Concentricity Issue

[Kevin Watts]

Good morning folks. i have an issue with concentricity, or how pcdmis calculates concentricity.
pcdmis 3.5 mr 1 ph 10t tp20
please see attached amature drawing. i need to verify concentricity of the 2 radius A and B. my alignment is shown on drawing. im creating seperate circles and concentricity reported by pcdmis is .0129". I can live with that but in checking my Y and Z position i am getting .0000 deviation in Z to each other and .0023" deviation in Y to each other. I dont understand how pcdmis gets .0129" concentricity. I have tried making seperate cylinders on each radius from more circles and still get apx the same. please enlighten me as to a better way to do this, or how pcdmis calculates. from documents on pcdmis it says it takes the 2 features and makes 1 a datum and checks perpendicular distance from centroid of one to the other and doubles the variance. if so the variance would only be .0046" right?
thanks in advance for the help.

[Wingman]

Concentricity is more than just a positional location. Concentricity also includes form. Check the form of the feature you are reporting concentricity. Is should show out of round. Also, if you are using Cylinders the parallelism between the cylinders will be picked up in concentricity.

[kbotta]

Good morning folks. i have an issue with concentricity, or how pcdmis calculates concentricity.
pcdmis 3.5 mr 1 ph 10t tp20
please see attached amature drawing. i need to verify concentricity of the 2 radius A and B. my alignment is shown on drawing. im creating seperate circles and concentricity reported by pcdmis is .0129". I can live with that but in checking my Y and Z position i am getting .0000 deviation in Z to each other and .0023" deviation in Y to each other. I dont understand how pcdmis gets .0129" concentricity. I have tried making seperate cylinders on each radius from more circles and still get apx the same. please enlighten me as to a better way to do this, or how pcdmis calculates. from documents on pcdmis it says it takes the 2 features and makes 1 a datum and checks perpendicular distance from centroid of one to the other and doubles the variance. if so the variance would only be .0046" right?
thanks in advance for the help.

I know I have on occasion beat this to death - but "According to ASME Y14.5M-1994..." Pc-dmis does NOT calculate concentricity correctly. You will not get the correct results there. You will in effect get a positional result.
Regards,
Kev

[Wingman]

Wait, something is wrong here. According to Original Post the value reported was NOT the Positional. If it had been positional then he would have gotten the value he expected and would never have posted it to the forum.

To Kevin Watts
Try aligning your Primary Datum to the Datum Cylinder and reporting Positional to your reported Cylinder. If you get the same result 0.0129 the PC DMIS is reporting Concentricity as TP (I didn't know it was doing this - Hmm).

[cmmguy]

PC DMIS is reporting Concentricity as TP (I didn't know it was doing this - Hmm).

Concentricity sort of looks like TP

[Wingman]

Yes but it should take into account the form / parallelism etc where TP has always reported Centroid of a feature. Concentricity should figure as many median points as possible and take the max deviation from the nom and double it. That is how other software I have used figured it and I assumed PC DMIS did the same.
TP has always used the centroid of the feature with no regard to form / orientation etc. Hmm

[cmmguy]

What CMM software finds the median points of diametrically opposed points on a feature and calculated Concentricity that way... This is almost impossible.

I believe that most softwares use the centroid and doubles it.

[why me]

Debate

Ahh the old Concentricity, TP, TIR Debate you'll find almost as many answers to this question as they’re is inspectors/programmers.
The answer will even vary from Book to Book

[Wingman]

What CMM software finds the median points of diametrically opposed points on a feature and calculated Concentricity that way... This is almost impossible.

I believe that most softwares use the centroid and doubles it.

LK DMIS for one (unless my memory fails me completely and I don't have a copy handy to verify). However, I distincly remember doing journal concentricity and being able to pinpoint deviations in concentricity from journal to journal.
Perhaps most do use the Centroid - not sure. When I get time I will check to see what PC DMIS has to say about concentricity of two cylinders when I purposefully skew one about 0.5mm.

I will also try this in PC DMIS. I will create a Primary Alignment from Two 'parallel' Cylinders (think Journals on the ends of a roll). I will then report Concentricity of each Cylinder to the created cylincer. If Concentricity reports as TP I will get 0. If it reports as something other than TP I will get some other value.

[cmmguy]

LK DMIS for one (unless my memory fails me completely and I don't have a copy handy to verify). However, I distincly remember doing journal concentricity and being able to pinpoint deviations in concentricity from journal to journal.
Perhaps most do use the Centroid - not sure. When I get time I will check to see what PC DMIS has to say about concentricity of two cylinders when I purposefully skew one about 0.5mm.

I will also try this in PC DMIS. I will create a Primary Alignment from Two 'parallel' Cylinders (think Journals on the ends of a roll). I will then report Concentricity of each Cylinder to the created cylincer. If Concentricity reports as TP I will get 0. If it reports as something other than TP I will get some other value.

Why I say that none do it is because it is almost theoretically impossible to find two points that are exactly diametrically opposed to one another. Concentricity is very difficult to evaluate per the definition.

[Wingman]

Why I say that none do it is because it is almost theoretically impossible to find two points that are exactly diametrically opposed to one another. Concentricity is very difficult to evaluate per the definition.

Well I can tell you that I got something other than TP and LK sure passed it off as Concentricity - LOL. If you are in fact correct then they had me (and a lot of others) fooled with what they were doing.

Still, I will conduct these tests because they will draw a line between TP and something else.

[kbotta]

Have conducted same test's, 2 yrs ago (v 3.2063mr3) same result in TP as Conc. He may have another issue if the reported value is questionable.
Also, It is entirely possible to have an eliptical feature check "ok" (in theory) therefore it does not control SURFACE form (but rather median point form)
0.02
Kev

[Wingman]

Well the original post certainly has another issue with the Primary Alignemnt. When he reports TP he is usint the Y+ Alignment. When he reports CCEN he is using the 1st Cyl as the Datum.

On the testing issue though, OK. Saved some time. No need to test PC DMIS. Now I am curious about my memory of LK as I remember it DID CCEN as CCEN. So, I will have to send a Couple Programs to some friends that use LK and have them run them so I can see if my memory is failing me. Unless somebody out here also has an LK and want to do this test???

[kbotta]

Well you can answer it quickly : were your points diametrically opposite? 180 deg apart? legth of feature? rotating around feature?

This is an extremely tough spec to varify, and very time consuming.
(The standard even suggest's taking another route.)
kb

[Wingman]

Well, thoeritically the points were diametrically opposed, yes. And, the spec suggests other methods for several things. Plus, you can mathematically extrapolate 'best guesses' of diametrically opposed points even if you don't have exact points. Of course, the more exact you are the closer you will be to correct.
At any rate, a guess using parallelism to add or subract from a centroid is closer than True Position. Like I say, I can't be sure because it has been a while but I am pretty sure that LK DMIS calculated CCEN differently than TP. Other factors were involved because I can remember obtaining the exact points from the features that were throwing CCEN out.
I would be very interested in seeing how LK DMIS does this.