hexagon logo
Register Sign in
  • Replies 9 replies
  • Subscribers 222 subscribers
  • Views 106 views
  • Users 0 members are here
Options
Related

Variation on Cylinder due to probe touching angle

RobOntario

I learned from 101 that for cylinder, the probe should vertical touch the point ( we use analog probe with 2 mm header), the first 3 or 6 even 8 points compose a plane. I believe this definition is good for compensation of the probe header. Now we have a product with an around 5 mm diameter wire which welds on the part. On the drawing the wire could have 2 mm center variation on one workplane. The program measures it as a cylinder( touch 8 points in 180 degree) with a fix angle which I believe has 10 degree away from vertical cut the cylinder (around 80 degree to the plane). I wonder how much variation does this measurement bring it due to my concern and anything we could help reduce those kind of variation?

Thanks
  • good day,

    On the drawing the wire could have 2 mm center variation on one workplane

    hm do you mean "position" ?
    why do you measure a "wire" in the first place ? is this relay relevant with 2mm ? could you test the "function" of this part?
    is this "wire" bended per hand ? variation of measurement depends heavy on how it is produced.
    i think your CMM is the smallest error in this chain of production, and can be ignored.
    i can't realy tell if this is a good way to measure this without the drawing

    print out the position or xyz-values from cyl12. than you can see if this match your CAD up to 2mm
  • if you want to exact the most accurate location & diameter of this wire:
    you should try to establish a true as-found type of axis of the wire (not depend on its theoretical location), by sampling a few hits with prudent prehit/retract distances (two circles for example),
    with the sample circles, you can establish an axis and locally within the routine align to it
    once locally aligned to the actual measured axis of the wire, you can then measure the circles with your hits normal to the wire.
    doing so would eliminate the potential cosine error, making the location and diameter most accurate.
  • I'm not sure what you are talking about when it comes to using vertical touch points. I'm interpreting this as the 101 course said to measure the cylinder with the probe oriented in the same direction as the cylinder. This may be ideal, but it is not necessary. Quite often you just have to use the best probe angle you can get to measure a cylinder. Having the probe at a different angle may mean that you can't measure all the way around with one probe angle, but depending on your needs, you may get by with measuring just half of the cylinder, especially if you don't need to evaluate the form. If you need to measure all the way around and need 2 or more probe angles to access all the feature, that is fine, just be mindful that you are adding a few microns of error from the probe head's imperfect repeatability and the larger span you will be using of the machines measuring volume.

    You may be concerned about the probes actual touch accuracy if the probe is at a different angle from the cylinder. This is quite negligible, especially if you have an analog probe. If you have a touch trigger probe it takes a little more force to take hits that are in the same direction of the probe vs hits that are square to the probe, but with an analog probe that is all compensated for. If you ever find yourself using a touch trigger probe and want to see how much this affects your measurements, you can measure your qualification sphere as a sphere with several hits at several levels and report/graph the form error. You would be able to see how much form error the probe and stylus combination has due to the different amounts of force needed to trigger a hit a different vectors. Normally this is just a couple of microns.

    As for taking your first hits at the same level, that is just something to do to help PC-DMIS figure out the orientation of the cylinder when you are using manually measured hits to define a cylinder. It helps if you take a lot of manual hits at several levels. The more hits you take the more likely the software will correctly identify the type of feature you are measuring and the proper location and vector. You will know right away if the software misinterprets your intent. You'll end up getting the wrong feature or it will be pointing the wrong direction in the graphic display window. You may be able to override the guess, but in many cases you will just need to delete the feature and try again with more hits.

    All that said, louisd gave some good advise for measuring that feature accurately. Considering how small that feature is and how much it is allowed to shift in position, a local alignment will help a lot.
  • in reply to Henniger123
    Sorry for the delay. Our supplier made the wire ( They call it a wire, a little bit strong). We have drawing for this.
  • in reply to louisd
    Yes, a little bit busy for this, I think it could be more accurate.
  • I'm not sure what you are talking about when it comes to using vertical touch points. I'm interpreting this as the 101 course said to measure the cylinder with the probe oriented in the same direction as the cylinder. This may be ideal, but it is not necessary. Quite often you just have to use the best probe angle you can get to measure a cylinder. Having the probe at a different angle may mean that you can't measure all the way around with one probe angle, but depending on your needs, you may get by with measuring just half of the cylinder, especially if you don't need to evaluate the form. If you need to measure all the way around and need 2 or more probe angles to access all the feature, that is fine, just be mindful that you are adding a few microns of error from the probe head's imperfect repeatability and the larger span you will be using of the machines measuring volume.

