Qualitative flatness testing, cheap. (pics)

[BobWarfield]

I just can't get over the power of a simple laser level. Unfortunately, it's making me greedy for even more. It's nice to know how not flat my surface is, but where is it not flat? And how perpendicular is it to that other surface over there? Is it parallel to that one as well? Basically, can I measure everything there is to know about my machine tool, whether to adjust it, rebuild it, or create it from scratch?

So, the Internet is a dangerous place. I went looking and looking, and I did find a couple of interesting hits. I'm not done searching, but I did think I would share some information. As nearly as I can tell, the key to using lasers for metrology (hey, that's a big new word, eh?) is something called a PSD, which stands for "Position Sensitive Detector". This little gomer is designed to tell us, with extreme accuracy, exactly where the dead center of a laser beam is hitting the sensor. Nifty, eh?

Now using said device, we can do all manner of machine tool alignment and checkout. I like Hamar Laser's description for machine tools the best:

http://www.hamarlaser.com/howitworks...ng_Centers.htm

As near as I can fathom, once you have the PSD to tell what the laser is doing, the rest is gravy, and it seems like you can use a cheap laser level for this purpose. My next question was, "Where the heck am I going to get a cheap PSD?" On this one, the jury is still out. Colleges are fiddling with them in student experiments:
http://physerver.hamilton.edu/people...deflection.pdf

How about measuring the deflection of the laser as a sound wave passes over it in air? Obviously these things are pretty sensitive! A little more research turned up this article about how to build one:

http://www.circuitcellar.com/library/print/0303/5.htm

They use it to map a surface plate to 0.0001", which sounds right up our alley. In fact, that then led me to look to the end of that article and find out the gadget they show how to build is available commercially:

http://www.aculux.com/Alignment_pricing.htm

Not exactly cheap at $750, but it is still a pretty amazing deal compared to what real CNC laser alignment tools cost.

There's probably more and better to be had from deeper Google dives by better minds than I, but this definitely whetted my appetite.

Best,

BW

[Evan]

I am going to do a little experimenting tonight with some really simple optical apparatus made with some prisms from a pair of junk binoculars. If it works as I think you will be able to reference two surfaces to each other, or at least two ends of a surface like the bed of a lathe or milling machine ways. I calculate the possible relative precision at about 20 arc seconds or better.

[gearedloco]

My laser is a third of a bubble out of plumb!

Or maybe piece of tile is a lot flatter than I think it is. And I don't think its flat, because pushing my dial indicator mounted on a magnet base (it's the only one I have right now!) shows variations on the order of +/- .003 or so. The fact that it's a magnetic base shouldn't matter, because the tile isn't magnetic, the magnet is "turned off," and the bottom of the base seems to be quite flat - it doesn't rock when sitting on the tile or on my mill table.

The problem is that I only get 1 line on the target. I scattered salt on the tile and adjusted the laser level such that it was as low as I could get and still light up the salt on all parts of the surface, and with about half the width of the line on the edge of the tile and the other half on the target in line with the top of the tile.

The tile is black with a shiny surface and is about 16" square. The laser level is on a tripod a couple of feet from the front edge of the tile, and the back edge of the tile is about 6" from the target.

I took some photos, but I haven't had a chance to get them out of the camera. If it seems worthwhile I'll try to get them posted tomorrow.

Ant thoughts about where I went wrong, given my poor description of the setup? Hmmmm - could it be I have the spacing of the elements bass aackwards? Should the tile be farther from the target and closer to the laser?

-bill

[Evan]

The tile is much too close to the target screen. It needs to be at least several feet away and preferably further from the target than the laser so the direct beam and the reflected beam have some chance to diverge.

[vpt]

That is great Even! Kind of similar to body work lights. They cast a shadow line on the car and you can find dents by where the shadow bends and distorts.

[DeckelFan]

I realise that this thread is more than ten years old, but I find it interesting and would like to see the pictures/photos that should accompany it. Am I correct that these should appear in-line with the text? Assuming that is correct, do others here see those images? Is there anything I can do to recover/view those missing images? I'm new here, so apologies if this is stupid/obvious.

[RB211]

I realise that this thread is more than ten years old, but I find it interesting and would like to see the pictures/photos that should accompany it. Am I correct that these should appear in-line with the text? Assuming that is correct, do others here see those images? Is there anything I can do to recover/view those missing images? I'm new here, so apologies if this is stupid/obvious.

Evan used to host all of his content on his own servers. Evan for what ever reason decided to delete his online presence and remove himself from this forum.
I got excited for a second thinking he returned.

