Announcement

Collapse
No announcement yet.

Measuring Length with Stepper

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts


  • Measuring Length with Stepper

    I am wanting to measure the length of a cylindrical part after it has been trimmed. Parts are cylindrical in shape, and wall thickness is approx. .010" (think cartridge cases). Thoughts were originally to use laser, ultrasonic, or prox. sensor, but end-on I don't know if the exposed material will be enough to get an accurate measurement (+/- .001" is desirable). Other idea is to do like CNC machines with tool-length scanners, and use an AD Fork Sensor mounted to a liner slide with stepper, sense when the sensor beam is broken, then retract until the beam "reappears", then read counts from stepper to determine position (length). From this I can output a sorting function based in "in-spec", "long", short".

    Any ideas or suggestions on ladder?


  • #2
    If it was me, I think I'd do this with vision.

    Comment



    • #3
      I'm presuming you mean like my second idea using the fork-sensor?

      Comment



      • #4
        I believe that CG is speaking of a vision system like Datalogic, Cognex etc makes. They uses an actual camera to look at your part, and then uses software to calculate dimensions etc based on the images it acquires.
        Circumstances don't determine who we are, they only reveal it.
        Jason Wolthuis

        Comment



        • #5
          Vision System sounds overly complicated for what I need to measure. Right now we are hand-measuring with calipers as a QC check which works perfectly fine. I envision the PLC doing the QC check as parts are completed, then sorting them. I was thinking something simple like a tool-length setter on a VMC (Vertical Machining Center). HAAS uses the Renishaw Probe, other companies use simple continuity as the "limit switch" when the tool touches a "pad". I did review the video on AD's website about "homing" stepper motors, and this is exactly what I am looking for (option 4), just need to be able to write the "home" to a tag that I can then manipulate.

          Comment



          • #6
            All the other mechanics sounds more complicated to me, honestly. Vision is actually fairly easy (though the hardware can be expensive). Are you trying to do 100% inspection real-time with a manufacturing operation (in which case it would need to be fast)? Or just sample a single part out of every every 100 or so to make sure the process is on track? Or measure all parts but in offline batches, where throughput isn't as critical?

            Comment



            • #7
              I didn't use a fancy linear slide, but I did use a glass scale employed by DROs in CNC machines for this project. Knowing what I know now, I'd use the uC32 MCU from this other project as it has a quadrature encoder built in and fast enough processor I doubt it would skip any measurements. The Arduino's slow clock means you can move the slide fast enough the cpu won't see the quadtrature output change.

              Given I have played with stepper motors and measuring, I see the big problem you are facing being resolution. For the TPS test bench, I have a 1:3 gear reduction from stepper to the TPS input shaft. One step is almost a degree of rotation. If you want 0.001" resolution, I am seeing some sadness trying to make a leadscrew work reliably unless everything is about twice the size one would think you need. Making your "zero" reliable might prove somewhat maddening if the leadscrew has anything to say about it.

              I hate how easily stepper motors skip and made the code in my TPS bench very sensitive to catch any chance where the stepper skipped. If it thinks there's any chance of a skip, it trashes that test and you have to retest. I already thought how to redo the first bench using an air piston, force sensor, and stepper to turn the sprockets. Oh, and then add an inkjet to print the value on the belt itself...

              Chris

              Comment



              • #8
                An option to overcome the stepper issue would be something along the line of the Schneider Lexium M Drives with HMT, cost effective and no skipped steps.

                Comment



                • #9
                  Originally posted by cacycleworks View Post
                  I didn't use a fancy linear slide, but I did use a glass scale employed by DROs in CNC machines for this project. Knowing what I know now, I'd use the uC32 MCU from this other project as it has a quadrature encoder built in and fast enough processor I doubt it would skip any measurements. The Arduino's slow clock means you can move the slide fast enough the cpu won't see the quadtrature output change.

                  Given I have played with stepper motors and measuring, I see the big problem you are facing being resolution. For the TPS test bench, I have a 1:3 gear reduction from stepper to the TPS input shaft. One step is almost a degree of rotation. If you want 0.001" resolution, I am seeing some sadness trying to make a leadscrew work reliably unless everything is about twice the size one would think you need. Making your "zero" reliable might prove somewhat maddening if the leadscrew has anything to say about it.

                  I hate how easily stepper motors skip and made the code in my TPS bench very sensitive to catch any chance where the stepper skipped. If it thinks there's any chance of a skip, it trashes that test and you have to retest. I already thought how to redo the first bench using an air piston, force sensor, and stepper to turn the sprockets. Oh, and then add an inkjet to print the value on the belt itself...

                  Chris
                  Awesome -- you're actually printing TPS reports!! Will they have cover sheets?

                  Comment



                  • #10
                    I just did something like that, pulling labels off of a spool, did a register read when the sensor breaks, and another when it comes back, and subtract the smaller from the larger. I don't know about holding .001", but it does work.
                    anything is possible with the right sized hammer.....

                    Comment



                    • #11
                      Don't know if it would work for you but we built an inspection table for a customer using the OPT2001 laser distance sensor for ADC. It has been in service for 2 years. We had a service call on last month the measurement would very by 0.002" it was caused by wear in the table top. We had to surface grind the table back flat. The sensors have a resolution of 0.000787 with a linearity error of 0.000079". the down fall is the measuring of 3.15". but it worked very well for the inspection table, we mounted two sensors in a large thick piece of steel with a hole in the center that was surface ground to flat. We drilled location pins in the steel to stop the part, the sensors look up at the part thru the hole. We used a Do More PLC with a C More HMI with recipes for the different part numbers and the allowable difference. Then we compare the recipe value to the actual in the PLC and displayed the measurements on the HMI with Good or Bad. We found the sensors to be very accurate and repeatable.

                      JW

                      Comment

                      Working...
                      X