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

[Paul Alciatore]

I have been trying to understand the various techniques that are used in generating and regenerating or renewing a surface plate. I think I have a good appreciation of the three plate method. And I can understand how a master flat can be used to scrape a new, flat surface by comparing the two. I have also watched more than one internet video where the technique used by the plate lapping companies will check and then lap a worn plate back into spec. And that is where I do not understand exactly what they are doing. Perhaps this is by design as they may want to keep their technique to themselves. But I do want to understand what is going on.

Two instruments seem to be used: a repeat-o-meter and an autocollimator with some kind of target. The repeat-o-meter is a way of checking to see if the curvature of a surface is uniform. BY ITSELF, it does not insure that the surface is flat. Only if it has the same curvature across it's entire surface. So these pros seem to say they do not rely on the repeat-o-meter for the entire measurement. What they seem to do is to use an autocollimator with some kind of target, which is probably a form of a mirror, perhaps with an additional trick or two.

The thing is, an autocollimatoris not basically a way for measuring surface flatness or the deviation from flat, at least not in that orientation. It is an instrument that is used to measure angles, usually small ones. One explanation that I read detailed the measurement of surface flatness with an autocollimator as the taking of a series of angular measurements of a "base" mirror by moving that base mirror by the effective length of it's base at a time and using the measured ANGLE of that base mirror at each position, along with the length of it's base to calculate the amount of rise or fall of the mirror in that distance. In other words, the autocollimator was actually measuring the angle and then trig was used to get the flatness or lack thereof. It was doing exactly the same thing that the repeat-o-meter does and nothing more. Perhaps with more accuracy, but still the same thing.

The problem that I see here is that the surface plate pros were NOT moving the "base" mirror by the length of it's base. They were moving it by much larger distances so any calculation based on the length of it's base would be wildly inaccurate. Or conversely, if they were using the actual length of these movements to calculate the deviation, then they were missing more than half the positions along the lines they were checking. And there is absolutely nothing that says that the curvature was constant along that distance. In fact, the very fact that they were getting different readings, says that the curvature was changing.

Among others, the YouTube video by OxTool seems to illustrate this.

https://www.youtube.com/watch?v=EWqThb9Z1jk

So, what is going on here? Are these "pros" just being deceptive? Perhaps trying to hide their trade secrets? Does their "base" mirror have some kind of optical trick in it? Is there some other kind of trick? Or what? Perhaps they calibrated the repeat-o-meter on a laboratory grade plate they keep on their truck, out of sight, and then go more by the readings of that secretly calibrated repeat-o-meter while the autocollimator was just for show. You know, dazzle the dumb hicks in those shops with the magic, high tech, expensive, optical instrument, but actually work with the plain-Jane repeat-o-meter.

[JoeLee]

I've watched that video before, a couple times in fact. What I still don't quite understand is how does he know that his cast iron plates a perfect? He also mentioned something about self correcting. How do they self correct? He also mentions something about seasonal change..... or how a plate can vary form summer to winter. iow about his repeat o meter or auto colliminator isn't there also some variation taking place with those tools in regard with temperature and stabilization?? I would think with all this being so critical shouldn't he leave his inspection tools in the same room with the surface plates over night so every thing is at the same temp.? I say this because he has these tool in his truck, out in the hot sun perhaps or AC. Something has to be changing there due to temp differences.

I also wonder what he is seeing when he looks through his scope. The target that is being looked at appears to be a mirror. Are there graduations on it are there graduation lines in the scope? How is this tool calibrated?

JL...............

[J Tiers]

......
The thing is, an autocollimatoris not basically a way for measuring surface flatness or the deviation from flat, at least not in that orientation. It is an instrument that is used to measure angles, usually small ones. One explanation that I read detailed the measurement of surface flatness with an autocollimator as the taking of a series of angular measurements of a "base" mirror by moving that base mirror by the effective length of it's base at a time and using the measured ANGLE of that base mirror at each position, along with the length of it's base to calculate the amount of rise or fall of the mirror in that distance. In other words, the autocollimator was actually measuring the angle and then trig was used to get the flatness or lack thereof. It was doing exactly the same thing that the repeat-o-meter does and nothing more. Perhaps with more accuracy, but still the same thing.

