Calibration question

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

  • Calibration question

    Hi at all.
    I'm italian, so, sorry for my english
    I have some question about the calibration.
    I have a tip, Ø0.5mm, witch which i have made some programs that use 0.5mm/s of touch speed and other that use 0.3mm/s. Now, to calibrate this 2 different touch speeds i will use, in the program to calibrate tips, 2 autocalibration code, with 2 different parameters that contains the 2 different touch speeds. Is it possible to calibrate the same tip with different in the same program?
    I mean, will PC-Dmis will consider that i calibrated the same tip with 2 different parameters or it will consider only the last calibration i've made?

    If the answer is no, the only solution is to create two different name of the same tip but with two touch speeds, am i right?

    Last question. To choose the right MASTER PROBE, which one is better between a ball Ø5mm long about 20mm and a ball Ø1mm long about 20mm. The Ø1mm ball is the tip that i use for most of my programs

    I hope I explained myself better.

  • #2
    Originally posted by andrea_ABAZIA View Post
    Hi at all.
    I'm italian, so, sorry for my english
    I have some question about the calibration.
    I have a tip, Ø0.5mm, witch which i have made some programs that use 0.5mm/s of touch speed and other that use 0.3mm/s. Now, to calibrate this 2 different touch speeds i will use, in the program to calibrate tips, 2 autocalibration code, with 2 different parameters that contains the 2 different touch speeds. Is it possible to calibrate the same tip with different in the same program?
    I mean, will PC-Dmis will consider that i calibrated the same tip with 2 different parameters or it will consider only the last calibration i've made?

    If the answer is no, the only solution is to create two different name of the same tip but with two touch speeds, am i right?

    Last question. To choose the right MASTER PROBE, which one is better between a ball Ø5mm long about 20mm and a ball Ø1mm long about 20mm. The Ø1mm ball is the tip that i use for most of my programs

    I hope I explained myself better.

    It will only remember the latest calibration - having two separate probe files (names) is an okay solution, but could you not just use 0.4mm/s for both? Also that's a pretty slow touch speed. Typically 2mm/sec to 5mm/sec is normal.

    For master probe the Ø5 x 20mm long would be best.
    Automettech - Automated Metrology Technology

    Comment


    • #3
      Originally posted by NinjaBadger View Post


      It will only remember the latest calibration - having two separate probe files (names) is an okay solution, but could you not just use 0.4mm/s for both? Also that's a pretty slow touch speed. Typically 2mm/sec to 5mm/sec is normal.

      For master probe the Ø5 x 20mm long would be best.
      Thank you for your response.
      Up to a week ago, i was using 3mm/s for all my tips. Then, after a little problem with measuring a little diameter, Hexagon's assistance told me to use (especially with ball like Ø0.5mm ) touch speed slower than 3mm/s. Now you're saying me the opposite What i have to do?

      Comment


      • NinjaBadger
        NinjaBadger commented
        Editing a comment
        Sorry I wasn't clear - I wasn't saying it was wrong I was saying it was unusually slow and was wondering what the reason for it was.

        If it works, it works! And a tiny diameter with a tiny tip then yes, it might well be the best approach.

      • louisd
        louisd commented
        Editing a comment
        That small of a probe stylus mandates a slower touchspeed.
        The tungsten carbide shaft of a 0.5mm stylus is so delicate (0.3mm)... the frequency of failure would be very high at any touchspeed over 1mm/sec. in my opinion.
        I run our 0.3mm and 0.5mm probes at 0.5mm/sec max touchspeed.

      • derekvegeta
        derekvegeta commented
        Editing a comment
        The spec given in Hexagon's 101 and 201 courses (as of 2019) were .8 mm/sec touch speeds for any probe with less than 2mm diameter.

        In theory the slower the touch speed, the more accurate the result (as long as the calibration speed matches the program's touch speed): However the difference between .8 mm/sec and anything slower MAY not be significant enough to sacrifice run time.

    • #4
      andrea_ABAZIA,
      I agree with Ninja, use 0.4 mm/s touch speed in both programs for simplicity. I believe that Hexagon suggests a slow touch speed for a small probe because they are very fragile, not because of accuracy concerns. You could always the run the calibration with the two different speed settings and compare the results. My guess is there will not be any noticeable difference.

      Comment


      • #5
        Ok, now i understand.
        What about the prehit/retraction, does i infulence the result? I mean, with 0.5mm/s of touch speed, a prehit of 0.5mm is ok? Can i use 0.3mm?
        Normally i use 3mm of prehit, but in some case, in small spaces i have to use 0.5,0.3 or even 0.1mm of prehit.

