hexagon logo
Register Sign in
  • Replies 7 replies
  • Subscribers 221 subscribers
  • Views 134 views
  • Users 0 members are here
Options
Related

Alignment Questions and Customer Requirements

The Hermit
I have some questions for the GURU's. Hopefully this illustration is sufficient. Envision this as if you were looking from the Z+ axis, down. All three cylinders are at the same Z for location. The requirement by this customer is that I make a pierce point with each of these cylinders, on their respective planes. In this case, two on datum A and one on datum B. I'm then instructed to create a plane out these, for this purpose, in the ZAXIS. I'm then to use that plane to level to, rotate to the two A datum pierce points and then zero on the D point.

This is where I run into my problem. What, if for example, one of the Z locations of those cylinders is 0.010" off from the other two? The have nothing on the print indicating relationship between those three dowels, other than construction of the plane. They use this alignment scheme on nearly all their parts, some as large as 20 inches long, and their positional tolerancing is from 0.0008"-0.0010" at RFS, regardless the size of the part. In some cases tolerancing will be tight the smaller the part gets. They also require that we send reports with each PPAP part and require absolutely no out of tolerance dimensions on the report. We spend hours in some cases to achieve that, as for some reason the cmm more often than not will get a diameter out of tolerance. Most diameters are toleranced from ±0.00045" - 0.0010" and can vary from 0.500" - 6.000" in size. These tight diameters all get scanned and the roundness ranges from 0.0000"- 0.0001" Most times when verifying diameters with a two point bore gauge, we get them good. I report these using least squared, in Legacy. Reason being, if I switch to use max-inscribed or min-circumscribed, or report with UAME (I may have those backwards as I always forget which one is which) the cmm numbers conflict with the bore gauge, which operators are required to use for in process inspection.

This is the second part of my issue. The customer runs these on their Zeiss to confirm our results. They have failed PPAP's for 0.0003mm, either positional or size! That's the error of the cmm, if I'm not mistaken. Then they have error in their cmm, and there's error in the CNC as well. Here is the last prickly part of this problem, the customer refuses to allow hard gauging to be the arbiter on whether or not the part is correct. All that matters at the end of the day is if they get a green report on their cmm.

I really don't know where to go with these parts anymore. This has been a two year battle and we're not any better off today. We can run the same part 3 times in a row without moving it from the cmm, and get three different results, which understandably causes mistrust with operations. The only thing I can see as potentially having an impact on results is that we run these by jimmy-rigging fixturing on the cmm. A vise, jack screws in some case, whatever we can use to make the part stable. Some of these parts weight 200 pounds, others less than 3 pounds. Am I wrong to believe cmm fixtures designed for the specific parts would yield better accuracy and more reliable consistency?

Thanks in advance for your input.




​

​
Attached Files
  • The callout seem to indicate that the centerline of datum D is primary and the three cylinders are locking rotation. So, the method actually used is wrong. You would need a tertiary datum in order to fully constrain the alignment though, as it is missing in the X-direction.

    If the parts are large and heavy, maybe it would be better to lay them on the granite on datum A or datum B? Depending on the material, some part can "sag" under their own weight when placed vertical.

    Datum (and datums) sequence should be defined after function so my suggestion would be to follow it unless there is a problem with the design (which is different issue).
  • As 3 points are not on the same plane what are you using to constructing the 3 point plane ? Offset plane ? using the 3 points as a datum plane there should be no error in Z axis.

    Is the CMM in a temperature controlled room ? Are the parts allowed to soak in the room for 24 hours.

    Is your touch speed the same as your tip calibration speed ?

    Are you using the shortest probe possible to measure the part ?

    When working such tight tolerances all these things can have a factor on results.
  • The callout seem to indicate that the centerline of datum D is primary and the three cylinders are locking rotation. So, the method actually used is wrong. You would need a tertiary datum in order to fully constrain the alignment though, as it is missing in the X-direction.

    If the parts are large and heavy, maybe it would be better to lay them on the granite on datum A or datum B? Depending on the material, some part can "sag" under their own weight when placed vertical.

    Datum (and datums) sequence should be defined after function so my suggestion would be to follow it unless there is a problem with the design (which is different issue).

    I would agree it's wrong, or, at a minimum, unconventional, but it's what they require me to use. Laying it down is possible with the features I need to get.
  • As 3 points are not on the same plane what are you using to constructing the 3 point plane ? Offset plane ? using the 3 points as a datum plane there should be no error in Z axis.

    Is the CMM in a temperature controlled room ? Are the parts allowed to soak in the room for 24 hours.

    Is your touch speed the same as your tip calibration speed ?

    Are you using the shortest probe possible to measure the part ?

    When working such tight tolerances all these things can have a factor on results.

    The dowels in question are all at the same Z location, theoretically anyway, so an offset plane is not necessary. I agree there "should" be no error in Z, but I can't convince the customer of that. Temp is another issue. We don't let them sit overnight, but that's really an impossibility in daily inspections. They "could" make three shifts of bad parts in the mean time.
  • Sorry i misunderstood so the cylinders are in the side faces.

    Could you post the program ? one of us might be able to spot something that may improve your alignment.
  • Sorry i misunderstood so the cylinders are in the side faces.

    Could you post the program ? one of us might be able to spot something that may improve your alignment.

    I'll have to clear that with management first. I've never posted a program before. Is there anything I have to do to it first? I will say that initially, I created alignments that seemed obvious. The customer directed me to construct the alignment in the manner relayed in this post. I've never seen anything like it before, even in a GD&T manual and I've been programming for nearly 20 years. I think they designed them in house, specifically for their parts. I had Hexagon Apps support in house for two days. I had them look at these alignments too. There's no way I can find to get them to work with the Geometric Tolerancing panel, either. Apps support failed as well.
  • There are some things to consider I guess.
    Sounds like they are making a Plane, Line, Point alignment? (kinda, with lots of built in error and almost zero part to part repeatability)

    Without seeing the drawing, if they are "drawn" in line with each other, its possible they have a cheeky internal spec explaining that all things drawn with implied center lines must be inline within xxxx amount.

    Without seeing your code there are some questions.
    Are you doing a manual alignment? If so is that followed by a DCC alignment, and if so (It pains me to say this, but a second DCC alignment to further refine things)

    Have you checked the accuracy of your machine? Something as simple as a ZZ ring gage held the proper angle in a vise, preferably around the size of the features you are checking. Never hurts to double check things, especially when your honor is called into question.

    Good tooling will make a big difference in repeatability of your measurements.
    For the large parts you will need your CNC guys to help you out, for the smaller ones you could just 3d print them on a cheap printer. Looks like an Ender 3 is less than $200 on Amazon, that will be just fine for most parts and your tooling will cost you a few pennies. (and quite a few hours of waiting for it to print)


© 2024 Hexagon AB and/or its subsidiaries. Privacy Policy | Cloud Services Agreement