The problem that I see here is that the surface plate pros were NOT moving the "base" mirror by the length of it's base. They were moving it by much larger distances so any calculation based on the length of it's base would be wildly inaccurate. Or conversely, if they were using the actual length of these movements to calculate the deviation, then they were missing more than half the positions along the lines they were checking. And there is absolutely nothing that says that the curvature was constant along that distance. In fact, the very fact that they were getting different readings, says that the curvature was changing.

Among others, the YouTube video by OxTool seems to illustrate this.

https://www.youtube.com/watch?v=EWqThb9Z1jk

......

"Dazzle the hicks"..... CONSPIRACY THEORY!

I think you are misunderstanding the purpose and operation in this context.

The device projects an image in a beam of light, which is reflected back to a viewer. Any change in the angle of the mirror relative to the light beam will change the location of the reflected image.

If you operated it on a large sphere, the location of the reflected image would change as the mirror moved away "over the curve" of the sphere

The ideal case is where the image does NOT move in any relative position of the mirror. On a large perfectly flat surface that would be the result.

On a large surface where there is a step change in the middle, the result would ALSO not change, as long as the higher and lower parts of the surface were parallel, and you did not happen to put the mirror on the "step" itself..

The Repeat-o-Meter is the OPPOSITE. It will NOT see a sphere, but it WILL see any CHANGE of slope to the surface. So in the case of the step change, it would flag the step, but it would miss the previous example of a sphere.

The answer is that you NEED BOTH. they are complementary, which is why they are both used. What the autocollimator misses, the Repeat-o-Meter detects, and vice-versa.

In a way, the autocollimator is like a level. The level would also change indication over a large sphere, and it would also miss the "step", unless set exactly there.

[CalM]

Well described JT!

And in regard to "flatness of the working lap, Laps are like broken clocks, They are right twice a day.

That is, The "flatness is constantly being chased between concavity and convexity. I does not require a flat lapping plate to correct an out of true reference flat (Granite or other). It DOES require KNOWING WHAT YOU ARE DOING. And that is the topic of this post.

Observation of successful methods is a good beginning. Knowing the principles of the tools is another. Application of the two is the first steps to achievement.

K&E have /had a good reference on the use of autocollimation for accurate work. It was on line some years back. Worth a brows to find if it still is available.

Here is something from Taylor Hobson
https://www.taylor-hobson.com/-/medi..._lowres_en.pdf

Davidson Optronics once offered informational content as well.

It is a rather deep and crafty field

ps.
If anyone is need of a DoAll AAA grade 4'X5' granite, let me know. I have an extra ;-)

[QSIMDO]

Keith Rucker has one which he explains fairly well.

[eKretz]

Cal, it would probably behoove you to fill in any possible takers by advising of the plate's location...

And great concise explanation JTiers.

[Paul Alciatore]

J, thanks for the response. Others have praised it and I too appreciate it, as far as it goes. But I may get a bit nit-picky. Please do not take offense.

First, you seem to be saying that the target that they use with the autocollimator is just a good mirror. Probably a front surface mirror with a flatness at 1/10 wave or better. It would need to be adjustable in angle in it's holder. But the exact nature of this target is one of my questions.

I am going to add my further comments inside the quote of your response, in red.

"Dazzle the hicks"..... CONSPIRACY THEORY!

I think you are misunderstanding the purpose and operation in this context.

The device projects an image in a beam of light, which is reflected back to a viewer. Any change in the angle of the mirror relative to the light beam will change the location of the reflected image.

We are talking about the autocollimator here. That is what ALL the literature says; it detects ANGULAR changes. I have that.

If you operated it on a large sphere, the location of the reflected image would change as the mirror moved away "over the curve" of the sphere

Well, yes, it would. So, if they move the target mirror to several locations along the line on the surface plate which is spherical, the error would become greater and greater. So the fact that they are leaving the autocollimator in one position while moving the target to progressively greater distances is what is distinguishing it from a repeat-o-meter where the reference base AND the probing tip are both moved, keeping the distances between them the same. OK, that is a good point to remember.

The ideal case is where the image does NOT move in any relative position of the mirror. On a large perfectly flat surface that would be the result.

Yes, that would be the case on a flat surface. However with the way they do it, taking readings at only a few points along the line across the surface, they could easily miss an up or down "step" by taking successive readings before and after that step, but not on the step itself. Oh, OK, you seem to say that in the next paragraph.

