hexagon logo
Register Sign in
  • Replies 23 replies
  • Subscribers 221 subscribers
  • Views 240 views
  • Users 0 members are here
Options
Related

Looking for opinions

The Hermit
I'm not currently at work, so I can't recall the exact model of my probe. I think I'm using the HSPX (?), with the large, magnetic base, maybe 1.75" diameter? I'm sure many, if not all of you have had this happen at least one time, if not many, many more. When you rotate, your probe catches the part, the fixture, or even the side column on the cmm and knocks the probe off. Here's my question. Do you automatically recalibrate, or do you kind of base it off the tolerance of whatever you're checking? I'm just curious how others approach this.

Thank you
  • I have an internal verification program that checks all of my tips in relation to the master tip for a quick gut check.
  • in reply to SingularitY
    So if you have, for example, 3 different probes and 8 angles, you'll verify all of them? Wouldn't that be as time consuming as just recalibrating? But beyond that, if your positions are 0.020" at MMC, would you be concerned with accuracy after a tip falls off? I'm talking about production parts, not necessarily PPAP's. Would you mind sharing your verification program?
  • well, if it is opinions you want, the internet is definitely the right place to come...

    So that I'm not working with a probe that bends or fractures, I use primarily ceramic, so it doesn't survive hitting something when rotating the head.

    If it was steel, and would survive, I would clear the results for the entire build and requalify. If it bends, not only will it be off when you use it, but it won't relate to the other probes at all. Every measurement you take between probes will be wrong.
    Are they enough "wrong" to matter with your tolerance? I couldn't tell you, but I'd requalify the entire probe build if I had a probe survive that.

    I don't requalify more than every couple months.
    I know a lot of people are going to scream I'm a heretic and want to burn me at the stake as a witch. Sorry.
    Going in to that decision is the fact the CMM is next to me, and no one uses it without my supervision. Ever.
    That doesn't mean I micromanage what they are doing, it means I know if they crash or spin the head and knock the probe off.

    I have far more problems letting them requalify, where they don't qualify the master probe first saying the sphere moved and then say no on every other qualification, so the damned rack doesn't relate between probes and it takes me an hour to prove that to them. Then I have to clear the results from the rack, which costs me time. A lot of time. We have nine probes in the racks. It is painful to qualify.

    That said, we also compare results between different machines.
    Sure, not qualifying can cost you accuracy, but three machines aren't all going to lose qual in the same vectors as each other, so when the results match, you are safe.

    If people from the floor ran the machines, or if we had a night shift or something, I'd qualify a lot more since I wouldn't know what was happening with my machines all the time.

    To your question about tolerance, going down that road means you are allowing uncertainty, due to a large tolerance. This is acceptable to me.
    If you are allowed .020, and you measure .0199 are you saying the part is "bad" and verifying with something that has the uncertainty required to report a value that close to the limit?
    If you are going down this road, you should either review EVERY output and remeasure anything close to the limit, or adjust the limit. Allow .016 instead of .020. If you get .0159, you're in. If you are over .0161, adjustments need to be made, or you need to measure with another tool for correlation.
  • I have an internal verification program that checks all of my tips in relation to the master tip for a quick gut check.

    We do the same thing. We basically just check that all the probes correlate to the Master Probe using the A0B0 probe angle on each. A quick check to make sure nothing got bent or shifted.

    I sometimes wonder if we should go a step further and check two angles for each probe (one at A0 and one at A90). So far, I have resisted the urge to get that paranoid. Checking just one angle will catch most problems.

    We do this check every morning and whenever something unfortunate happens, like when a probe is knocked out of the probe module. All the probe angles are calibrated once a month or when it is clearly needed.
  • in reply to Cris_C
    once a day we CHECK A0B0, A90B90,A90B0,A90-90,A90B180. that gives me a good idea how simple but frequently used angles are preforming, how clean the tip and sphere are and if someone dropped or wacked a tip. it only takes 1 hour but checking only A0B0 only takes 10mins which we run at the start of every shift holding ± 0.0254 x,y,z & Ø soo ya its worth it. we have 13 tips total with hundreds of angles and only calibrate once a month or as needed individually. takes 5 1/2 hours to calibrate all angles with 12 pts on slow mode....
  • in reply to SingularitY
    Yes, what is this internal verification program you speak of?
  • well, if it is opinions you want, the internet is definitely the right place to come...

