Announcement

Collapse
No announcement yet.

Autocolliminator - How Is It Used To Check Deviation From A Plane

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

  • #31
    Originally posted by J Tiers View Post
    I do not see how it assumes anything. We already discussed the use of the "Repeat-O-Meter" to sweep the surface and detect any local changes. Between the two, the plate can be well checked. Either alone has issues.

    The autocollimator typically has no sensitivity to step changes, but good sensitivity to angles. The Repeat-O-Meter has poor sensitivity to small but cumulative changes, but good sensitivity to step changes.. The autocollimator is typically used to check a grid of spots. The Repeat-O-Meter inherently checks a continuous strip of spots (it does have a certain averaging effect due to the size of the "movable foot").
    Thank you JT for the comment (saved me typing.)

    Paul, It really doesn't matter if you like it or not.

    The two tools are used to characterize different qualities of a flat (or other defined surface)
    They each need the other to expedite the task given to the technician. If there was a better way, there have been a lot of really smart people that would have mentioned it before we came along.

    An interferometer could readily map the profile of any surface you might like, But the cost would be high to bring that technology to the stone. Not many are willing to bring the stone to the qualification lab.

    Just a note:

    I keep in my own lab, a K&E autocollimator /alignment telescope along with a matching infinity target. Also, a Davidson D-305 6 inch aperture Fizeau interferometer.
    I do not re-qualify or correct reference flats, But have produced some very fine TWD plano optics.
    As mentioned, There is an inspection grade stone of good size in the shop as well. Not that these "possession" add materially to this discussion. Just that I am versed in their workings.

    A side note, For optical work, the analog to the repeat-o-meter is the sphere-o-meter. The indicator is placed centrally. Check it out! They work for plano surfaces as well as sagitta.
    Last edited by CalM; 05-27-2020, 12:27 AM.

    Comment


    • #32
      Well, I can imagine a plate with a completely horrible set of errors that will show absolutely no indication from perfect with either and perhaps with both of the instruments. But once I go to that trouble you will just say that "normally" the plates do not wear that way. You will essentially be saying that I can not have, I can not postulate that kind of wear because it never happens. In short, you will then invoke ASSUMPTIONS about the plate.

      Now, I can understand how a technician, who is pressed for time, will not want to "waste" time on unnecessary checks. I may even use the techniques myself. But I don't have to like it.

      But the fact of the matter is, there ARE assumptions built into the use of the techniques that I have seen in the internet videos.



      Originally posted by J Tiers View Post
      I do not see how it assumes anything. We already discussed the use of the "Repeat-O-Meter" to sweep the surface and detect any local changes. Between the two, the plate can be well checked. Either alone has issues.

      The autocollimator typically has no sensitivity to step changes, but good sensitivity to angles. The Repeat-O-Meter has poor sensitivity to small but cumulative changes, but good sensitivity to step changes.. The autocollimator is typically used to check a grid of spots. The Repeat-O-Meter inherently checks a continuous strip of spots (it does have a certain averaging effect due to the size of the "movable foot").
      Last edited by Paul Alciatore; 05-27-2020, 01:23 AM.
      Paul A.
      SE Texas

      And if you look REAL close at an analog signal,
      You will find that it has discrete steps.

      Comment


      • #33
        Cal, A lot of really smart people have been wrong before and they will be wrong again. A lot of really smart people said that the earth was flat. A lot of really smart people claimed that there was an ether that was the medium for light to travel in. A lot of really smart people said that there was no limit to how fast particles or energy could travel. And a lot of the really smart people who actually worked to develop the atomic bomb were actually betting that it would not work before that first successful test.

        Fact is, a lot of the really major advances in science and technology have happened when someone questioned the opinions of "really smart people".

        I am not saying that the stock answer is wrong. I am just saying that there is no harm in questioning things. And when I dig into this subject, I find some really big holes in what the "really smart people" are doing. In fact, at least one source that I found stated point blank that the way to plot the errors in a surface with an autocollimator was to take readings that were spaced the distance between the feet of the target mirror. They did not say that shortcuts could be made. Just one length of the mirror's base after another. So perhaps there are some people who are really smarter than "really smart people". And it seems to me that if the technique in that source were used, then the autocollimator would produce an accurate plot of the errors of a surface plate in terms of actual distance. The repeat-o-meter is not really necessary in that case.

