Aloha

I'm a student from Sweden with an (un)healthy DIY spirit.:rolleyes:

For a long time a CNC router was something I would build "somewhere in the future". But as a DIY:er the number of projects would rapidly build up. And now I feel the time saving aspects of a CNC router could help allot.

I hope it's ok for me to use metrics.

I just recently (~2-3 days) ago begun to do in-depth reading about DIY CNC-routers. And I feel like I got a decent grasp on the subject. But as I lack the trial and error, practical experience I need to ask you a thing or two.

The issue is the old one: accuracy/money/size. I'm designing a three axis router with four fully supported linear rails for every axis. Maybe only two for the z-axis, time will tell. On top of that twin ball screws with two 260 oz motors each on every axle (again maybe not z-axle). The router will be used for circuit boards, wood, alu and possibly composites (glass and CF).

The problem is that I would like it to have 0.01mm accuracy on a routing area of 1 x 1 meters. Or let me change that. The problem being to do it with a students economy. I think I have a decent sources for rails, motors and most things needed with one big exception. The ball screw.

The screws on http://www.homeshopcnc.com have a good price but would they be accurate enough? The screws for x/y-axles will have to move about 1000-1010mm + mounting should end up somewhere around 1100mm?.

I can't get the precise measurements quite yet, but in a few days I will have access to a CAD application and then I can tell more precisely.

Also for another question: Do you think prefab carbon fiber rods of 12mm have the accuracy to be used as linear rails? Would still be fully supported. I'm hunting dynamic weight you see. :cool:

/Cheers

i don't have any experience with homeshopcnc stuff- but i am very interested in your build.

Welcome to the board, LUM. I see this is your first post/thread.

Welcome!

Scott

Homeshop cnc ballscrews are just fine. I started building my unit on jan 8th and had it finished by the 19th. Accuracy is fantastic. The ballscrews have a travel 0.2 inch per turn and with driver board and stepper motors (1600 steps per turn) you can break the distance down incredibly, basically each step will move the unit .003 of a mm. I used the non preloaded nuts and there is no physcial play that I can find. I can run 5 individual lines 670mm long and cut each one perfectly on top of each other time and time again. There product is very good
Steve

Sir,

Your accuracy of .01 mm (I think in terms of 1/1000 inch so this is .3737 of one thousandth) is impossible to meet over the 1 x 1 meter range. That accuracy requirement must be the SUM of all errors, straightness of X,Y and Z axis rails, orthagonality of the three to one another, accuracy of your measurement system, temperature effects, lost motion, elasticity of the various members etc. Some claim that to get a given accuracy, you must plan each menber to be 10x better!!

I said impossible, as it requires very expensive measuring equipment to verify the results, and very high grade construction of your machine. As an ameteur
an overall accuracy of 0.5 mm might be more reasonable and achievable.

Regards,
Jack C.

Sounds like you're aiming pretty high in all aspects, good luck! if you get halfway there you'll have a quite useable machine!

~Steve

jcc3inc: I was afraid of that. The straightness of X and Y I think I'll have under control. Also by keeping the dynamic mass to a minimum and having 4 motors (the 260oz ones have the best power to weight ratio of the ones I looked at) and two screws each for the X and Y axles should help some. Keeping the inertia minimum with good stopping power. I guess I should change my aim to at least 0.1mm precision. That is if the screws allow that.

(Hey aiming high for the sky! :D )

As far as Z goes I'm not 100% sure yet, the precision is not as crucial here either. I'm thinking two 12mm c-fiber rods and one or two screws with a motor each maybe. If I go for one screw I'm going to aim for a c3 milled at least.

I'm trying to keep the dynamic parts as balanced as possible too with the routing bit as close to the other axles as possible. Everything to minimize unwanted movement. Even have a few ideas with a moving counterweight but I think it'll add too much mass.

In time I will aim to change most of the dynamic structure to c-fiber or honeycomb sandwiched alu if I would stumble upon any...

The static structure I want as dead as possible. Possibly I could cast it in concrete or sandwiched construction filled with sand.

I'll probably use a Dremel for the routing.

The speed (IPM) is not too important if maybe I don't desire the slowest of the slow.

A new issue arose too, I was accounted for the controllers but not the drivers. 10 Gecko G203V:s cost a bit... :tired:

Other than that I will build a hush-box/dustcover (with dust suction) for the router.

P.S. The drawings are delayed. The SolidWorks machine had some issues with licensing. Think they said something bout the network. Now I'm booked on a AutoCad machine but with a delay of a few days. Good thing I guess is that I've spent many hours in AutoCad before ( long time ago in high school ) and only minuets in SolidEdge. Bad side is that I wanted to learn SolidEdge a bit better. Oh well...

.....an overall accuracy of 0.5 mm might be more reasonable and achievable.

I understand the point that you are trying to make but I think if this student does his homework, he can expect better results than this. True - hitting .01MM might be a bit of a challenge and might be a bit more expensive to achieve but it is achievable at the hobbiest level.

I agree that hitting .01MM over 1000MM will require accurate components - not likely to be affordable to a student but you never know.

I think .03MM over a meter would be a reasonable target for an ambitious student's budget.

Scott

.03mm would do good. It's mainly for circuit boards and some precision components.

Any idea on the carbon fiber rods? I figure as they're sold in bulk most of them are probably only made in a few places somewhere in Asia. Never seen a accuracy claim on them though...

While I'm at it. Is it hard to mill in copper? I know it's the wrong forum but I don't feel like a new thread is necessary. I ask because I'm thinking about DIY blocks for water cooling. Soft and good heat dispersion, shouldn't be too hard?

And if the stepper housing would be water cooled do you think it would be ok to have them in an tight enclosure? For sound dampening purposes. Or do they need to breath? Maybe a small airspace would do? I'm thinking small boxes of thick mineral/glass wool around the motors.