hexagon logo
Register Sign in
  • Replies 8 replies
  • Subscribers 5 subscribers
  • Views 250 views
  • Users 0 members are here
Options
Related

Creative suggestions on probing sheet metal

matl59
Im working with a gantry style machine and a large cylindrical part with a sheet metal section. In the sheet metal all the way around the circumference of the part are holes that I need to check positional accuracy of. I was going to use a shank type stylus but it appears that I'm limited to 5 degree incriments in PC-DMIS and this will not allow me to probe the holes at a perpendicular angle thus skewing their position left or right depending on the angle used...I've read in other posts just how much people dislike these styli for this very reason...

I can not probe the edges of the sheet metal with a normal ball type stylus without going to far in or out which also skews the results.

I've been told they've used plugs here in similar situations to allow for probing on a more substancial surface but I'm sure the accuracy of probing the plugs is also questionable. We'll probably be goin this route but I'd be curious if anyone had any thoughts. There is over 100 holes so thats alot of plugs to have made not to mention the amount of time lost during the installation and removal of the plugs for each part probed...

Wish I was able to calibrate and program the probe for all the crazy angles I'd need. Then I'd give the shank probe a try.
Parents
  • I'm thinking the part is as you describe. It is a very difficult problem. I think there is a solution, but it will be very difficult to do and even more difficult to explain. It will no doubt take some tinkering/experimenting preferably with a "known" part so you'd know when your getting good results.

    I'm thinking it may even require iterative measurement involving lots of alignments.

    For an example:

    Measure and align to the cylinder. We'll call it Z+ for this example.
    Measure the hole knowingly shanking the probe (only want rough location).
    Align to the cylinder and use the shanked hole to fix rotation about the cylinder. Rotate to X+
    Change workplane to X+.

    There are two possible avenues of exploration from there.

    1) autocircle with 1 sample hit above or below the hole along the cylinder's vector. Edit the x value of each hit individually to get it to hit the edge. This will work if you think you can get it to hit the edge (thicker "sheet metal" like maybe .1 inch thick or more). X values could probably be calculated knowing the hole's diameter and the cylinder's diameter, but I think it would be a lot of work.

    2) Edge points (with sample hit(s)) all around the hole, then construct a circle from them. This might work better for thinner "sheet metal", but the sample hits might prove problematic as they won't be "normal" since the surface surrounding the hole isn't "normal"

    Now, you might need to go through the alignment and measurement step again to get the best results.

    Following that, lather, rinse, repeat for all the holes. Unless you can figure out a way to copy/paste with pattern or loop it (I'm not so sure that works when alignments are involved).

    I have no idea if either approach would work. I'd really need a part to play with to get an idea of that.
    About the ONLY thing I could think of would be to set the SAMPLE HITS to 4 (if you can, can you?) and set it to take 4 hits in the hole. BUT, set the ANGLE to "45,45" so that all 4 hits would be at the same distance from the center line of the cylinder, if ya follow me. I mean, if you have 2 up/down and 2 to the sides, then the up/down are at the same depth. If you rotate the angle 45 degrees from there, then all 4 hits SHOULD be at the same depth, KWIM? BUT, you would either need 1 or 4 sample hits, 3 would screw it up.
Reply
  • I'm thinking the part is as you describe. It is a very difficult problem. I think there is a solution, but it will be very difficult to do and even more difficult to explain. It will no doubt take some tinkering/experimenting preferably with a "known" part so you'd know when your getting good results.

    I'm thinking it may even require iterative measurement involving lots of alignments.

    For an example:

    Measure and align to the cylinder. We'll call it Z+ for this example.
    Measure the hole knowingly shanking the probe (only want rough location).
    Align to the cylinder and use the shanked hole to fix rotation about the cylinder. Rotate to X+
    Change workplane to X+.

    There are two possible avenues of exploration from there.

    1) autocircle with 1 sample hit above or below the hole along the cylinder's vector. Edit the x value of each hit individually to get it to hit the edge. This will work if you think you can get it to hit the edge (thicker "sheet metal" like maybe .1 inch thick or more). X values could probably be calculated knowing the hole's diameter and the cylinder's diameter, but I think it would be a lot of work.

    2) Edge points (with sample hit(s)) all around the hole, then construct a circle from them. This might work better for thinner "sheet metal", but the sample hits might prove problematic as they won't be "normal" since the surface surrounding the hole isn't "normal"

    Now, you might need to go through the alignment and measurement step again to get the best results.

    Following that, lather, rinse, repeat for all the holes. Unless you can figure out a way to copy/paste with pattern or loop it (I'm not so sure that works when alignments are involved).

    I have no idea if either approach would work. I'd really need a part to play with to get an idea of that.
    About the ONLY thing I could think of would be to set the SAMPLE HITS to 4 (if you can, can you?) and set it to take 4 hits in the hole. BUT, set the ANGLE to "45,45" so that all 4 hits would be at the same distance from the center line of the cylinder, if ya follow me. I mean, if you have 2 up/down and 2 to the sides, then the up/down are at the same depth. If you rotate the angle 45 degrees from there, then all 4 hits SHOULD be at the same depth, KWIM? BUT, you would either need 1 or 4 sample hits, 3 would screw it up.
Children
No Data
© 2024 Hexagon AB and/or its subsidiaries. Privacy Policy | Cloud Services Agreement