        By the way, I do not need to look up the sphere-o-meter. I have used them and I do understand how they work - equations and all. I am not so sure that it compares to the repeat-o-meter. They both use three points to establish a reference plane and they both use a fourth point where a measurement is taken, but In one case the fourth, measurement point is surrounded by the three points in a very symmetrical manner while in the other that fourth point is completely outside of the three reference points. To my mind they are designed for and actually do completely different things. It would be hard to measure the radius of curvature with a repeat-o-meter and likewise hard to measure the errors of a flat surface with the sphere-o-meter.

        For those who may not be familiar with a sphere-o-meter:

        https://en.wikipedia.org/wiki/Spherometer



        Originally posted by CalM View Post

        Thank you JT for the comment (saved me typing.)

        Paul, It really doesn't matter if you like it or not.

        The two tools are used to characterize different qualities of a flat (or other defined surface)
        They each need the other to expedite the task given to the technician. If there was a better way, there have been a lot of really smart people that would have mentioned it before we came along.

        An interferometer could readily map the profile of any surface you might like, But the cost would be high to bring that technology to the stone. Not many are willing to bring the stone to the qualification lab.

        Just a note:

        I keep in my own lab, a K&E autocollimator /alignment telescope along with a matching infinity target. Also, a Davidson D-305 6 inch aperture Fizeau interferometer.
        I do not re-qualify or correct reference flats, But have produced some very fine TWD plano optics.
        As mentioned, There is an inspection grade stone of good size in the shop as well. Not that these "possession" add materially to this discussion. Just that I am versed in their workings.

        A side note, For optical work, the analog to the repeat-o-meter is the sphere-o-meter. The indicator is placed centrally. Check it out! They work for plano surfaces as well as sagitta.
        Paul A.
        SE Texas

        And if you look REAL close at an analog signal,
        You will find that it has discrete steps.

        Comment


        • #34
          Your mind is made , no sense confusing the issue with facts.

          The youtube jocky is the smart one, but not the people who actually do the stuff every day.

          I really can't (won't) add anything more.

          Comment


          • #35
            Originally posted by Paul Alciatore View Post
            Well, I can imagine a plate with a completely horrible set of errors that will show absolutely no indication from perfect with either and perhaps with both of the instruments. But once I go to that trouble you will just say that "normally" the plates do not wear that way. You will essentially be saying that I can not have, I can not postulate that kind of wear because it never happens. In short, you will then invoke ASSUMPTIONS about the plate.

            Now, I can understand how a technician, who is pressed for time, will not want to "waste" time on unnecessary checks. I may even use the techniques myself. But I don't have to like it.

            But the fact of the matter is, there ARE assumptions built into the use of the techniques that I have seen in the internet videos.


            Go ahead and describe it.

            It will likely have to be something that consists of just what I ALREADY SAID is not going to be detectable, which is a series of small steps, each of which is not measurable with the repeat-o-meter (R-O-M). If there were any angles, then they would need to be smaller than the autocollimator (AC) resolution.

            I make NO assumptions about the wear pattern. To my thinking, ANY wear pattern is fair to assume.

            What I DO question, is whether the resulting errors of the plate are significant. If the error is not detectable as a step-change with the R-O-M, and is not detectable with the AC, then either A) you have the wrong instruments, OR B) you are OK that way, and they will detect anything that is of a concern to you.

            Then, if they do not detect the issue, it is too small to be of concern. If you WERE worried about something smaller, then refer back to "A)"

            There has never been a perfect granite flat ever made, most likely, and every gauge block on earth is actually defective. They just are not bad enough to be a problem.

            EDIT:

            One such pattern would be a series of steps in the surface, spaced such that the R-O-M can "see" only one at a time, each of a magnitude that is not reliably detected, and having tops that are exactly parallel to the original surface. That is a very special case, but so be it. The surface could be a set of "stairsteps", or it might be such that it makes a sort of "potato chip" (saddle shape, or twisted saddle, perhaps.... I don't know if there is a "standard chip").

            So, the parallelism means that the A-C might not detect them, unless the base was across a step, and we defined them as being too small for the R-O-M to reliably detect. So only the A-C may be able to find them.

