Inspecting a 3D part with Vision Probe

  • Filter
  • Time
  • Show
Clear All
new posts

  • Inspecting a 3D part with Vision Probe

    We have a Optiv 321 TP that we haven’t used in about 2 years. I took the vision course at Hex when be bought the unit. It was used in a cell to check 1 part. I used the tactile probe %100 because I couldn't get the vision probe to work. The part was a shiny cylinder on its side and I just couldn't get any good measurements. I’m determined the get it checking something. I’ve only been successful with using bottom light and the tp20 for measurements. No probe rack or rot table. I can only check a less than half the part which isn’t real useful.

    I attached a drawing the circled the dimensions that I want to check. Can this be down with just the vision probe? I currently have the part with the wide base laying flat on the glass so I can see the large ID in the center of the part. It has a tight tolerance so I need to check and I have not tried to check the part on its side.

    Please let me know what ya'll think.
    Attached Files

  • #2
    I don't see any circled dimensions, but that's irrelevant.

    from the looks of it, if you can get the part to be on it's side within a reasonable straigtness, then yes, those dimensions should be easy to do.

    EDIT: maybe i'm reading those dimensions wrong, how big is this thing?


    • #3
      the large OD is about 3.2in and the overall height is about 1in. Its my actual drawing so I removed some of the decimal places.

      Its just that I see other users posting pictures of parts on the vision forum and I wonder what they are actually inspecting. Like your posts on the 3D printing fixtures thread. I see you have a probe rack and your fixtures are not clear so I'm wondering how you are usiing/programming the optiv. My boss thinks i'm just missing something and don't want to do it.
      Last edited by acgarcia; 07-02-2019, 02:26 PM.


      • #4
        easily do-able.

        I'm getting ready to leave for the day but tomorrow morning i will mock something up in CAD so you can have a visual of how to set it up.

        you have to wrap your mind around the fundamental difference between inspecting in 2D and 3D, if you haven't already.

        Your print shows a cut-section at AA and that's EXACTLY what your camera will see (minus the internal dimensions, which is where the probe comes in.)

        I'm happy to help with this, but tomorrow. lol.


        • #5
          Looking forward to it.


          • #6
            My apologies.

            Today was a lot more hectic than i anticipated. I'll be back at work friday and i might have a chance to work something up.

            again, i'm sorry man.


            • #7
              no worries dude, its just one of my many projects.


              • #8
                Alright, so i threw this together real's not comprehensive as you will need to find something to keep it from rolling the other way, maybe some sort of spring loaded plunger clamp from R&R or something similar if you don't have any R&R stuff.

                but basically, if you are able to use a clear gridplate then you can just the backlight to capture all the edges EXACTLY as you see them on your blue print. Your camera should be looking at the part perpendicular to the axis of the ID cylinder.

                Generally speaking your best focal point will be when the Z vlaue of the camera matches the center point of the ID.

                For example, if you manually focus on the top of the part (the OD), and you jog the machine down 1.6 inches you should be at the best focus point for all of the outside profile. From there you can measure lines and such to capture any dimensions you would need. The standoff between the focal point and the physical camera varies depending on your camera and zoom so just be cautious when you are jogging down so that you don't hit the part with the camera.

                I hope this helps.
                Attached Files


                • #9
                  I will definitely try, thanks a million!


                  • #10
                    Anthony, I have a question.
                    IF.... that part were rotated 90° to place the 'flange' to the glass/acrylic plate, up on 3 of exact same gage blocks, can the dims be measured in Z?
                    I say only because for me? I feel the machine is just not as keen in Z (using the optics)
                    My experience on a 443 Dual Z is that probe and optics do not correlate really well in Z, and why I mention using gage blocks to set Z 'zero' is that the Hexagon correlation device is a very shiny ring which is used during cal/correlation of the 2.
                    Does that make any sense?
                    ** also by meaning that close correlation should be a .0001 or .two. Or am I dreaming? This machine (2015) was very expensive in relation to other vendors' offerings.
                    Just would love to hear your opinion(s)


                    • #11
                      The 2Z443 was my baby. I spent an unreasonable amount of time with that machine so i get what you're saying. Unfortunately, due to the surface imperfections and how cameras work, we will never get the Z-value repeatability that we will with a probe. BUT

                      let's not get confused with what i'm proposing here: I'm simply proposing that we are measuring the profile of the part using backlight, not any top/ring light, which is where the z-value inconsistencies come into play. When we backlight the part i'm absolutely confident we are measuring properly as we are TECHNICALLY just measuring x-y values.

                      As far as addressing what you're saying in laying it down and measuring those values in Z, no i would not personally do that as a simple type 1 study would prove the repeatability to be absolute garbage because what you're really doing in measuring the distance in Z is relying on the focus point of the camera. Keep in mind that the Z-Value of your surface points is simply the focal point. So if your focus parameters are off, your Z value will be also.

                      I hope that makes sense!

                      I have seen some generally acceptable correlation values (no more than a handful of microns) between the LSP and camera on the 2Z443 but it does take some tweaking.

                      That particular machine is pretty expensive but it's so versatile there's really no competition to it.


                      • #12
                        1. The Performance 443 Dual Z IS versatile!
                        2. The Ring light can make magic, but I certainly agree with you. For me? The Top light is only good to put my finger under the camera to judge my last manicure.
                        3. In XY, I did put it to silly tolerances just because my company doesn't want to pay a cal company to cert Ring Gages. ($35-40 a pop. Seems reasonable price to check our air gages.)
                        They calibrated a .80000 Master XXX Ring ± .00001. OK. As you mention.... it takes some tweaking, but I matched the cert and repeated KINDA.
                        Actually. I used opposite points X and Y. The Leitz probe is pretty good.
                        The Mycrona...uh Falcon...uh...Hexagon (who evers) Optic device is nice (ours a 3.1 X zoom) but Demon Vision software needs a more critical Illumination Cal utility. The measurement can be made happy dance by many tenths depending on the light. Pixel exercise with Glass artifact and illumination with white paper. I'm not left feeling comfortable.
                        Just my experience.
                        But Anthony? Thx for a great informative post!

                        p.s. I asked Boss if I could put the ring on our (pick one of our multiple) Zeiss Contura.... to compare. He didn't want to open a barrel of serpents!
                        Last edited by sealevel; 07-09-2019, 06:41 PM.


                        • #13
                          the top light is good if it's calibrated properly, but as you say, the light calibration could probably use some improvement. How? I'm not so sure, maybe the current calibration is as good as it gets...?

                          To be frank, you probably WOULD see a lot of difference in the gages between the Zeiss and hex machines but you have to take into account the sensor calibrations, measurement methods, measurement calculations, scan forces (if scanned), etc. Simply throwing it on there and expecting it to be exactly the same wouldn't be wise. BUT, i agree. I wouldn't want to go down that path either.


                          Related Topics