unilateral) profile tolerance of 0.2mm to the "top" surface (call it surface B), relative the bottom one (Call it surface A). I'm not including a picture because this is an imaginary part with a fairly simple shape. Hope that's OK.

I'm trying to figure out if there's a reasonable way to decide how many points I should measure on Surface B to validate my profile tolerance (i.e. with the goal of guaranteeing that the thickness of my disc will fit into the notch it's design to slide into). I imagine there are statistics and metalurgy knowledge that could indicate the part's tendency to warp, but... my inclination shy of that is to just pick a number from my gut. Influencing factors are that my part will be static, so I don't need to worry about rotational balancing, but I will have rotating parts next to it.

Any thoughts? Hopefully I've made myself clear, but if you have questions, or would work better with a picture and more details, let me know and I'll be happy to provide.

Thanks!

 

[DonkeyPhysics]

I did read about the ASME Y14.43-2011 standard, but that seems to only cover the building of gages for checking parts at MMC & MMB. My case specifically requires measurement by CMM.

 

[KENAT]

Y15.5 does reference a measurement spec though I believe it may be set up for gauging etc. not CMM.

Trying to establish a datum by taking an average of multiple points on a part face does not actually approximate the ASME definition of a datum plane as I understand it.

The datum plane should be derived from effectively the '3 high points' of the plane.

One way to better approximate this while using a CMM may be to set the part up against the 3 'flat' gauge plates and then generate you datum planes from the gauge plates. Something like that anyway - I'm not an inspection guy.

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[MikeHalloran]

<rant>

The whole business of using CMMs to define 'theoretical planes' from point clouds has set back metrology a couple of decades, just as a spike in production of fax machines set back electronic communications.

Forget theoretical planes. You can't measure them, and you can't rely on them as datums because you'll never derive the same theoretical plane from the same object twice.

Forget associating a datum with a part. A datum is a feature of a perfect fixture. The set of datums in a particular fixture should locate a particular part in the same way every time the part is removed and replaced.

</rant>

A profile tolerance defines two theoretical surfaces, within which a given surface should lie, completely. You can't know in advance how many points you have to measure to assure that; you have to survey the surface of the part. CMMs should be good at that, and they are, if you disallow using theoretical planes derived from part features as datums; that just confuses everyone, and adds expense and time to what should be a straightforward process.



Mike Halloran
Pembroke Pines, FL, USA

 

[DonkeyPhysics]

Mike, I hear you, and I see your point. Perhaps I shouldn't have mentioned that possibility as a "thing" that our CMM operator knows how to do. Ultimately, it's not directly related to my main question, though.

But ok then, it seems that there's really no set standard for how many points to measure on a surface (be it curved, flat, mathematically-defined, etc) to validate that it complies with the profile tolerance requirement. So, short of using a gage, I could basically tell them to validate the part at x number of locations (distributed approximately evenly over the surface), where X can be any number that my conscience deems appropriate?

You mentioned I should "survey the surface". I'm curious what you mean by that. Are you talking about a visual inspection to spot and decide on areas to measure? Or does a Faro arm typically have a feature to do a sort of analog sweep of the surface? That is, is there some mode where one can say... start recording at one point, then sweep the arm back and forth across the length and breadth of the surface (with the measurement tip remaining in contact) and have the system record a... line...or line graph, the entire time, such that I'd be able to pick out the extremeties? Or is taking a point cloud the only way to accomplish that? If I'm checking a profile tolerance of, say, 0.2mm, that's not something I can just visually spot from looking at the surface (i.e. to know where to measure). So, having some sufficiently fine way of surveying the surface is my ultimate goal here.

 

[MikeHalloran]

I interviewed at Faro once, thought I had the job, but had a 'no thanks' in my inbox before I got home. I came close to guessing how they got the resolution needed for metrology, but I don't know if they had or have a 'survey' function.

Should be do-able in software, though. ... say the hardware guys.

Fast way to get one limit of the surface is a reference surface and Prussian blue.
The blue, or some kind of gel, might also give you a good guess on the other limit, too.
Sometimes fancy technology is not the best solutiion.




Mike Halloran
Pembroke Pines, FL, USA