How can I find level within .0005 over 72"

Hello All!

I need to set 2 parallel rails that are 72" long with .0005" accuracy.

I have an electronic sensor that will detect a .0001 change in the position of a 4mm red laser spot. (Maximum laser spot size 8mm). If I can get the laser perfectly level, I can
sweep the laser to any heading I want, and using the sensor get the rails to within .0001" of the center of the beam.

My problem is: how do I get the laser perfectly level to begin the process?

I'm thinking, hang a weighted silver mylar strip from a tripod at say 100 ft from the laser on a calm night, attach the laser on the top of a pendulum bar, and bounce the beam off the mylar, adjusting the laser tilt relative to the pendulum bar until the return beam hits the exit lens of the laser.

Would this work? Would it be repeatable? Better Solutions?

Does a pendulum always hang perfectly plumb?

Cheers!

Bill

 PhysOrg.com engineering news on PhysOrg.com >> Researchers use light projector and single-pixel detectors to create 3-D images>> GPS solution provides 3-minute tsunami alerts>> Single-pixel power: Scientists make 3-D images without a camera
 Hi even with no aerodynamics effects, again because of rotation of earth, your pendulum will not rotate perfectly in plumb path (unless you use it on the equator) to see why it will not rotate perfect, check Foucault pendulum in Wikipedia.
 I make the max accuracy of a standard machine level (as used for setting machine tools) about half your requirement. You may be able to find a more accurate one. Incidentally have you considered the effect of curvature of the earth on your laser line? Good news - I make the correction 0.00001 inch over 72 inches.

How can I find level within .0005 over 72"

Hello All!

I guess my main question is...

Will the pendulum (mounted in a 2 way gimbal mount) always hang at the same exact angle, when removed and returned to the same geographical location on the earth?

Spirit levels accurate to .0005" at 12 inches are common, but at 72 inches the error is .003", which is too much error for my rails.

I guess there are devices that will solve my problem, but my budget is small.

 Could you not float your laser level on the surface of a liquid such as water in a bucket and if set as level to your satisfaction then it should remain level with rotation.
 Personally, I would use a flexible transparent plastic tube filled with water. You would of course, have a convex "dome" on the water surface at each end due to the surface tension, but you could take your level from the highest point on the dome (which would be in the centre). I would have thought this would give you sufficient accuracy. You didn't tell us what the application was that you are building - Is it something to do with an astronomical telescope mounting ?

 but my budget is small.
Have you considered hiring or borowing?

A few years ago I was looking for such a machine level, and as you say they are expensive.

I did eventually find one on Ebay, but meanwhile I talked to several local 'model makers' and machine shops. You would be suprised what people have.

As to water levels, the Agrimensores of ancient Rome used 32 feet long water levels to measure to your sorts of accuracy when constructing aquaducts and viaducts across valleys.

 Recognitions: Gold Member Science Advisor forget about level,,level is useless regarding precision machining and measurement..i think wha t you really want to do is inspect the 72 " ways fro flatness and straigtness and then parrellism.. level is NOT Flatness and you most ceretainly can use a laser to inspect these geometric features...you have to " buck in" the laser at two points ( usually the extreme ends of the way the ninsepct the dimensions in between...i sold laser alignment gaging for years
 Hello All! I did forget to mention my application! I am going to construct a cnc router to mill molds for ceramic parts. Since the molds are made in two or three parts, the mating surfaces need to be +-0.0005" to minimize leakage. I need to be able to "dry fit" the steel gantry and base frames before welding. I'm thinking that the best way to do that is to shim the steel boxtubes level, and measure the distance between the diag corners first. If all the tubes are level, the frame will be true. Plus or munus a few thousanths for the frame is fine..but the rails that the cutter head travels on must be true +-0.0005", or the machined mold parts won't fit properly. If I'm going to build it, I want it to be accurate. I like the idea of using water as a reference. I'd guess water is always perfectly level. As I mentioned, I have an electronic sensor chip ( Pacific Silicon QP50-6-18u-SD ) that will find the center of a cheap red laser pointer beam within 0.0001" X & Y dimensions at the same time just using two 9 volt batteries and two \$10.00 volt meters (pretty cool!) Once I get the laser beam level, the rest is all ice cream! Thanks for all the suggestions... All of us together are smarter than any one of us! Best Regards, Bill
 Recognitions: Gold Member Science Advisor once again..LEVEL is not FLATNESS and you need to have mating sufaces FLAT to have proper seal...get off the level thing...level is a perpendicular line to earth and you can indcue bend in a structure if you level to earth over long distance..due to the curvature of the earth...you nee to measure for flatness and this must be done with an intrinsic datum that is accurace , that is t osay..the laser beam..if you insit on using level,,,i garuntee you will have leaks
 Hi Mike, I understand. The rails have to be flat, the milled surfaces can only be as true as the rails. I was mentioning levelness only as an aid in manufacturing the frame that holds the rails. I understand that the final machine dosen't need to be level to create true parts. The accuracy (flatness) of the rails control that. Thanks for reading.. Bill
 Ceramic moulds huh? Will things get warm? If so your rails will not maintain their dimensional accuracy. This is a real issue. I remember a continuous welded viaduct (350 yards) where the dimensionl changes after half an hours sunshine could be measured in inches.

