Homemade surface plate - how flat is it?

re: Flatness of parallels.
That no visible light penetrates junction of one to another is misleading. Width of bar width acts like a window blind; light may penetrate undetected, until you get perfect eye alignment to the 'joint' and light source. Plus, fluorescent lamps do not have the wave length or beam of an incandescent bulb.
And wider the bar, the more perfect it appears.
The straight edge profile most adept displaying lack of contact is the 'knife edge' variety. The name is misleading, actual contour is an apex with a small radius. That maintains its alignment throughout the length; [ie] the apex, radius, angle and tangents do not vary in relationship.
In lieu of a knife edge, try a drill or reamer blank. What "was" straight might prove disappointing. Another method is with 3 feeler gages; this works with wider material. Simulate the Airy points [https://en.wikipedia.org/wiki/Airy_points] dividing length by .57735 to locate two of the shims. use the third to prove openings prevail equally; middle and ends.
 
brino said:
Much logic indeed! That's a very interesting project.
My first thought for that application would be abrasive cord type of band-saw, or even water jet.

Is that within the capabilities of an inexpensive microcontroller (like arduino) or fully analog, or SOC?
Anything you feel like sharing will be greatly appreciated.

-brino
Click to expand...
Concretions contain their specimen surface (fern leaf, shrimp, etc) in their central plane, which is a mechanical weak point. You must let them open up in that plane or you ruin the specimen. The old freeze-thaw method is to leave them out in buckets of water all winter. Every year, some open. Some never will. Some also open up in summer if left to soak and then dried quickly in the hot August sun. Hammering opens many but ruins very many too. So, I thought I'd give resonant vibration a try. If nothing else, I'm learning a lot trying!!
I have an Arduino. You have to program them in C, which is tedious and creates error-prone code. I hate coding in C. I also own a PyBoard, which is progammable in Python, a vastly more powerful and productive language that produces working code much faster. I like coding in Python. Besides, you provide main memory via a SIMM socket, so instead of, say, 256K of memory, you can have 64Gb! Blows Arduinos away.
Right now, I don't see a need for the automation, and I'm not sure I'll ever get the time to do it anyway. I think I can do it with straight simple analog circuitry, which I've been working with since 19-ought-66.
 
I have three 12" diameter pieces of glass that are flat to a couple of waves of light; fine ground down to 9 micron; but not polished.

They are intended to be optical flats when I ever get back to that project.
 
I bought the Autocollimator, the associated mirrors and paraphernalia. Currently I am in the process of fabricating the “repeat o meter” that Bob Korves mentions. I am not skilled in the trade but the instruction book is very detailed and I have learned a bunch.... The equipment is a small fortune but I don’t regret the purchase, and once I get my plate properly measured and lapped I will in my mind have the satisfaction of knowing that I am further down the road than where I started, and better for it.


Sent from my iPad using Tapatalk
 
Toolmaker51 said:
re: Flatness of parallels.
That no visible light penetrates junction of one to another is misleading. Width of bar width acts like a window blind; light may penetrate undetected, until you get perfect eye alignment to the 'joint' and light source. Plus, fluorescent lamps do not have the wave length or beam of an incandescent bulb.
And wider the bar, the more perfect it appears.
The straight edge profile most adept displaying lack of contact is the 'knife edge' variety. The name is misleading, actual contour is an apex with a small radius. That maintains its alignment throughout the length; [ie] the apex, radius, angle and tangents do not vary in relationship.
In lieu of a knife edge, try a drill or reamer blank. What "was" straight might prove disappointing. Another method is with 3 feeler gages; this works with wider material. Simulate the Airy points [https://en.wikipedia.org/wiki/Airy_points] dividing length by .57735 to locate two of the shims. use the third to prove openings prevail equally; middle and ends.
Click to expand...

My homemade straightedges are heavily chamfered as well as slightly rounded, close to what you describe. I do mulitangle viewing (had to do it a lot when making the surface plate), and when possible use a north window as light source in a dimly lit shop, otherwise good old incandescent bulb. If any little speckles of light show through, that shows it very well because your eyes are dark adapted. So far, it's been very easy to see what I expect to be a 0.2 mil gap, some effort to see 0.1 mil, and dicy to see 0.05 mil.
Early on when grinding the surface plate, I discovered that a 4 mil gap is quite invisible when you are out in full sunlight!
I have tried some shim testing and results seem pretty good. I'm sure they're good to 0.2 mil. But I'll keep it in mind for future comparisons to flat surfaces. In fact, sometimes shimming is necessary. My spark plug feeler guages only go down to 3 mil, so I have a collection of material samples that I've miked. Two 2.1 mil end shims plus a 1.8 mil test shim leaves a 0.3 mil gap.
Most of my smaller drill bits are definitely not trustworthy for straightness measure, and the rest are only about 3-4" long, but worth a try for small test pieces.
Thanks for posting.
 
You must log in or register to reply here.
Back
Top