Actual Part Center of a Pattern 2

dtmbiz,

Did you mean to specify perpendicularity instead of position?

If you call up a positional tolerance of zero at MMC, an accurate pattern of Ø4mm holes would be your datum. Your Ø1mm tolerance is more challenging. Perpendicularity is not meaningful to me, at least from a fixturing point of view.

JHG

Ok ,lets corretly use the position symbol. Haste makes waste of course, trying to fit this in with my work.

Ok, lets use your senario drawOh. My main point though is to define the pattern center of a produced part so that it can be positioned on the CMM.

The actual part holes were overlayed on the perfect model so I know they fit. Two holes are high and outward of center, while the lower two are inside and above the center plane (line). The holes vary in size as produced. Now where is the center of the pattern in order to postion on a machine to start measuring?

Are you inferring that hole pins are used for setup on a CMM?

Since datum B is not referenced with the MMB modifier "M" this is indeed tricky. You are saying that the profile is anchored to the center of each hole regardless of their size and orientation. If the "M" modifier were after "B" in the profile callout, then things would be much easier (for function, at least).

But if this is how it really is, then the datum could be thought of as the center of the four-hole pattern but taken as a best-fit through the holes while still maintaining proper position/orientation. This is called "regardless of material boundary" (formerly RFS) so if a hole or two wander off as in your actual sketch, the datum is still supposed to be based on the centers as designed. To understand this, think of a fixture of four expandable pins that are located exactly 20 x 40. They expand as much as they can until they hit the edges of the holes; but they will not self-center into each hole. The entire part may skew a little, but the datum can still be thought of as a perfect rectangle.

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

Belanger,

I did not mention a major factor which is that we are working to the 1994 standard. I have heard material conditions are signifcantly changed in the 2009? std.

I do want the profile anchored to the holes.

Considering the 1994 std, is your reply the same?

Yes, the reply is the same. The only difference is the verbal terminology used: what was called RFS on a datum is now RMB and what was MMC on a datum is now MMB.

But the symbology on the drawing and the interpretations are the same for both standards.

So again, the hard part for the CMM is to probe each of the 4 holes but then mathematically construct a perfect 20 x 40 rectangle that "plugs into" the actual pattern.

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

John-Paul,
Assuming there is MMB specified on datum B within profile FCF. How CMM would work then?

Not quite sure of the actual "keystrokes" in PC-DMIS or other programs, but the CMM operator would somehow have to punch in the hard number of 3.0 (in this case) for the diameters -- that's the "boundary" of size minus the geometric tolerance -- and also punch in the locations of 20 and 40. Most likely the actual hole boundaries will be larger than 3.0, and the CMM should somehow be able to then "float" this fixed datum structure around relative the real part's outline to verify profile. But again, that's only if the "M" symbol is after the "B" in the profile callout.

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

Referencing a feature-of-size pattern datum feature at RFS was not defined in Y14.5M-1994, but is defined in Y14.5-2009 (using the RMB nomenclature). JP described the simulator behavior pretty well in his post earlier today.

Some CMM software can establish an optimized datum reference frame using a "soft gaging" analysis like what JP described in his last post. Most CMM software, however, will fit the pattern of actual axes to the pattern of true axes using a least squares "best fit" algorithm. This algorithm minimizes the sum of the squared deviations. The advantages of the least squares algorithm are that it is repeatable, stable, and computationally simple. The disadvantage is that it does not get the same result as the physical gage or soft gage!

Evan Janeshewski

Axymetrix Quality Engineering Inc.

Interesting... I hadn’t realized that a pattern used as a datum was only defined in 1994 with MMC.

I am attaching upgraded sketch to focus on my question. I changed from perpendicularity symbol to position and removed the profile FCF. I am only concerned with inspection the 4 holes called out as datum B. How is the pattern center defined on the actual part to place on the CMM. I believe you have satisfied my questions with the expandable pins and Evan’s software /” best fit”. Just a little more clarification please.

Belanger,
It is my understanding that datum material modification modifiers are not legal with in a profile FCF,
and obviously illegal in the FCF tolerance block. Do you not agree?

If I am getting this correctly, as shown the CMM setup would need to utilize adjustable pins or best fit to software generated zones. I presume the best fit is accomplished by physically moving the part around on the CMM to fit in the zones. Is there a template or marking of the zones on the CMM table or is this done by display readings from the CMM, or is it that the part is placed on the CMM table, probed and then the software is “zero’d” in on the part position?

With the upgraded sketch I hope to get an answer about the verification of the 4 datum holes, pattern center. I would think it is pretty much as described on the preceding posts. As shown the callout is at MMC to datum A. If actual virtual gage pins were used in conjunction with the CMM, would the company be looking at cost of pins for each part configuration, size, tolerance, and positional tolerance for every hole pattern? At this point I am thinking that this would be the case.

A callout for profile can indeed have the "M" symbol if it appears after a datum reference letter. You're right that it definitely can't appear after the number, though.

To really get what you want, you still probably want a "position" symbol instead of perpendicularity on the four holes. (The rest of that feature control frame remains the same.)