Recognitions:
Gold Member
 Quote by wmingin I need to set 2 parallel rails that are 72" long with .0005" accuracy. [...] My problem is: how do I get the laser perfectly level to begin the process?
 Quote by Ranger Mike forget about level,,level is useless regarding precision machining and measurement..i think wha t you really want to do is inspect the 72 " ways fro flatness and straigtness and then parrellism..

I concur with Ranger Mike, the only concerns are straightness and parallness of the 72'' ways.

If I hazard a guess I think you ask about levelness because you need a perfectly repeateble reference.

To my knowledge, when a manufacturer needs a reference for straightness the entire measurement rig is set up on a stone slab that is within a specified tolerance for flatness.

Flat stone slabs like that are manufactured by specialized companies. Slabs are ground against each other. If you would work the same surfaces against each other the whole time you will probably end up with one hollowed out and the other bulging up. To achieve high quality flatness multiple surfaces are exchanged. For instance, a production run could be 4 slabs, each with one high grade surface. Periodically slabs are exchanged, cycling through all combinations.
I'm guessing that the measuring equipment to verify the flatness of the finished product will be expensive.
I think surfaces like that are the only way to have a high grade reference for flatness.

You stated the level of accuracy required for the molds. (More precisely: for the mating surfaces of the constituent parts of the molds.)
To my knowledge, that is towards the top end of what is achievable at all with CNC machines. Top end machines like that cost tens of thousends of dollars, I wouldn't be surprised if they are in the 100,000 dollar range.

So it's really hard to see how you could possibly achieve your goal to 'construct a router to mill molds for ceramic parts'.

 Hello All! Thanks for all the input. I guess I'll just fool around and see how close I can get. If y'all are interested, I post photos of my results.. ( success and/or failure ) Thanks again for all the suggestions... B.

 Quote by wmingin Hello All! I need to set 2 parallel rails that are 72" long with .0005" accuracy.
$$0.0005 \, \mathrm{in} \times \frac{25.4 \, \mathrm{mm}}{1 \, \mathrm{in}} = 0.0127 \, \mathrm{mm} = 12.7 \, \mathrm{\mu m}$$

This is of the order of:
$$\frac{12700 \, \mathrm{nm}}{550 \, \mathrm{nm}} = 23$$
23 wavelengths of green light.

 Hi Dickfore, The 550 nm value represents the wavelength of green light? B.

 Quote by wmingin Hi Dickfore, The 550 nm value represents the wavelength of green light? B.
Yes, it does. The point is, at the level of accuracy you specified, your rails are no longer straight lines, but you might notice any unevenness in their fabrication, or bending due to minimal stresses that they are subjected to.

Namely, the shear modulus of an aluminum rod is $G = 25.5 \, \mathrm{GPa}$. This means you may get a transversal deflection of 0.0005'' over a distance of 72'', which is a relative deflection of $\delta = 6.94 \times 10^{-6}$. This requires a shear stress of the order of $2.55 \times 10^{10} \, \mathrm{Pa} \times 6.94 \times 10^{-6} = 1.77 \times 10^{5} \, \mathrm{Pa}$. Suppose the cross-section of the rail is a square with a 1 '' side. It means the cross-sectional area is $(2.54 \, \mathrm{cm})^{2} = 6.45 \times 10^{-4} \, \mathrm{m}^{2}$. Thus, the deflection force is $1.77 \times 10^{5} \times 6.45 \times 10^{-4} \, \mathrm{N} = 114 \, \mathrm{N} = 25.6 \, \mathrm{lbf}$.