hexagon logo
Register Sign in
  • Replies 13 replies
  • Subscribers 222 subscribers
  • Views 290 views
  • Users 0 members are here
Options
Related

Perpendicularity of cylinder

Unitn
Hi everyone,

I am new to the forum and by searching the topics I couldn't find the answer to my question.
I need to measure the perpendicularity of a hollow cylinder according to its planar surface used as datum. I have used two different approaches in this regard, but I obtain very different results.
in the first approach, I measure 5 circles at different levels along the cylinder height [using auto circle feature] then I construct a cylinder from these 5 circles [constructed feature] and finally I chose Perpendicularity Dimension to measure the perpendicularity.
in the second approach, I use Auto Cylinder feature[using Adaptive Cylinder Concentric Circle Scan] to construct the cylinder geometry as showed in the attached image, where the scans are performed at the same positions along the axis compare to the previous approach. by measuring the perpendicularity according to the same datum as the first approach, this time I get perpendicularity values which are much worst than the previous case. [in first case I get 0.006 mm while in second case it is 0.098mm].
I would appreciate if you can help me find out which of these two approach is correct, or is there a better solution to measure the perpendicularity precisely?

Thank you in advance.
Attached Files
  • I think the problem in the first case is that you are constructing a cylinder from circles. I think you need to construct the cylinder from the circle hits.

    But, I also don't do any analog scanning, so I can't say that there isn't a problem there. I would think that your adaptive scans are the better (more correct) result.
  • I think the problem in the first case is that you are constructing a cylinder from circles. I think you need to construct the cylinder from the circle hits..

    You can construct a cylinder from constructed or measured circles, just not auto circles according to the help file anyway. I am guessing it will just use the original hit data.


    to the OP, you could check rather crudely which answer is closer to reality by measuring a circle at the top of the cylinder and aligning to it (set your origin on the centre of the circle). Then measure a circle at the bottom of the cylinder, do a quick linear dimension and see how far the center of the bottom circle deviates from the center of the top circle.

    I would say that the auto cylinder method is probably the best way to go about it.

  • Just remember when you dimension. PERP of Cylinder TO DATUM-A-, not the other way around.
  • I would just do an Auto Cylinder scan back to -A- like Schlag suggested.



    This will separate the circles at each level automatically. Just click the little plus sign, if you need to analyze each level separately.




    either way Xactmeasure or Legacy
  • in reply to VinniUSMC
    "Construct cylinder from circles" is a special case in PC-DMIS, and uses all the measuring points of the circles, so it shouldn't be the source of the problem. BF or BFRE in the construction might give small differences, but I doubt it could be as much as is the case here.

  • in the first approach, I measure 5 circles at different levels along the cylinder height [using auto circle feature] then I construct a cylinder from these 5 circles [constructed feature]
    ...
    in the second approach, I use Auto Cylinder feature[using Adaptive Cylinder Concentric Circle Scan] to construct the cylinder geometry

    An (non-)obvious difference between those methods is that for the constructed cylinder, the length will be the actual span of the hits, while for the autocylinder the length will be taken from the CAD surface(s) involved. As you can never put actual hits over all of a surface, this will always generate a smaller perpendicularity value for the constructed (or measured) cylinder, compared to the autocylinder, even if you construct a cylinder from the autocylinder's hit points!

    Using the (P) option in the FCF, you can tell PC-DMIS the computation length explicitly, to eliminate this difference.
  • in reply to VinniUSMC
    Ok. I don't do much work with cylinders of late. I just thought I recalled having read something about it being not so good. Thanks for the clarification Anders.
  • Measure a plane around the feature...level and make that an origin...and then measure an auto-cylinder and then report the perpendicularity of that feature back to the plane.... pretty straight forward
  • Thanks everyone for your useful comments.
    As I realize from most of the comments, the correct approach to measure the perpendicularity is to use the Autocylinder feature. afterwards, first I tried to construct a cylinder only by two scans at top and bottom of the cylinder and I got a perpendicularity value of 180 um, then I introduced more scans between the two previous ones by increasing the number of rows, and every time it happens that I get different values for perpendicularity [up to 250 um]. I was wondering what could be the reason for this differences and also if there is any recommended standard guideline for the number of scans regarding the height of the cylinder.
    Thanks again for your valuable comments.
© 2024 Hexagon AB and/or its subsidiaries. Privacy Policy | Cloud Services Agreement