Single axis.
Max load of 3000 N along the axis.
Travel distance is 1 mm (not a typo)
Accuracy 1-5 microns.
Controlled by some sort of electronic device.

Zero backlash, zero side to side play and zero up and down play. Approximately 20cm x 30cm .

Budget- 600$

Any ideas would be great. Pretty open ended.

Sorry to say, but what are you trying to build?

The best I can tell is that he wants this setup to prove that such a setup is possible. He will then build a bigger setup based on this one.

We were building a CNC engraver, by building I mean finishing, for our senior project, but the two advisor's said they felt it did not include enough design aspect. So they threw this little number on us.

The accuracy and range are well within the bounds of commercial "steppers" used to pattern semiconductor wafers. These kind of systems rely on laser interference to measure distance for feedback.

The load is pretty impressive.

I was going to suggest piezo electric movement like they use in Atomic Force Microscopes, but I don't think it will fit into that force requirement.

Maybe you can use the change in length of a rod with temperature for the movement. Those are quite strong forces, and would overcome the substantial clamping needed for zero backlash in multiple axis arrangements.

3 kN for engraving? I got to be deep text then. :)
Even the smallest linear guides takes the load, you can run one single rail with one or two carriers. They are very tight and with a ballscrew you'll have no problems keeping the tolerances, if we talk inches though (Microns of a mm is caused by temperature).
I suggest you check old threads in the other forums, for example in the Engraving (http://www.cnczone.com/forums/forumdisplay.php?f=347) and the Metalworking machines-subforums (http://www.cnczone.com/forums/forumdisplay.php?f=1).

Regards,
Sven

Microns of a meter. 1/10000 inch is 2.54 microns I think.

The device is seperate from the engraver. They basicaly gave us two different tasks to complete. The engraver is well under control, but the slider is giving us head aches.

Thank you HarryN. I am going to start researching along those lines. I like the thermal idea because it didn't even occur to me.

Microns of a meter. 1/10000 inch is 2.54 microns I think.


Yes, my fault. I messed up some measurements (and it was supposed to be only "m", not "mm"). Better eat breakfast before I post. ;)
Microns, we're still talking constant pressure/temp/humidity environments. But maybe it isn't important if the movement is repeatable, is it?
I've been working with a CE Johansson Saphire, could hardly guarantee that small tolerances with that monster either. Correct me if I'm wrong, but did you really mean 600 dollars?

There are handful of designs that would work well, but some are clearly better than others depending on your design requirements.

-What is the approximate range of travel of the larger, scaled up design? Some architectures are not easily scalable. For 1mm I may look into using a cam mechanism, but this is not very practical if the ultimate travel is 100mm.

-What kind of speeds are you taking about? 1, 10, or 100 mm/second?

-Matt

600$ is the price he gave. We have not priced anything out yet, but we know it is not going to be enough. The ball screws are going to cost a fair bit, even for just a few inches. We are in the proposal stage as of right now. We have access to CNC mills and a machine shop and several techs so labour is not an issue.

The cam mechanism is under consideration, but like you said it can not be scaled up. The first time he ran this by us I though of using a high end RC aircaft servo. Then we asked about the loads involved.

Travel time was not important. Most likey less then .5mm/second as he wants to be able to fine tune it. The important part is that he wants to be able to jog back and forth a few microns at a time.

Just build up something like these for your closed loop measurements:

http://www.zygo.com/?/products/zmi/

I found this just typing it in a search engine. These measure to a couple of nm, so well beyond your actual needs. Probably some used ones out there.

Most of these are laser based and use the concept of measuring the constructive and destructive interference of light waves. Basically, you get a "peak" and "trough" brightness in the light signal.

From a practical standpoint, the shorter the wavelength of the laser, the more accurate (and more expensive). Blue light is approx 460nm (about 1/2 micron) and red is nominally 630 ish nm. 1000 nm = 1 micron. Red solid state lasers are practically free now days, while blue / green cost real money.

Just a tidbit, this method is how the official standard meter is defined.

Keep in mind, you only need movement in 1 dimension, over a very short distance. I think the hardest part will be picking your reference points to mount and measure from. Small temperature changes will be both your friend, and your challenge, regardless of movement method selected.

http://www.zygo.com/?/products/zmi/


Site has been booked marked. Thank you.

I have talked to two people in the school outside of the project group/advisers. They both know their stuff. One teaches machine design and the other is a electrical tech. Both told me more or less that the expectations of the project are high given the amount of money. The tech said go ahead and design to the expectations of the 'customer'. Then spec each piece of equipment according to what we come up with. The price can then be asset and then the customer can either give a thumbs up and pay or scrap the project.

I have little experience as I am just a 5th year ME, but I would say something like this would be 1500-2000$ minimum.

I hope you don't mind, but I would like to encourage you toward success with your existing constraints. Let's not worry for now so much about what people will do with something "someday" and focus on the basic specs.

University projects tend to start with a "basic concept" and continue to evolve and improve over time, with small incremental budgets.

Specs you have posted

- 1 mm travel in one
- essentially no slop in any other direction
- 3000 N
- approx 1 micron travel resolution
- speed does not matter
- student sort of budget - goal $ 600.

Let's for fun, start with the idea of using thermal expansion as the driving force for the movement. Since it is easy to obtain, and has a fairly high coefficent of thermal expansion, let's try Al.

From this link http://hyperphysics.phy-astr.gsu.edu/hbase/tables/thexp.html

Al will expand approx 24 x 10 - 6. (please double check all work)

If we start with a 1 meter long block of Al at 20 C, and heat it to 120 C, then

100 C x 1000 mm x 24 x 10-6 = approx 2.4 mm. Technically, Al's expansion is slightly non linear over this range, but this is well documented.

So, to pull this off, all you need is
- An Al block 500 mm long
- 2 pieces of granite - one for the base, one for forming an L shape at the end
- set the block of Al on the granite, firmly fasten one end only to the piece of granite on the other end (L )
- Set up a heater and a good insulation blanket to bring the temperature up uniformly

If I am thinking clearly, which is always questionable, it will expand approx 1 micron per degree C. That should be easy given some stabilization time.

A big block of Al firmly mounted on one end to a granite block will have very little backlash or ability to move in the other 2 axis.

Ok - now go get an A - or prove me wrong. :)

It goes outside the dimensional boundaries, but is clever and simple. I am taking what you did there and am running with it. I hope I don't trip.

Force/micron=(2.4mm/1000mm)*75GPA*x^2/2400micons

For =3000N/1 micron, x would be 20cm square.

So a max force of 3000 N would move it a maximum of 1 micron if the block was 20cm square.

I hope I did that right.

Don't forget to send us some feedback, pictures are always welcome.

Hi

Please double check all of my calculations. It has been a while since I had to actually depend on the engineering training, and I did the math in my head. Too many / not enough zeros are always possible in my work.

If it were me, I would use a factor of 10 x the cross sectional area that you think you need. This will help even out the temperature distribution, and reduce backlash. Yes - backlash.

Your calculation shows part of the design challenge that I did not think about - deformation of the Al block itself. When it is extending, it might be up against a force of 3000 N, but when in retraction, the force could be considerably less. This could easily add 1 micron of "backlash" to your system. If for some reason, the load went from compressive to tensile during this shift, the backlash effect would be at least 2 x this.

If my thought process is clear this morning, increasing the cross section by 10 X will reduce backlash by an approximately equal amount. (assuming you can distribute the force across the face).

Good Luck, and we are all curious how it works out.

The $600 budget may be enough. If he wanted 1-5 micron precision over a longer distance, it would be a different story, but over 1mm, I think a lot of standard components will work. The key will be to heavily preload everything.

Four linear motion rails can be oriented in ways such that when you connect the blocks together, they will be under heavy tension in all directions. Ballscrew nuts and support bearings are easy to preload as well.

Use leverage( re: Archimedes " Give me a lever long enough and I can move the world"),

Set up a ballscrew to move a 4-bar linkage.
4-bar linkage to keep everything parrallel.

The point of application ( 3000N at 1mm) would be close to a given pivot point. The ball screw would be far from the pivot point.


For distance measurement put a small mirror on the linkage. Shine a cheap laser pointer on the mirror. As the linkage and mirror move, the landing point of the laser will move. Set up some cheap optical sensorsto catch the reflected laser, equally spaced apart, for example at 10cm spacing. In front of each sensor is a piece of lucite that is 10cm long. So as long as the laser is shininng anywhere in the 10cm zone, the sensor is activated. Calibrate each individual sensor to a distance moved.


keep us posted

Use leverage( re: Archimedes " Give me a lever long enough and I can move the world"),

Set up a ballscrew to move a 4-bar linkage.
4-bar linkage to keep everything parrallel.

The point of application ( 3000N at 1mm) would be close to a given pivot point. The ball screw would be far from the pivot point.


For distance measurement put a small mirror on the linkage. Shine a cheap laser pointer on the mirror. As the linkage and mirror move, the landing point of the laser will move. Set up some cheap optical sensorsto catch the reflected laser, equally spaced apart, for example at 10cm spacing. In front of each sensor is a piece of lucite that is 10cm long. So as long as the laser is shininng anywhere in the 10cm zone, the sensor is activated. Calibrate each individual sensor to a distance moved.


keep us posted

That is a good concept, especially for monitoring the location. It will take some thought and care to locate the sensors in a way which captures the leverage action of the "large movement side" of the lever for movements on the "position side" of only 1 micron, but the general idea seems sound.

It will take some thought and care to locate the sensors in a way which captures the leverage action of the "large movement side" of the lever for movements on the "position side" of only 1 micron, but the general idea seems sound.

Locate the sensors on the opposite wall in a large room like a gymnasium, this is a student project after all.

You guys are talking a little bit over my head on this one but I had a thought and thought it couldn't hurt to post, maybe it will give someone else some ideas.

Since the 3000N load is along the axis you are working with you can orient the axis up and down, therefore gravity will eliminate any backlash.
Use four pins and bushings to keep the weight from shifting, then just use a compound pulley system to lift and lower the weight, if you have enough pulleys you can lift and lower huge weights. You can measure the length of pull of the cable or chain or whatever you use and calculate the distance of travel for each micron of movement of the weight.

You could also use the same principle in other axis, but you would have to attach the cable to both sides of the weight so that it pulls from either direction.

Jim