I'll let others who know more about CMMs answer the particulars there, but I think you're on the right track.

For the last question, yes, each unique hole size/position would require physical pins made to those unique conditions (if it is decided to go that route of physical simulation).

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

You reposted the same file, dtmbiz.

To clarify a couple things; the fcf controlling the pattern of datum features must be modified at MMC, per '94, but does not have to be modified at MMC for '09. The datum established by the pattern of FOS, when referenced in an fcf, SHOULD (good practice) be referenced at MMC also (with the exception of the datum references in a profile control).

Once you secure the workpiece on the CMM bed, you don't move it ... it would change your results. The CMM needs to sequentially probe each of the pattern's datum features of size, then internally processes the data to get the least squares fit. If you move the workpiece, any inspection done to that point is voided and you will have to re-establish your datums again. Costly and induces errors.



Jim Sykes, P.Eng, GDTP-S
Profile Services TecEase, Inc.

Belanger,

I think you have helped me understand the basics here.

The GDT reference chart for ASME Y14.5M 1994 by "Effective Training Inc. that I am looking at for profile reads;

a column for "datums used"; yes ( profile of a surface); no (profile of a surface -coplanar)

column "MMC or LMC modifier"; NO

column "Bonus Tol"; NO

I take it regarding what you have written that all 3 columns are for the FCF tolerance block?

I am not sure how it would work with at MMC if you can move the part around lets say on a hard gage with virtual pins, then the tolerance zone inner / outer "bands" move too? or are increased in size to include allowance for datum shift?

Jim,

It looks to me as I "appended" the new pdf which shows the original with the new on the bottom, at least thats what I see.

Thank you very much for your reply it helps alot, same to all !

Yup, sorry. I re-loaded the new file four times with the same old file being brought up from my buffer instead of it actually loading the new file. As I was working on another thread, the pdf image changed to show the new content. Aren't computers wonderful.

Jim Sykes, P.Eng, GDTP-S
Profile Services TecEase, Inc.

dtmbiz,

There are several cans of worms here, so be warned ;^)

In a profile FCF, the considered feature cannot have a material condition modifier but the datum features can.

Most CMM software can inspect the position tolerance on the 4-hole pattern relatively easily. Usually, a temporary coordinate system is first created using 2 of the holes. Then the software tweaks the translation and rotation of the coordinate system until the best (i.e. smallest) actual value of position is achieved. So you generally don't need physical pins for this.

Things get more difficult when you have to inspect another FCF that references the 4-hole pattern as a datum feature at MMC. The datum reference frame (coordinate system) is defined in the set of 4 perfect virtual condition pins. If there is any clearance between the 4 actual holes and these pins, the part can be shifted around to get the best actual value. This shifting is where things get difficult for the CMM software. So you can either build a physical fixture with 4 pins, create the CMM coordinate system on the fixture, and physically more the part around to get the best actual value for the FCF. Or you can get "soft gaging" software that essentially does the same thing by digitally shifting the measured point cloud around relative to digitally defined virtual condition pins.

As you suspect, the virtual condition geometry is generally different for every position tolerance so making physical fixtures can get very expensive.

Now just a few thoughts on the pattern center. First of all, don't concern yourself with the pattern center! A position tolerance defines zones for the individual features in the pattern and does not directly control the pattern center. Second, the pattern center can only be defined on a pattern of perfect holes and not on a pattern of real holes. Figure 4-22 on p. 69 of Y14.5M1-1994 implies that the actual part has a "hole pattern axis" but that is wrong. It only works because the actual part in that diagram has 4 holes that form a perfect square. It doesn't work on real parts like the one in your diagram. CMM software can calculate a "pattern center" using least squares algorithm, but it's not the same as the center you would get if the proper 4-pin fixture was used. So the center of the real 4-hole pattern is the center of the perfect 4-pin fixture that the holes fit over. Whew.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

There was a very long thread a year or two back, I think maybe an fcsuper thread, relating to practical application of hole patterns as datums for more patterns of more than 2 holes.

Not sure if that would be of much help.

Really appreciate everyone's time to reply because it really has helped me, particuarly in this thread.

Evan, it may be "whew" for you to write; however, it does help make this topic clearer to me. Thank you

See, more worms!!! OK, when dealing with irregular patterns of datum fos, as Evan says don't get hung up on the "center". Oversimplifying what I go through in classes, I tell seminar participants that they can pick an arbitrary point in space that is located wrt the pattern datum fos by means of basic dimensions, and use that as their origin of measurement. It's a long topic that we've gone through on this forum a number of times. Try these threads for starters:
thd1103_237732, thd1103_235809.

Jim Sykes, P.Eng, GDTP-S
Profile Services TecEase, Inc.

dtmbiz,

Some of the GD&T referene charts I've seen have an asterisk next to that column saying that profile tolerances can't use MMC/LMC, and then a footnote clarifies that it's OK for the datum references. I'm sure a chart by Effective Training would have some sort of allowance for modifiers on the datum references in profile. If not, I can provide a version that does





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