On a large surface where there is a step change in the middle, the result would ALSO not change, as long as the higher and lower parts of the surface were parallel, and you did not happen to put the mirror on the "step" itself..

The Repeat-o-Meter is the OPPOSITE. It will NOT see a sphere, but it WILL see any CHANGE of slope to the surface. So in the case of the step change, it would flag the step, but it would miss the previous example of a sphere.

OK, by saying it will not see a sphere, you are in agreement with my analysis on this point. "The Repeat-o-Meter is the OPPOSITE." Well, it is the opposite in that it reads in linear measure while the autocollimator reads in angular measure. It "flags" a step only because, in normal use, the repeat-o-meter is swept across the surface. It takes readings at an infinite number of positions along any line that is is moved across. Now, if the target mirror for the autocollimator were to be swept across that same line, it too would pick up or "flag" any steps that are present. So it seems to me that the primary difference is due to the WAY the instruments are typically used and not in any inherent differences in them. And yes, those inherent differences do exist.

The answer is that you NEED BOTH. they are complementary, which is why they are both used. What the autocollimator misses, the Repeat-o-Meter detects, and vice-versa.

Even at this point, I have to think that a repeat-o-meter that is calibrated on a known good flat, would do the whole job and the autocollimator would not be necessary. Am I missing something here?

In a way, the autocollimator is like a level. The level would also change indication over a large sphere, and it would also miss the "step", unless set exactly there.

That gets back to another of my points; in the videos they only take readings at a few points along the lines. So they can easily miss a hole or a step. I suspect this is a shortcut to save time because time is money. It also makes me suspicious of the process. It probably works 90% of the time, but it does not seem to be a proper way to do it.

[eKretz]

Your points are why the two must be used together. The autocollimator will miss a step but the repeat-o-meter will not. If the repeat-o-meter reads near zero change, or near perfect uniformity, then there are no steps for the autocollimator to miss. If the autocollimator and repeat-o-meter both read near zero change, there's nothing to correct. The repeat-o-meter set to a perfectly flat surface could be used alone, but how do you guarantee that any surface you bring along for a reference is still perfectly flat once you arrive? Both instruments together make a sort of self-checking reference. I agree that ideally the autocollimator mirror movements should overlap for maximum accuracy, but if all the steps are showing no change, it makes as near as no difference. Perhaps they would spend a little more time and care with that if they detected more change between readings.

[J Tiers]

The Repeat-o-Meter is NOT a "calibrated measurement device". It is a calibrated detector of CHANGE.

If it were calibrated, yes it would, or at least could, detect a spherical surface. But it can only detect the difference over it's small footprint. A series of tiny errors that are each undetectable, would still end up making a larger change over a longer distance, which the autocollimator detects.

The autocollimator, as you noted, has as large (long) a "footprint" as you want, so long as you can still read the image accurately. You will also note that with double the path length, it doubles the sensitivity vs just the distance to the mirror..

Yes, the mirror has to be adjustable or perfectly made, to ensure it is able to get the image on-screen on a long flat surface.

[CalM]

The collimation retro reflection mirrors mimic the angle of the mirror base relative to a pre-established plane.

Draw yourself a 2D cartoon . The arc of a large spherical dish. then draw is several short segments along that arc.

At one end of the arc, the segment will tip one way, at the other end, the tip will be opposite. in the center, there will be no tip at all.

That tip is displayed in the eyepiece of an autocollimator.

Do the same thing with the repeat-o-meter. The reading displayed will remain the same regardless of the position of the instrument on the arc.

Two different types of error. Two different instruments used to characterize the surface.

And though the stepped placement of the reflection mirrors may seem large, The experienced technician knows where to look ;-)

Wear and error are often predictable based on use and accessability of the flat.

(eKetz, S. VT)

[tlfamm]

Some K&E literature here:
https://www.brunson.us/blog/post/ali...talogs-manuals


Other resources:

https://oaktrust.library.tamu.edu/bi.../T3pg17-22.pdf

https://www.vermontphotonics.com/VT_...plications.pdf

https://www.spectrum-metrology.co.uk...raightness.php

[JRouche]

I have been trying to understand the various techniques that are used in generating and regenerating or renewing a surface plate. I think I have a good appreciation of the three plate method. And I can understand how a master flat can be used to scrape a new, flat surface by comparing the two. I have also watched more than one internet video where the technique used by the plate lapping companies will check and then lap a worn plate back into spec. And that is where I do not understand exactly what they are doing. Perhaps this is by design as they may want to keep their technique to themselves. But I do want to understand what is going on.

