Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

White Light scanning

  1. Jul 17, 2010 #1

    Ranger Mike

    User Avatar
    Science Advisor
    Gold Member

    How does white light scanning work?
    I know the system is used in industry to measure parts for dimensions like diameter, size.

    I know there is a white light source. There is a camera mounted at a known angle relative to the white light source. I was told the signal is obtained from the center of energy of the white light source which means the system can measure features smaller than the size of t he white light dot. Also the white light source can be targeted at surfaces up to 85 degrees form perpendicular and still yield credible data.

    does anyone have more insight regarding this ?
     
    Last edited: Jul 17, 2010
  2. jcsd
  3. Jul 17, 2010 #2

    minger

    User Avatar
    Science Advisor

    Center of energy? White light scanning is really just glorified pictures. The idea is that if you set up a part in a fixture of known dimension, then you can use stereo cameras to triangulate position of the part inside of the fixture.

    White light scanning is pretty cool and it has some really nice applications. Firstly, the scanning is typically much faster than a CMM (depending on what you're doing exactly). Secondly, some features are much easier to measure (think of the profile of an airfoil). Lastly, the data is output in such a way that it can be read into a modeler.

    The last point is particularly neat as you can overlay the geometry and see where you have thinning or out of tolerancing. This can greatly help with manufacturing.

    The downside is the accuracy. We recently spend some time talking to some white light guys. We really couldn't get the precision that we needed. IIRC, we had a radial diffuser, and no matter what, we couldn't get better than +-0.002" or 0.004. While that sounds nice, we typically have parts toleranced to +-0.0005, so the precision was unacceptable.

    Aside from that, there is significant costs associated not only with the hardware, but the post-processing software.
     
  4. Jul 18, 2010 #3

    Ranger Mike

    User Avatar
    Science Advisor
    Gold Member

    Thanks for the reply..
    The Germans have taken it to the next level. They developed a cmm with white light sensor mounted ion two axis rotating head. They claim volumetric accuracy of +/- 8 micron certified with the same ISO 10360 standard used to certify CMMs. I was wondering about the camera and how it gathers its data. The stand off distance and triangulation is pretty straight forward.
    you are correct about speed and ability to generate a cloud of points that cam be exported to software that can compare the data to the CAD model.,,
     
  5. Jul 20, 2010 #4

    minger

    User Avatar
    Science Advisor

    Those crazy Germans; +/- 8 micron is pretty impressive for a system like that.
     
  6. Jul 20, 2010 #5

    Ranger Mike

    User Avatar
    Science Advisor
    Gold Member

    a si understand it, the white light scan head has a light source and two cameras mounted at known angles from the source. from what i read, the cameras pick up the XYZ coordinate of the center of the white light dot. since these cameras ate at different locations from the white light source we have triangulation to determine the exact location of the white light dot. the whole scan probe mounted on a two axis motorized probe head which is mounted on XY and Z precision guideways..these have hi resolution scales and read heads for positioning the probe head. it is old cmm techonolgy with new age sensor..trick huh??
    i am trying to find out exactly how the camera determines the white light spots center of energy??
     
  7. Jul 21, 2010 #6

    minger

    User Avatar
    Science Advisor

    From what I understand, it doesn't necessarily read a single point in space per "snapshot", it reads the entire "viewable" surface. In order to do this, you must prep the part with markers. You typically place these markers (they are just like clay dots) on features or corners.

    When the camera scans the image, it uses the markers to get the orientation of the park. It then uses the image from that camera, and a secondary (or tertiary) camera to locate the depth of each individual point in the dataset it just created.

    What you're describing sounds a little more like laser scanning.
     
  8. Jul 21, 2010 #7

    Ranger Mike

    User Avatar
    Science Advisor
    Gold Member

    close...the machine is calibrated prior to taking measurements. a datum sphere is used to this. the part to be inspected is placed on the table. to write a part program ( used to automatically measure the part ) the operator must joy stick the white light probe over the production part and manually take measurement of a plane. lien and a point to set up a part coordinate system. ( the machine was moving in machine coordinates before this.) once this is done the operator can import a CAD model of the production part and any data points take on the production part will be compared to the CAD Nominal points and we can see manufacturing trends from this. ( drift out of tolerance, etc..)
    so markers no longer are required. I think the camera picks up the white light dot and reads the brightest area which would be the center regardless of point shape, ( ellipse, oval) so the white light can scan at very steep angles to normal and get good results..
    i am trying to find info for a graph of this center of brightness ( energy) ..should be bell curve distribution i think
     
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook