Fit To Datums

[parrottk]

Can someone give a good explination of the difference in true position when the fit to datums check box is on or off?
When I have the check box on all holes match nominals perfectly. (NO DVIATION AT ALL).
Check box turned off shows deviation that matches the actuals in the measured features.
when the check box is on the part looks to good and questions arise in regards to the accuracy of the gage and or the programer "ME".
thanks for the help.

[Matthew D. Hoedeman]

The reason the holes look so **** good is becuase instead of PCdmis using the center of the datum holes as 'master', it uses the hole + MMC to SHIFT the part around. It will shift it as much as it can (and still keep the datums on an imaginary pin) to put the dimensioned hole as close to nominals as it can.

BEST SUGGESTION:
Create an alignment using the datums for the TP callout, then simply TP the hole WITHOUT using datums so you can see how the hole actually and truely relates to the datum holes. The way Pcdmis does it, they SHOULD also be reporting the location of the shifted datum so you can see what it is doing. I think it is pure ***** the way it does it. The MMC bonus for the datum holes SHOULD show up as a bonus to the TP for the hole you are dimensioning (and I think about 99% of the other users feel the same way). It USED to do it this way, way back in the dawn of pre-history, but it made too much sense to all the users, so they changed it.

[William Johnson]

When I have features that have MMC,MMC,MMC I will tolerance the feature twice. one with use datums on and one without. This allows me to see if the part is in print according to the print with use datums on, and will also show me how much deviation that I have with use datums off. With use datums on PCDMIS will translate and rotate to simulate a hard gage fit which is the proper way to check it but will not tell you how much deviation that you have. Also when you use the use datums feature you must pick all 3 datums in the proper order.

[JamesMannes]

The reason the holes look so **** good is becuase instead of PCdmis using the center of the datum holes as 'master', it uses the hole + MMC to SHIFT the part around. It will shift it as much as it can (and still keep the datums on an imaginary pin) to put the dimensioned hole as close to nominals as it can.

BEST SUGGESTION:
Create an alignment using the datums for the TP callout, then simply TP the hole WITHOUT using datums so you can see how the hole actually and truely relates to the datum holes. The way Pcdmis does it, they SHOULD also be reporting the location of the shifted datum so you can see what it is doing. I think it is pure ***** the way it does it. The MMC bonus for the datum holes SHOULD show up as a bonus to the TP for the hole you are dimensioning (and I think about 99% of the other users feel the same way). It USED to do it this way, way back in the dawn of pre-history, but it made too much sense to all the users, so they changed it.

Mathew,
As of 4.1 the option is available to us to show datum shift.

[Jan d.]

In V4.1, PC-DMIS allows you to actually put in the TP exactly as it is specified on the print. But I share Matt's concerns.

I ended up outputting TP in 3 different ways:
1) No MMC on anything. In other words, RFS.
2) Only MMC on the TP tolerance.
3) MMC on TP call-out AND datums.

1 gives me more process control information. 2 can sometimes behave like you describe and there is no way to get any process info from that.

I think that PC-DMIS evaluates 1 and 2 correctly. After staring at the results of hundreds of holes, I am quite convinced about that. However, I am still not convinced that PC-DMIS always evaluates number 3 right. I have not finished that debate with Hexagon yet.


Jan.

[Matthew D. Hoedeman]

But I still want it to show it the way it USED TO, show the MMC of the datum as a bonus to the MMC of the feature. I want the datums as datums, not floating-datums, and the location of the hole to be shown as a deviation FROM THE DATUMS, not the 'best fit on a pin' location and how much the datums were shifted.

What happens when the datum hole is MMC as well as the dimensioned hole, but the assembly method uses a tapered pin that goes in the datum? The datum hole WILL BE ZERO, not shifted, and I need to know, WITH A SINGLE DIMENSION, how far out of relationship the dimensioned hole is. THIS is how Pcdmis used to do it, and if this is the type of dimension you need, then you HAVE to do an alignment BEFORE dimensioning the hole and god help you if you need to MMC bonus from the datum hole for the dimensioned hole, you can't get it any more, unless you do a bunch of assignments and add it to the MMC tolerance, then try to explain THAT to the customer.