    So that I'm not working with a probe that bends or fractures, I use primarily ceramic, so it doesn't survive hitting something when rotating the head.

    If it was steel, and would survive, I would clear the results for the entire build and requalify. If it bends, not only will it be off when you use it, but it won't relate to the other probes at all. Every measurement you take between probes will be wrong.
    Are they enough "wrong" to matter with your tolerance? I couldn't tell you, but I'd requalify the entire probe build if I had a probe survive that.

    I don't requalify more than every couple months.
    I know a lot of people are going to scream I'm a heretic and want to burn me at the stake as a witch. Sorry.
    Going in to that decision is the fact the CMM is next to me, and no one uses it without my supervision. Ever.
    That doesn't mean I micromanage what they are doing, it means I know if they crash or spin the head and knock the probe off.

    I have far more problems letting them requalify, where they don't qualify the master probe first saying the sphere moved and then say no on every other qualification, so the ****ed rack doesn't relate between probes and it takes me an hour to prove that to them. Then I have to clear the results from the rack, which costs me time. A lot of time. We have nine probes in the racks. It is painful to qualify.

    That said, we also compare results between different machines.
    Sure, not qualifying can cost you accuracy, but three machines aren't all going to lose qual in the same vectors as each other, so when the results match, you are safe.

    If people from the floor ran the machines, or if we had a night shift or something, I'd qualify a lot more since I wouldn't know what was happening with my machines all the time.

    To your question about tolerance, going down that road means you are allowing uncertainty, due to a large tolerance. This is acceptable to me.
    If you are allowed .020, and you measure .0199 are you saying the part is "bad" and verifying with something that has the uncertainty required to report a value that close to the limit?
    If you are going down this road, you should either review EVERY output and remeasure anything close to the limit, or adjust the limit. Allow .016 instead of .020. If you get .0159, you're in. If you are over .0161, adjustments need to be made, or you need to measure with another tool for correlation.

    As you calibrate once a month, it likely would have a much greater impact if you didn't recalibrate after dropping a probe. We calibrate per part, per shift, only calibrating the angles for that part. So though a probe may occasionally get knocked off, it's specific to the part it's being used on. I have not given my opinion, but it is that you don't "have" to recalibrate, unless it's a tight tolerance part. In twenty years, I've never actually seen a steel probe bend from a 6" or 7" drop. If it happens to fall on the floor however, that's quite a distance and would get recalibrated. If we're looking at features with a 0.020" positional tolerance, and dropping a probe makes the difference between 0.010 or 0.015, the part is still in tolerance, and I save calibration time. No foul, no harm. It will be calibrated again the next time it comes in. If it's out of tolerance, then that would require a recalibration and recheck. I'm speaking production parts, not PPAP's.


  • We do the same thing. We basically just check that all the probes correlate to the Master Probe using the A0B0 probe angle on each. A quick check to make sure nothing got bent or shifted.

    I sometimes wonder if we should go a step further and check two angles for each probe (one at A0 and one at A90). So far, I have resisted the urge to get that paranoid. Checking just one angle will catch most problems.

    We do this check every morning and whenever something unfortunate happens, like when a probe is knocked out of the probe module. All the probe angles are calibrated once a month or when it is clearly needed.

    We did that at the first shop I ever worked in. The problem was that if something happened, we had to recalibrate everything again! It's so much more efficient to calibrate only the angles you need at the time you need them.
  • in reply to The Hermit
    When a TP20 gets knocked loose, it is often no issue what-so-ever. Sometimes, even with the ceramic, it doesn't matter (if the probe is close to the table).

    The HP-S-X1H probe on the two new machines, the magnets are SOOO much stronger. The ceramic sometimes shatters without the holder even popping off the probe.

    But, no, a short drop is also a deciding factor, I've had the head up and dropped the probe 25-30 inches, which ends badly more often than the short drops.

    So, height of drop is another area where a little common sense can go a long way on determining your actual risks.
© 2024 Hexagon AB and/or its subsidiaries. Privacy Policy | Cloud Services Agreement