Laser Aligning Machine Tools

[BobWarfield]

I have been reading with some interest a couple of threads in a couple of places. First is the elaborate discussion of laying out centerlines by JerryFlyGuy on the Zone at:

Layout of centerlines

Second was an article on another board about how to check flatness to an amazing tolerance (0.0001") with a laser level:

http://www.homeshopmachinist.net/bbs...ad.php?t=18917

Given that commercial CNC machines are often set up with lasers, I couldn't help but be attracted to the idea of how to do more with the humble laser level. There is a fascinating article about how to do the CNC alignments here:

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

It seems that everything revolves around the innate properties of laser beams together with a magical gadget called a PSD = "Position Sensitive Device." PSD's can basically precisely find the center of a laser beam and tell you when they move how far and in what direction (or at least how far on an X,Y grid). Pretty cool, eh? Except these gizmos cost like $10K. How does the average CNCZone reader get their hands on one?

More reading, and it seems they're available more cheaply. For example, this article tells how to build a PSD around a $150 chip:

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

Better yet, the fellow that wrote that article will sell you the gadget described, already built for $750:

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

Still not cheap, but we're a heck of a lot more in the ballpark than we had been at $10K! Now this gadget doesn't look any more complex than the Shumatech DRO or some of the other goodies we routinely play with like Gecko drives. I'm not sure if some other firm sells one more cheaply, but it seems tantalizingly within reach to be able to lay out our plasma tables or mills so that the linear slides are just about dead on.

Just a thought,

BW

[Mcgyver]

Bob, that's a good collection of links. I'd read Evans article on the laser for flatness and skimmed the circuit cellar article. I really like the idea of a a diy linear encoder, I think it would be well received and have lots of applications. On my last electronics order i bought some hall sensors to play around the ball bearing (newall) idea, but haven't progress more than assigning a parts drawer.

didn't realize the circuit cellar one was that accurate, but if its delivering .0001 that's awesome. Thanks for compiling the list and providing some mental stretch - when can we see a prototype

[Geof]

Why not build your own Position Sensitive Device for a laser beam using some phototransistors, a steel ball and a few other circuit components.
A laser beam does not have an absolutely even intensity across the beam; it is a gaussian distribution peaking at the center and falling off to the edges and for a good laser is symmetric around the axis. If the beam is incident on a very precise reflective convex surface with the beam axis passing precisely through the center of the convexity an infinitesimally small portion of the beam exactly on the centerline will be reflected back along the axis. Most of the beam will be reflected back as a diverging cone with the amount of divergence being related to the beam diameter and the radius of the convexity. If a cross of four phototransistors, fixed relative to the reflective convex and centered on the beam axis is placed around the beam 'looking' at the reflected cone they will each detect the same intensity. The distance across the cross and the radius of the convexity would determine what this intensity was compared to the beam maximum. When this cross of detectors and the convex reflector are moved off axis the intensity will diminish for one detector and increase for the one opposite. If opposite detectors are wired into a bridge circuit, with the detector assembly centered the bridge could be balanced and the off axis motion would produce an imbalance which could be used to measure the off axis direction and distance. The sensitivity could be very good; if the cross size and radius mentioned above where adjusted to yield a detetcted intensity of about one half beam maximum then moving approximately half the beam diameter off axis would take the intensity at one detector down close to zero while the opposite one would double. The resolution would be determined primarily by the instrument used to measure the bridge imbalance and a good voltmeter would give a sensitivity of one part in a hundred. If the beam diameter was around 0.05" this gives an off axis sensitivity of around 0.0005" or better.

Below is a sketch showing the layout of the components.

[BobWarfield]

Yes Geof, your design looks plausible.

The advantage of the Hamamatsu chip is that it provides more resolution versus just one bit from the ball bearing reflection. The distance by which the laser is offset from the center is encoded as a voltage on these chips. The particular design discussed for the aculux device uses a 12 bit analog to digital converter to convert that to numbers and a PIC device to further process the Gaussian distribution and generate a human friendly interface. I have to admit, being able to zero the device, move it, and then read the magnitude of the motion to 0.0001" accuracy is a pretty handy format for our purposes.

Interestingly, while the cost for an assembled unit is $750, the Circuit Cellar article provides all of the information, including the PIC controller program, to build your own. You can buy the PSD chip itself and preamp PC board for $175 from aculux. I'm not PIC maven, but it sure looks to me like the additional costs for the PIC board with LCD readout couldn't be too much. Here is one for $289, for example:

http://www.cippsites.com/Merchant4/m...re_Code=melabs

Again, I am no PIC maven, but it sounds like a product like this could be assembled for somewhere on the order of perhaps $500 anyway. Not sure what price would attract people, but it seems like a pittance given what it is capable of if one were sure it really worked well.

Alas, it needs an aspiring entrepreneur ala a Mariss or Shumatech sort of venture to make it happen. I'm not one who can pick up a soldering iron and bang one of these gizmos out.

Best,

BW

[Geof]

Hook the design I propose to two digital voltmeters, mount the detector on something with a known resolution such as the table of an old fashioned knee mill with a DRO installed and construct a nomogram; voltage to displacement.

[Lotus14]

Another idea is to look around for old equipment. I have a complete laser alignment system that was built by HP back in the '80s. It uses a laser beam that has been polarized into two axes. One part, lets say the vertical beam is a reference, the other beam; the horizontal beam does the measurement by being reflected off of a mirror. The two beams when the leave the laser are the same wavelength, and because they are created from the same beam, their waves are aligned; in other words the peaks and valleys are aligned. But, when one is sent out to a moving mirror the waves are no longer aligned, and the detector can figure out the difference, and convert it into a distance. This can be accurate to a quarter wave of light, about 550 nanometers. There are newer versions of this gadget, but I would expect you could find older parts of this system for not too much money. The model I have is a HP-5150, if I remember correctly; it came off of an old step-and-repeat camera used to put the image on silicon wafers. Look around, I got mine for practically nothing, at a surplus sale at a big chip maker, but judging by the manuals, it was used for a bunch of stuff including machine tools. I use mine for inspection, and other measurement tasks.