            Basically ANY placement of the mirror is countered by the statement "but they were spaced so you missed all of them", however, I suspect that random placements would increase the possibility of finding something, assuming it is within the sensitivity of the A-C.

            That error type would not be detected, possibly in many cases, or even in all.

            The question is NOT whether that wear is possible, but rather if the total change across the surface is too large to be acceptable. If so, then the wrong R-O-M was used, you need one with more sensitivity.

            That is a different problem entirely. There is also the issue of how bif=g the plate would have to be in order to have that happen. Clearly not with a 9" x 12" plate.....more like a several foot wide by several more long plate.

            That problem exists with any error size.... you have to know that the instruments you use are capable of finding any error you cannot tolerate.

            Otherwise you find yourself in the predicament of the person who once asserted here on the forum that he could just use a magnifying glass on the dial of the caliper, to get better resolution. He forgot, or ignored, the fact that the mechanism itself may not (probably does not) repeat well enough to be reproducible at that level. If it did, it would be made to have that level of accuracy and resolution.
            Last edited by J Tiers; 05-27-2020, 03:32 PM.
            CNC machines only go through the motions

            Comment


            • #36
              I would suggest that there are two different operations being discussed, not one.

              1) When re-lapping it is not neccessary to quantify every single deviation from flat. You are trying to reduce them and will re-measure more than once in the proceedings.. Lapping is an 'area process' not a 'spot process' like scraping. In this phase, the repeat-o-meter gives valuable additional information about the size of any small local variations. To some extent, the visible pattern of grindings on the plate can give a similar clue...

              2) When you are calibrating a surface plate, you will be making a map of the surface and need to step the mirror along the lines so that the steps are the exact spacing of the feet as close as possible. At this point, if it's after the lapping, only the maximum repeat meter variation is really of relevance.

              Location- Rugby, Warwickshire. UK

              Comment


              • #37
                Maybe this older video from Quality Digest will shed some light?
                https://www.youtube.com/watch?v=NWObTpn6dTk
                25 miles north of Buffalo NY, USA

                Comment


                • #38
                  Originally posted by Mark Rand View Post
                  I would suggest that there are two different operations being discussed, not one.

                  1) When re-lapping it is not neccessary to quantify every single deviation from flat. You are trying to reduce them and will re-measure more than once in the proceedings.. Lapping is an 'area process' not a 'spot process' like scraping. In this phase, the repeat-o-meter gives valuable additional information about the size of any small local variations. To some extent, the visible pattern of grindings on the plate can give a similar clue...

                  2) When you are calibrating a surface plate, you will be making a map of the surface and need to step the mirror along the lines so that the steps are the exact spacing of the feet as close as possible. At this point, if it's after the lapping, only the maximum repeat meter variation is really of relevance.
                  Per the video, the actual Federal spec for plates does not (here is where Paul A can jump in and criticize the procedure) cover a total point by point evaluation over every bit of the surface.

                  As far as I can see, the issue is that the spec requires both the level/Planekator/autocollimator measurement, AND the Repeat-O-Meter. In the video, the flatnness was OK, but the R-O-M detected local issues that actually lowered the grade of the plate.

                  It seems that since the goal is to meet the spec for a given grade, the procedure is to map it and sweep it, determining what the issues are, then lap the spots that need it, and proceed to check again to see what happened. Lather, rinse, clean, and repeat until it meets the spec.

                  I'm not sure I see the difference in goal and/or procedure for the "two different operations".

                  I suppose that if the Federal specs do not satisfy you, and you roll your own, you can require whatever winds your boat....
                  CNC machines only go through the motions

                  Comment


                  • #39
                    https://www.youtube.com/watch?v=Juo7ciTf5Sw

                    Comment


                    • #40
                      Nice video of the process and calculations. And it's good to see the offering of different ways to assess what the numbers mean.

                      It wouldn't be hard to do this in any shop with a spreadsheet to do the calculations. The only weakness I see in small shop use is the assumption that nothing shifts as you're doing the measurements. My shop floor is wood, so I know that just walking around changes a measurable amount. If there's no control for that, the numbers as you progress can be misleading since your measuring both the change in the plate surface plus any shift in the floor the plate is sitting on. Having a second level parked at the first position would let you cancel out floor changes, provided you don't have to walk back and forth to check the stationary one
                      .
                      "People will occasionally stumble over the truth, but most of the time they will pick themselves up and carry on" : Winston Churchill

                      Comment


                      • #41
                        i find it revealing, that flatness not only depends on the data but also on the definition. of course the deviations are small (0.2µ from memory) and would decrease with the number of points, but if you measure a 2 µ plate with a 20 millionths instrument that number suddenly looks big.