[Paul Alciatore]

I was quite interested in this post when Evan first put it up. You may notice that I did not comment at that time and that was because I was not sure about the technique and weather or not it had any merit. I did and still do suspect that it does, but I just was not sure. And in many discussions (disagreements) with Evan, I came to the conclusion that I did not want to get into any more without at least a fair amount of certainty on my part. His technique was basically to observe the reflection of a laser light that struck the table at a very low angle. I have used a similar technique to make a rough judgment of a surface by observing reflections of lines or edges at a similar, low angle. Unlike Evan, I never tried to work out the math involved. I never tried to quantify it.

Anyway, this rang a bell and I checked my files. I had downloaded the original three posts that he started the thread with and that includes the photos except for the first and fourth ones which somehow did not get saved. I know that Evan had some issues when he left the board and while I may not know what they all were, but I do respect them. He may have pulled the plug on his image server simply as a part of other things that occurred in his life at that time or he may have wanted to keep those images private. I do not know. In any case I am sure that he owns the images and copyright may very well have been invoked.

I would be willing to send the downloads to anyone who is willing to also honor his unspoken motives. Just send me a PM with your e-mail and a statement that you will not post them here or anywhere else or use them for commercial purposes.

[old mart]

Were there any pictures in the first post?

[Hopefuldave]

Informative pictures, yes...
I thought I'd saved them somewhere, but not found 'em yet - they may be on a long-dead hard drive

Dave H. (the other one)

[old mart]

Now I understand, they would not have survived 14 years, especially with the change of format.

[J Tiers]

The idea here is that the laser light, if reflected from a perfect surface, would all reflect at the same angle. So the line reflected onto the screen should be exactly the same thickness on the screen as it is at the original beam.

If part of the surface is tilted, the tilt will reflect that part of the beam at a different angle to the rest, and the line on the screen will be thicker, and possibly partly separated from the original line.

. Limitations:

The magnification factor needs to be such as to make the smallest error you want to detect deflect more than the thickness of the beam. That way, a defect that is at the far end of the surface will be deflected out of the original beam line and become visible and identifiable.

The technique may not be very good for directly detecting magnitude of errors. It is good for detecting a flat surface vs a non-flat. But an extraneous line on the screen might be from a defect location and exact magnitude anywhere on the surface. It might be a big one at the far end of the surface, or a smaller one near the middle, etc. You would need to look at the line to discover what part is missing (because it got reflected elsewhere), and it is not clear that the image would be crisp enough to see that and make a measurement. Scattering would affect the clarity of the image as well.

The laser must have good beam quality, and have no tendency to spread out vertically on it's own. Any spread has the potential to mess up measurements of the amount of error, particularly if it is not even across the entire beam (or fan pattern)

The spreading fan pattern is not the most desirable, since it makes the image distorted horizontally like a fish-eye lens.. There would be some difference vertically, due to different path lengths across the fan-shaped beam as it reaches the screen.

Horizontal effects of errors are not very detectable. Probably two images at right angles are needed

It seems that there is something missing.

. A possible solution for certain problems:

Collecting the image on a SLANTED screen would spread it out and might very well give a reasonable "map" of the surface. Missing areas of light would be defects, and THEN, the distance the reflected error image is from the place it should have been would be a measure of the error size. It would need to be corrected for the differing path lengths due to the slant, but that is a somewhat secondary consideration.

[dian]

http://www.aspe.net/publications/Ann...T/KULAWIEC.PDF

[J Tiers]

http://www.aspe.net/publications/Ann...T/KULAWIEC.PDF

That is fundamentally different, as it is an interferometer type instrument. What Evan described was a direct measurement unit, which is more capable of being made in the shop, and which depends on direct measurement of the produced image for determining error magnitude.

That 1999 paper describes a method which is based om interferometry and which produces a detailed map of the surface to a high accuracy. But it is much less easily made..

[Paul Alciatore]

I said it above but perhaps it got lost in my verbose message.

I downloaded Evan's original, three posts, complete with the photos (the first and fourth photo did not come through in a usable format). As I said above, I want to honor Evan's wishes, whatever they were, when he took his image server down. So I am not going to post those photos now. But I am willing to share then with individuals if they also agree not to post them anywhere or to use them for commercial purposes. Just PM me with a statement to that effect and your e-mail address.

I said before that I am not sure about the efficacy of this method and I have not analyzed it or tried it. That is why I did not reply to this thread before. It seems to be highly dependent on the skill of the user both in setting it up and in interpreting the results.

Now I understand, they would not have survived 14 years, especially with the change of format.