        Comment


        • RanDawgg
          RanDawgg commented
          Editing a comment
          Prehit/retraction won't effect the results at all. Unless the prehit is so small that the CMM can't travel in the appropriate vector before contacting the feature, but .3 should be fine. I would be nervous about 0.1mm, but you can test it yourself by measuring the same feature with different prehits, and look for variation.

        • AndersI
          AndersI commented
          Editing a comment
          Prehit in combination with measuring speed may affect the result, if the distance is too short for the machine to accelerate to the set measuring speed.

      • #6
        What type of CMM are you using ? Some CMM'S are really effected by prehit distances.

        Comment


        • louisd
          louisd commented
          Editing a comment
          Yes CMM controllers might be less accurate or slower to effectively locate to your coordinates in time to prevent crash. A trick when using a really tight prehit/retract value, is to set a move/point close to the intended feature (within 3mm), immediately prior to taking the tighter prehit/retract hits.

          Also, what probe sensor are you using?
          You can get away with a smaller prehit/retract on TP200's as they are piezo-electric sensors.
          TP20's not so much, they need defined travel and duration in order to open the circuit repeatably.
          For scanning probes: (SP25, HP-S-X, or LSP) I feel a more prudent prehit/retract is also necessary.
          Last edited by louisd; 11-13-2019, 01:06 PM.

      • #7
        No, it has nothing to do with accuracy either. Prehit is how far away from the hit point the tip starts from. For example, if you were just measuring a surface point with a prehit of 10 mm, then the probe would drive to a spot 10 mm above the point, then drive in at your touch speed until it hit the surface. Then it would retract back your retract value. You want your prehit big enough so that it doesn't accidentally hit the part before you want it to, but not so big that it takes a long time to drive there. If your prehit was 10 mm with a touch speed of 0.5 mm/s, it would take 20 seconds to drive there. That would add a lot of unnecessary time to your program. Make it big enough so that you aren't getting false touches from starting too close to your part, but not much bigger than that.

        Comment


        • louisd
          louisd commented
          Editing a comment
          I promise you it is relevant. If your CMM's "Positioning Accuracy" and "Probing Accuracy" are less capable than your prehit/retract distance is defined, it will certainly have an adverse impact. If you are trying to measure a feature with a 0.010" prehit and retract, and your probing/positioning accuracies are only capable of 0.03", and you run at full movespeed... what do you expect will happen? You literally have a 1 in 3 chance of a crash.

      • #8
        Originally posted by Schlag View Post
        What type of CMM are you using ? Some CMM'S are really effected by prehit distances.
        I run a DEA GLOBAL Performance with SP25M probes, but i don't need too much scansions, it's a rare case.

        Sometimes i work with little spaces, and 0.1mm of prefit it's the only solution.
        By the way, i will do some tests and keep up to date this post with my results.

        Comment


        • #9
          One trick I have used in the past when measuring is small spaces, is to change the target vector (this is when using auto vector points).

          Imagine a 1mm wide groove with a 0.5mm tip - typically you would have to use <0.5 prehit (some allowance for variation in the location of the groove)

          Now obviously you won't be probing normal to the surface - but that is just the driving vector - the Theo vector remains normal to the surface so probe comp happens along this vector.

          The thing you have to watch with this method is that if there's plus material on the part, you will probe at a higher Z value then defined, and if there's minus material on the part you will hit lower in z, but it can be useful in the right situation.

          Automettech - Automated Metrology Technology

          Comment


          • andrea_ABAZIA
            andrea_ABAZIA commented
            Editing a comment
            Sorry, maybe is for my english.. But i think i don't understand the solution :P
            You are saying that i have to change the surface vector, but what changes should I make?

          • NinjaBadger
            NinjaBadger commented
            Editing a comment
            I'm not saying you have to do anything.

            I'm saying you could change the target vectors to approach the surface at an angle.

            Auto points have three sets of values

            ACTUAL - What is actually measured
            THEO - The theoretical values
            TARGET - The target values (this defines where the machine actually drives to), usually their the same as the theos, but you can adjust them in certain situations.

            I've got to head out now but I'll try post an example later.

        • #10
          LouisD I get that, but I'm talking about measurement accuracy, not a crash. If you pick a prehit distance that is far enough to not risk a crash, your accuracy will not suffer if you arbitrarily double that distance for whatever reason.

          Comment


          • NinjaBadger
            NinjaBadger commented
            Editing a comment
            Not quite true. Your prehit needs to be big enough for the CMM to get up to speed it used during probe calibration.