Two instruments seem to be used: a repeat-o-meter and an autocollimator with some kind of target. The repeat-o-meter is a way of checking to see if the curvature of a surface is uniform. BY ITSELF, it does not insure that the surface is flat. Only if it has the same curvature across it's entire surface. So these pros seem to say they do not rely on the repeat-o-meter for the entire measurement. What they seem to do is to use an autocollimator with some kind of target, which is probably a form of a mirror, perhaps with an additional trick or two.

The thing is, an autocollimatoris not basically a way for measuring surface flatness or the deviation from flat, at least not in that orientation. It is an instrument that is used to measure angles, usually small ones. One explanation that I read detailed the measurement of surface flatness with an autocollimator as the taking of a series of angular measurements of a "base" mirror by moving that base mirror by the effective length of it's base at a time and using the measured ANGLE of that base mirror at each position, along with the length of it's base to calculate the amount of rise or fall of the mirror in that distance. In other words, the autocollimator was actually measuring the angle and then trig was used to get the flatness or lack thereof. It was doing exactly the same thing that the repeat-o-meter does and nothing more. Perhaps with more accuracy, but still the same thing.

The problem that I see here is that the surface plate pros were NOT moving the "base" mirror by the length of it's base. They were moving it by much larger distances so any calculation based on the length of it's base would be wildly inaccurate. Or conversely, if they were using the actual length of these movements to calculate the deviation, then they were missing more than half the positions along the lines they were checking. And there is absolutely nothing that says that the curvature was constant along that distance. In fact, the very fact that they were getting different readings, says that the curvature was changing.

Among others, the YouTube video by OxTool seems to illustrate this.

https://www.youtube.com/watch?v=EWqThb9Z1jk

So, what is going on here? Are these "pros" just being deceptive? Perhaps trying to hide their trade secrets? Does their "base" mirror have some kind of optical trick in it? Is there some other kind of trick? Or what? Perhaps they calibrated the repeat-o-meter on a laboratory grade plate they keep on their truck, out of sight, and then go more by the readings of that secretly calibrated repeat-o-meter while the autocollimator was just for show. You know, dazzle the dumb hicks in those shops with the magic, high tech, expensive, optical instrument, but actually work with the plain-Jane repeat-o-meter.

I love that story so I will not dissect it.

I got lost a lil around the "I read detailed the measurement of surface flatness with an autocollimator".

I in fact Paul like most if not all of the ideas. FYI. I have a very flate 12" dia plate. Welcome to use. Whatcha got? Im looking for an XRF gun for use (cant afford to buy one idiots). Anyway.

I have a light source and a 12" round. Autocollimate all you want. Its just sitting idle anyway.

Never really knew how to use it. Along with the other rounds. This cat is BIG though. Belongs in a lab. Not my garage.

You guys have seen it. And the frigin Lamp. supposed to put out a specieal wave length light or some chit. I dont know.

I that disk is flat and I really want to use it as a flat surface for finish polishing with small grain.

I dont want to muck it up so I keep it in the big ol wooden box its always been in.

Giant chunk of who knows what? Glass, plastic, quartz? Dunno.

Nice wooden case though, I can use that. JR

[garyhlucas]

Years ago when Volt Ohm Meters were analog with a moving needle the good ones like high end Simpsons had a mirror surface on the scale that you read. The purpose was to get rid of parallax error that you would have if you were not looking straight on. What you would do is close one eye and line up the indicator with its reflection in the mirror by moving your head. The reflection could only be directly behind the indicator when your eye is exactly perpendicular to the mirror.

So I think the autocollimator works the same way. The farther away also increases the sensitivity because the ratio of beam length to mirror angle increases.

[Bented]

Bought a new small granite surface plate this past March, it came in a red box along with this.


If you have no method of verifying such measurements yourself you may only rely on what the manufacturer tells you, end of story.

[wierdscience]

So I think the autocollimator works the same way. The farther away also increases the sensitivity because the ratio of beam length to mirror angle increases.

Bingo! Because as we know, distance will amplify error.