                        Comment


                        • #42
                          It simply depends on what you reference to. If you take the opposite edges as reference points on a plate with wear at one side, it will look different than if you realize it's very flat over most of the surface until it drops off. As an egregious example, if it's not level (earth gravity) but you measure the data across the plate it may show that it continues to get worse across the plate finally ending with a plate that's several thousandths high side to side. The calculations help it make sense, but you need to know the frame of reference to fully understand what the data tells you.
                          .
                          "People will occasionally stumble over the truth, but most of the time they will pick themselves up and carry on" : Winston Churchill

                          Comment


                          • #43
                            When I was training I was told by the trainers that the autocolumator generates the plane, datum being its three feet, the mirror just tells you howfar above or below the surface is, the mirror was moved along the surface and results plotted on the graph, fairly simplistic I know but that’s how I learn, side note we did a big iron table as the training exercise, the instructor waited till just before lunch to do the exercise, chalked the grid, did the measurements, bloody big hump in the table, he got us to do it again, another hump 12” further along, repeat another, drove me bonkers till the penny dropped, the sun was shining into the skylight and illuminating the table, as the light moved along the table the expanded hump moved!
                            the record was apparently 5 goes before noticing, I got 3!
                            mark

                            Comment


                            • #44
                              I was thinking along these lines. Consider a surface that has a wave nature, like a sine function. The wave length is a whole number fraction of the distance between the legs of the autocollimator target mirror's feet. That would ensure that the target mirror would never deviate from it's original angle. Now further imagine that that wavelength is also a whole number fraction of the common factor distance between the repeat-o-meter's feet so it also would not ever show any deviation. The autocollimator mirror and the repeat-o-meter would each just ride up and down on that wavy surface but never changing their angle or showing any difference in height. In theory, such a wave shaped surface could deviate by any, arbitrarily large amount from flat.

                              Don't bother saying it. I know that such a surface is not a very likely wear pattern in the real world. And I know that there are problems extending it from one dimension to two. But it does demonstrate that at least one possibility could exist. One thing I have learned about geometry is that there are an infinite number of different possible shapes that a surface could take. I am not at all sure that there aren't some that could fool both of these instruments, especially if they are not used in strict conformity with the conditions that the original designer's of these tests had in mind.

                              And I do know that when you start getting down to measurements in the tenths and millionths range, things can quickly go squirly. One of the stated, but perhaps not well known properties of the repeat-o-meter is that the mass of the metal under the DI's tip serves to dampen the jumpy nature of the readings that the DI would show without it. In the videos that I have seen of it in use, it is swept rapidly across the surface. What might it show if it were allowed to settle for a period of time at each location before the readings were taken? What other hidden points are at work with that device? And what hidden points may also be at work with the autocollimator?

                              I want to understand more about exactly what is happening with these instruments. One or a few internet videos do not answer all my questions. I may have to not only use each of them, but also actually build them so I can see the effects of any changes in their construction.

                              And by the way in response to another comment, my mind is not made up or closed on this. In fact, it is wide open.



                              Originally posted by J Tiers View Post

                              Go ahead and describe it.

                              ...<snip>...
                              Paul A.
                              SE Texas

                              And if you look REAL close at an analog signal,
                              You will find that it has discrete steps.

                              Comment


                              • #45
                                Do not forget the "X" pattern that is used with the instruments. That has the property of changing the effective distance between "feet" of the autocollimator mirror OR the Repeat O Meter relative to the "waves", and so should allow one or both to catch the deviation.

                                The standard was set up to catch just about anything that could be messed up.

                                The "rapid sweep" is not necessarily correct. It depends on what "rapid" is considered to be. It has to be slow enough that the eye can catch and observe the max and min of needle movement on the R-O-M. That ought to be slow enough that there will not be any effect of the moving part "skipping over the waves".
                                CNC machines only go through the motions

                                Comment

                                Working...
                                X