OK, 'they' (that same infamouse 'they' that always comes up) say the current way is the correct way (but 'they' said the same thing BEFORE they changed it, too!), BUT only if the datum hole is to be put on a straight pin, not a tapered pin. If a tapered pin is used, then the old way is the way it needs to be. BOTH case are correct use, but ONLY 1 is offered to us at a time (I think you have to go back to V3.2063 to get the other method). Why not offer us BOTH? Then they would not have had to ADD the 'datum shift' dimension. Hey, some people do not allow the datum shift (no matter WHAT the GD&T says, but they will allow the MMC from the datum), so I am still bent over a barrel if I need to MMC from the datum hole BECAUSE they do not any longer have the 'taper pin' TP method/option of reporting TP.

[parrottk]

Thanks guys

Now it makes sense to the powers to be.
only had to draw a few pictures to go along with your explinations.

[Stuart]

FIT TO DATUMS.... means if you have MMC on your datums, PCDMIS will 'shift, translate, rotate' your coordinate system using the BONUS on the datums to get the best possible True Position result. PCDMIS 3.7 does this by default. Using this option, it is possible to get a 0.000 True Position frequently, as it reduces the BONUS to improve the TP.

If you uncheck FIT TO DATUMS, then the coordinate system stays where it is (i.e. the true CENTERS of the datum holes don't move) and you get a True Position value, with a TOTAL available bonus.

The latter is what 99% of the industry is looking for.

Stu

[Matthew D. Hoedeman]

FIT TO DATUMS.... means if you have MMC on your datums, PCDMIS will 'shift, translate, rotate' your coordinate system using the BONUS on the datums to get the best possible True Position result. PCDMIS 3.7 does this by default. Using this option, it is possible to get a 0.000 True Position frequently, as it reduces the BONUS to improve the TP.

If you uncheck FIT TO DATUMS, then the coordinate system stays where it is (i.e. the true CENTERS of the datum holes don't move) and you get a True Position value, with a TOTAL available bonus.

The latter is what 99% of the industry is looking for.

Stu

This bonus is ONLY from the dimensioned hole, not any additional from the datums. AND YES! THIS IS WHAT MOST OF US WANT, NEED, AND REQUIRE.

[Thomas Stewart]

Do I understand correctly?

Did I understand you correctly Jan to say that "In V4.1, PC-DMIS allows you to actually put in the TP exactly as it is specified on the print."?

Are you saying that 4.1 is the first in a long time where, if you used the Datums (-A-B-C-) to align the part, you can actually recall not only the MMC for the feature but also the MMC for the Datums (-B- & -C-) as well without it shifting and turning the part in a seemingly unpredictable attempt to show Datum / feature relationship?

If not I am sadly disappointed because as we (in metal stamping) can agree, that is primarily all we need (in terms of this issue of course) to FULLY report diametrical features as the print describes.

If so... Sign me up for 4.1!!!!

[Stuart]

Thomas,

Yes, your assumption is correct. Unchecking the FIT TO DATUMS in PCDMIS 4.1 will give you the result you are expecting.

Stu

[Jan d.]

Are you saying that 4.1 is the first in a long time where, if you used the Datums (-A-B-C-) to align the part, you can actually recall not only the MMC for the feature but also the MMC for the Datums (-B- & -C-) as well without it shifting and turning the part in a seemingly unpredictable attempt to show Datum / feature relationship?

Yes! 4.1 is the first and it does exactly that. It allows for virtual conditions on datums. And boy, do I use that! If you do a report, you can see all the gory details of how it wiggles to get the best "fit" of the functional gauge (meaning: the smallest TP).

Be careful though: This may have been implemented correctly IF your datum has only 1 Virtual Condition. I am not 100% sure but it looks as if it has been done right.

HOWEVER: I have many FCF on datums that have 2 VCs. PC-DMIS seems to add all of them up giving me a humongous bonus. I think that is wrong (see an earlier post regarding this issue). So I am still quite hesitant to use MMC on datums with more than 1 VC.

My suggestion: try it and REALLY study the results. It may work if the TP is to a datum with just one VC.



Jan.

[Benedictj1]

This makes my brain hurt...few more years and I am playing guitar all day..