hexagon logo
Register Sign in
  • Replies 21 replies
  • Subscribers 221 subscribers
  • Views 338 views
  • Users 0 members are here
Options
Related

Reverse Engineering: Scan to CAD File

penguinzrp
Hello,

I am wondering, if it's possible, to scan a part without a CAD File and extrapolate data and measurements from the scan and export that data into a CAD File. If this is possible, how do you do that?

Thank you!
  • Yep. Probe comp works the same, except for the little detail where after the scan is completed PC-DMIS processes the ball-center-data it gathered and applies the comp to each datapoint.
    Without using CAD the comp can be fine for simple, smooth, low-rate-of-change surfaces or wildly wrong for compound complex surfaces. Even with CAD you can still have comp error if the part is too far from nominal.

    Scanning is fun and profitable, you should try it. So much data! Make beautiful full-color plots that perfectly illustrate just how bad the parts are - a picture is worth well over a million numbers in today's market.

    Also, it has this little time savings over TTP when Reverse Engineering - jobs that took me 40 or 50 hours using TTP take about 3 or 4 with scanning.

    Oh, I scan, I scan a lot. In fact, that is what I am doing right now, but this is 2-D scanning (trim details, on my special plate).

    BUT, my machine can not do anything BUT TTP (Sharpe32 controller).

    For 3D scanning, I use the "laser" on the 7-axis SI.
  • Oh, I scan, I scan a lot. In fact, that is what I am doing right now, but this is 2-D scanning (trim details, on my special plate).

    BUT, my machine can not do anything BUT TTP (Sharpe32 controller).

    For 3D scanning, I use the "laser" on the 7-axis SI.

    Ah, the power of the laser-equipped SI. I'm still a one-trick-pony CMM guy, haven't got learned on those fancy portable arm things yet.
  • Ah, the power of the laser-equipped SI. I'm still a one-trick-pony CMM guy, haven't got learned on those fancy portable arm things yet.

    Them fancy, new-fangled, portable thing-a-majiggers is the bee's knees, if'n yer usin' it on sumpin' that ain't PC-DMIS.
  • Them fancy, new-fangled, portable thing-a-majiggers is the bee's knees, if'n yer usin' it on sumpin' that ain't PC-DMIS.

    Actually, I am using it with Pcdmis (2012). Other than the random, frequent software crashes (like, without even having a program open, moving the arm and Pcdmis crashed), it's pretty nice.
  • Actually, I am using it with Pcdmis (2012). Other than the random, frequent software crashes (like, without even having a program open, moving the arm and Pcdmis crashed), it's pretty nice.

    That's at least half of what I am referring to. Slight smile

    PC-DMIS just doesn't handle memory very well. To be fair, it is much better now (2012) than it used to be.
  • One of the many reason the CMM is still on V3.7

    As for the memory, it's always been a problem with Pcdmis. And, of course, there are work-arounds. Most common one for me and scanning, when scanning something BIG and needing a "tight" point cloud is to do it in 'sections'. Scan a section, filter it and so forth, the save the program (sec-1), then save it as "sec-2", close & open sec-2, then change the filter of the existin cloud to 5mm so it is just a ghost of a cloud, then scan sec-2, then save, then save-as sec-3, close & open sec-3, re-filter, and so on. Pretty much, keep the cloud under a million points.
  • One of the many reason the CMM is still on V3.7

    As for the memory, it's always been a problem with Pcdmis. And, of course, there are work-arounds. Most common one for me and scanning, when scanning something BIG and needing a "tight" point cloud is to do it in 'sections'. Scan a section, filter it and so forth, the save the program (sec-1), then save it as "sec-2", close & open sec-2, then change the filter of the existin cloud to 5mm so it is just a ghost of a cloud, then scan sec-2, then save, then save-as sec-3, close & open sec-3, re-filter, and so on. Pretty much, keep the cloud under a million points.

    Yeah. And then with PolyWorks, I can scan a single stripe of a million points without blinking. Working with 20+ million points, slow, but solid. Of course, PW also has its own quirks and weaknesses.
  • Geomagic (what little I worked with it with the arm) is pretty good as well, HOWEVER, Geo drops 'excessive' or 'extra' points as it scans, so, if you were using Poly or Pcdmis, and just scanned non-stop for 15 minutes, without moving the arm, you would have BILLIONS (or so Slight smile ) points, in Geo, you wouldn't have but the 1 strip of points.
  • Geomagic (what little I worked with it with the arm) is pretty good as well, HOWEVER, Geo drops 'excessive' or 'extra' points as it scans, so, if you were using Poly or Pcdmis, and just scanned non-stop for 15 minutes, without moving the arm, you would have BILLIONS (or so Slight smile ) points, in Geo, you wouldn't have but the 1 strip of points.

    PolyWorks has some cool real-time features also. Meshing, filtering, scan-pass to scan-pass checking. Pretty cool stuff.

    I've heard that Geomagic/RapidForm is the boss of creating solid CAD from scans.

    My company still hasn't decided whether to go for IMEdit module, a CATIA module, or Geomagic/RapidForm for the last step of actually reverse engineering from pointcloud to solid CAD.
  • I used Geo for quite some time, back when we were using a Konica-Minolta scanning camera (80 pounds, took 'pictures' 640x480 points), and it had to be moved, then each 'picture' overlaid (by hand) to the previous, then "automatic" 'mesh' then repeat. Took freaking FOREVER (as in, a small detail would take 2-3 hours). Then, we got the arm, and the Geo work got moved to Cad since I really didn't have time for it, but in the 'move' I used it w/ geo to get him started on the use of Geo, and it was pretty slick. Alignments were kind of a bear, but we didn't have the 'inspect' portion of it (still don't). But, I just scan and dump XYZ files from the cloud and it's their problem now. NOW they are seeing WHY 'sharp' corners are really impossible. That has to be done in a "real" cad package, not a 'data gathering' package (like Geo & Poly really are). Yes, you can 'get close', but not really.
© 2024 Hexagon AB and/or its subsidiaries. Privacy Policy | Cloud Services Agreement