        • #11
          I've just made some test using my calibration sphere (is it a good idea?). I've compared the results of location (X,Y,Z) and diameter of the spere with SP25M and ruby ball Ø5mm, no scanning mode, 32 hits on 6 levels.
          These are my conditions and my results:
          1. ​​​​​​​movespeed=2mm/s - prehit/retraction=2mm - results X=0.000mm Y=0.000mm Z=0.000 D=15.875;
          2. movespeed=2mm/s - prehit/retraction=0.2mm - results X=0.000mm Y=0.000mm Z=0.000 D=15.875;
          3. movespeed=5mm/s - prehit/retraction=2mm - results X=0.000mm Y=0.000mm Z=0.000 D=15.875;
          4. movespeed=5mm/s - prehit/retraction=0.2mm - results X=0.000mm Y=0.000mm Z=0.000 D=15.875;
          5. movespeed=1mm/s - prehit/retraction=2mm - results X=0.000mm Y=0.000mm Z=0.000 D=15.875;
          6. movespeed=1mm/s - prehit/retraction=0.5mm - results X=0.000mm Y=0.000mm Z=0.000 D=15.875;
          7. movespeed=1mm/s - prehit/retraction=0.2mm - results X=0.000mm Y=0.000mm Z=0.000 D=15.875;
          8. movespeed=1mm/s - prehit/retraction=0.08mm - results X=0.000mm Y=0.000mm Z=0.000 D=15.875.
          What do you think about my results?

          Comment


          • louisd
            louisd commented
            Editing a comment
            How did you execute said tests?
            If you are showing zero variation, I would be more concerned that your test affirmed nothing. You should be testing until you see variation, or increase your resolution of the test (displayprecision=4)

          • NinjaBadger
            NinjaBadger commented
            Editing a comment
            Something looks fishy. Not even 0.001 variation anywhere!

            Were you re-calibrating before these runs or just changing touch-speed/prehit etc?

            Are you sure the dimension is marked?

        • #12
          I don't know why, maybe i was drunk when i wrote these numbers

          However i executed the same program with 5 zero and i calibrated my tip with movespeed = 2mm/s and prehit/retraction=2mm/s.

          These are my results:
          1. ​​​​​​movespeed=2mm/s - prehit/retraction=2mm - results: X=-0.00038mm Y= 0.00010mm Z=0.00045mm D=15.87289mm;
          2. movespeed=2mm/s - prehit/retraction=0.2mm - results: X=-0.00056mm Y=-0.00011mm Z=0.00038mm D=15.87276mm;
          3. movespeed=5mm/s - prehit/retraction=2mm - results: X=-0.00025mm Y=-0.00002mm Z=0.00046mm D=15.87265mm;
          4. movespeed=5mm/s - prehit/retraction=0.2mm - results: X= 0.00002mm Y= 0.00002mm Z=0.00054mm D=15.87233mm;
          5. movespeed=1mm/s - prehit/retraction=2mm - results: X=-0.00009mm Y=-0.00002mm Z=0.00041mm D=15.87277mm;
          6. movespeed=1mm/s - prehit/retraction=0.5mm - results: X= 0.00001mm Y=-0.00008mm Z=0.00043mm D=15.87283mm;
          7. movespeed=1mm/s - prehit/retraction=0.2mm - results: X=-0.00002mm Y= 0.00001mm Z=0.00052mm D=15.87260mm;
          8. movespeed=1mm/s - prehit/retraction=0.08mm - results: X= 0.00031mm Y= 0.00000mm Z=0.00051mm D=15.87225mm.

          Comment


          • #13
            That's much better! Great job.
            When you start looking at those results, you can see it's relatively inconclusive. #4 is almost as good as #7, #5 is almost as good as #6, in regards to coordinate location, but Diameter varies a lot as well... This is because you are trying to discern (at the moment) unconfirmed noise. There are other variables here which are forcing you to hit the center 2 rings of the bullseye, and not just the bullseye.

            To increase the accuracy, to get down to just the center ring of the bullseye, you would have to repeat this study many times, and assess how repeatable each test method is. I would develop a method to "score" the results. Determine your "Gold standard" values (X= 0, Y=0, Z=0, D= xx.xxxxxMM, RN= x.xxxxx) then tabulate the deltas of each measured value from the gold standard, maybe add them up or stddev them...
            Once you find what method is repeating the best, and what method is repeating the worst, you can hopefully make changes to your method, to increase the repeatability.

            -Maybe you need to calibrate the sphere with the movespeed matching your best method...
            -Maybe touchspeed has a much higher variable that you need to assess...
            -Maybe you need to assess and confirm the cal sphere diameter with higher accuracy...
            -Maybe you need to repeat the entire assessment with the cal sphere calibrated to match each of the defined variables (movespeed/touchspeed)...
            -Maybe your boss taps you on the shoulder and says the outer two rings of the bullseye is good enough, and to get back to work, lol.
            -Maybe you need to track the changing temperatures in the room, or how/when your heat/air conditioning cycles, or the direction in which your conditioned air flows across the CMM towards the return air vent.
            Last edited by louisd; 12-19-2019, 10:58 AM.

            Comment

            Related Topics

            Collapse

            Working...
            X