    You may be concerned about the probes actual touch accuracy if the probe is at a different angle from the cylinder. This is quite negligible, especially if you have an analog probe. If you have a touch trigger probe it takes a little more force to take hits that are in the same direction of the probe vs hits that are square to the probe, but with an analog probe that is all compensated for. If you ever find yourself using a touch trigger probe and want to see how much this affects your measurements, you can measure your qualification sphere as a sphere with several hits at several levels and report/graph the form error. You would be able to see how much form error the probe and stylus combination has due to the different amounts of force needed to trigger a hit a different vectors. Normally this is just a couple of microns.

    As for taking your first hits at the same level, that is just something to do to help PC-DMIS figure out the orientation of the cylinder when you are using manually measured hits to define a cylinder. It helps if you take a lot of manual hits at several levels. The more hits you take the more likely the software will correctly identify the type of feature you are measuring and the proper location and vector. You will know right away if the software misinterprets your intent. You'll end up getting the wrong feature or it will be pointing the wrong direction in the graphic display window. You may be able to override the guess, but in many cases you will just need to delete the feature and try again with more hits.

    All that said, louisd gave some good advise for measuring that feature accurately. Considering how small that feature is and how much it is allowed to shift in position, a local alignment will help a lot.


    Thank you for the reply. My concern is for cylinder, if the touching is not vertical, after the compensations added on, it could be an oval and no matter which way the software takes to calculate, the center would have some shift. For PC-DMIS, if it does not set the vertical condition for the first plane, in some case( 6 points, 8 points), the cylinder would be not identical.


  • Thank you for the reply. My concern is for cylinder, if the touching is not vertical, after the compensations added on, it could be an oval and no matter which way the software takes to calculate, the center would have some shift. For PC-DMIS, if it does not set the vertical condition for the first plane, in some case( 6 points, 8 points), the cylinder would be not identical.

    If you measure the cylinder directly as either a 'measured cylinder' or a 'auto cylinder' the vectors of each point will all be relative to the vector of the cylinder. I guess a different way to say it is that the software will compensate the hit angles based on the theoretical angle of the cylinder.

    If you prefer to measure 2 of more circles and then construct a cylinder from those, you can run into problems with the vectors not being right if the vectors of your circles are off. A good way to avoid that is to use the CAD to select 'Auto Circles'. Auto circles will define the correct vector no matter what workplane or alignment you are in. Then you can construct a cylinder from those. If you want to make that even more accurate, use the 'BF Recomp' method when constructing the cylinder. That constructs a cylinder using the probe tip center point of each hit and then compensates for the tip radius. It can remove some of the cosine error you may have if the cylinder's angle is off a bit from nominal. That said, the method louisd suggested is more accurate, but admittedly slower.



  • in reply to Cris_C
    Correct vectors are required for proper probe comp. IN THE CASE of spheres & cylinders, there is NO probe comp done until the final touch is taken, the sphere or cylinder is constructed AS IF there was no probe comp, THEN Pcdmis comps for the probe radius. The same is mostly true for a circle as well, BUT, the comp is done in a '2D' manner, based upon the theo vector of the circle UNLESS you use 3 surface sample hits, then those are used for the '2D' probe comp.
  • if you want to exact the most accurate location & diameter of this wire:
    you should try to establish a true as-found type of axis of the wire (not depend on its theoretical location), by sampling a few hits with prudent prehit/retract distances (two circles for example),
    with the sample circles, you can establish an axis and locally within the routine align to it
    once locally aligned to the actual measured axis of the wire, you can then measure the circles with your hits normal to the wire.
    doing so would eliminate the potential cosine error, making the location and diameter most accurate.

    Thank you. This is the best way to check the wire.

    I found I did ask 20 months ago for a similar question.

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