I'm trying to make a 5'X10'X3' (xyz) milling machine in order to make woodcarving and automotive body panel molds. Also trying to keep the prioce down as low as possible.

What parts do I need to and what manufacturers should I be looking for?

bearings?
ball screews?
stepper motor size? torque? type?
and others that i no doubt have missed...

That's a pretty *large* machine you have in mind. Why do you need the 3' z axis?

The problem is that the frame for such a wide and high working envelope must be very rigid. This means expensive.

If you could manage a z axis height of 12" or so then you stand a good chance of achieving your objective of keeping the price to a reasonable amount.

You might want to look at rack and pinion for the 10ft axis. A ballscrew that size would cost a fortune.

I saw an idea a while back, that might work for you. Gary is right about stiffness being important, so here's an idea for keeping down the price.

A garage or shop already has 6 surfaces (4 walls, ceiling, floor) which are built to be reasonably rigid. Attaching to those walls and ceiling might be an appropriate option.

I had my own idea, a while back, about using garage door parts (rails, drive screw, etc) to build a really big router which would hang from the ceiling. Since they're built to be linear motion components already (and can be beefed up and made of higher quality with relative ease) this should be a workable solution. By having an opening on both ends of your shop, you can have an "infinite" X axis length. Since your gantry would be hanging, you could build it as a truss for stiffness, but relatively light weight.

This should be a workable solution for you, and Z axis travel would be limited by the length of your rails, rather than the clearance available...the floor would be the absolute limit, obviously. Placing panels on "sawhorses," or some other type of support could be a workable solution, to raise it to whatever height you require.

What type of resolution do you require, and what level of accuracy? Are we talking about .001" or 1/8" accuracy? That will determine the cost of the required components.

-- Chuck Knight

The problem is with a Z axis that long the support for it would need to monsterous. Because of its size it would also limit what you could mill. I would guess all you could mill would be 3 foot tall volcanoes, due to the Z axis support structure. I guess if you had a rotationg tool holder on the end of your Z axis , you might gain some better manuverability. I dont know how you would program that though.

The other option is, of course, to raise your workpiece up *to* the router.

I agree that a 3 foot (that is what you meant, right?) Z axis travel is a LOT, but an 8 foot Z travel is just bizarre.

Having said that, even just "hanging" a router from a cable from the ceiling, would result in a very accurate Z height control. The problem would be with lateral movement, and that can be minimized by triangulation. Your Z axis might end up being 5 feet across, which means that for a 5 foot Y axis travel, your Y axis would have to be 10 feet wide...but this is not insurmountable, especially if you're building onto the existing ceiling and walls.

Using a truss design would give you "skeletal" support that would reduce the weight...they can be built "Erector Set" style with little trouble, and with pretty good results. Having it travel along a simple set of Y axis rails would seem almost trivial, so long as everything is built good and stiff.

If designed right, it might not be an insurmountable problem.

How's your project budget?

-- Chuck Knight

Thanks for the suggestions guys.
I 'm looking for more basic suggestions for the time being. I'll deal with stiffening the system later.
The accuracy would be ok if it is +/ 10mil.
Regarding the ball screws. Do they have to be ball screws? are there other options? ie acme screws?
what are the choices out there and dis advantages/advantages? Where can i find reputable manufactures? links so that I can look?

What manufacturers offer ways to control the backlash? Are there backlash compensation systems to remove this?
if so what are your recommendations so that I can start looking for what i need.


is there a formula to calculate the holding torque needed for "X" amount of wieght?

crazyman, for your application, I would not use a ball screw, Too expensive!. A timing belt would work well. and cost much less. Look here for an example,
http://www.cnczone.com/forums/showthread.php?s=&threadid=3816&highlight=cnc+painter

I am not sure I understand +-10 mil. Is that +-10millimeters? (+-.397") If so, you could tie a rope to this and probably be accurate enough. If you mean +- 10mills (+-.003937") over a 10' bed, you had better rethink the whole idea of low cost, DIY.

Routers like this would cost about $150,000+

crazyman are you intending to carve the molds themselves or carve foam to use in the making of the molds? The answer to this question will greatly impact how you have to build your machine. Iffin you'll tell me the answer I might reveal a design I wanted to build sometime for the ? same machining. I'll just let you build the machine and you can make all my parts as payment!!:D Hey am I a nice guy or what??;)

Mike

Well, I know Genie makes a garage door with an ACME screw drive. While it's probably not of the greatest quality, it's readily available. That would include rails (X axis), screw drive, and limit switches.

There's nothing wrong with ACME thread, or even allthread, if you just use it right, and know what to expect.

As for braking forces, the losses are high enough on anything other than a ballscrew, that they're self braking.

A timing belt would also work well, as would a rack and pinion drive.

I suggest you design your Z axis first, but with an eye towards the way you're going to move it around. "I'll work on stiffening it later" is not a good way to design this sort of thing, as everything is balanced. Keeping it light will reduce the mass, and consequently the inertia. That means that you'll require less power to move it, which implies cheaper screws and motors, which implies other stuff...etc, etc, etc. Like I said, it's all interrelated.

-- Chuck Knight

Oops, I forgot to address backlash.

Backlash exists in all systems, including ballscrews. There are just ways to compensate for it.

One way is a "spring loaded" nut which pushes against opposite sides of the thread...this is self compensating for wear. Another is a compressible nut. Basically, they're all just 2 part nuts that "squeeze" onto the thread, and eliminate the backlash.

Another option, though less attractive, is to compensate for it in software. TurboCNC has this compensation function built in, and from what I've heard it works well.

-- Chuck Knight

P.S. I used a plastic antibacklash nut of my own making.

chuck,


i am not sure a garge door system would have the accuracy to use for cnc work at all. i remember tge garage door at my parents house - it rattled, screached, etc. the tracks the bearings ran in probably had 1/32 to 1/16" slop in all directions. all the components were made of bent sheet metal if i remember correctly.

i have an idea for crazyman: get some 6" to 12" steel square tube\rectangular tube\i beams. whatever is easiest\cheapest, and also the best quality for price available easily. weld it into a large non-moving gantry. put another 2 pieces on the ground to be your moving milling bed. go from there.. for motors you would need big steppers or servos. budget at least $365 for 3 gecko drivers, and 200-500 for steppers or servos. then ou need a good large powersupply - another 200-350. this is probably about 1 thousand dollars allready, and you still need to get linear motion components, a spindle, tools to build this thing, possibly 3 phase power for whatever spindle you end up with, and if not, a convertor can add another 500 easily... your machine parameters are not those of a machine that can be built without great thought.

hers something that i came up with after almost no great thought:

get together $150 and goto lumber yard. buy the strongest, straightest 4x4 lumber they have. its gonna be like $15 to $20 per 8' piece. thats find. get like 10 of these, and take them home..

build a large, strong wooden scaffolding about 12'x6'x5'. make it as strong and accurately as possible. it should be within a 16th! if your wood isnt straight, consider getting a planer.
when it looks strong enough - make it stronger. this thing has to be built like a deck that 100 people could stand on, only much more accurately.

within this frame, you can now build a system based around whatever crackerpot thigns you can think of. for starters, i would think about using something silly like rewound alternators acting as giant servo motors. use steel roller chain as your linear actuators. just make it all real strong. put a big mean tensioner on the thing. use gears from your old 10 speed as idlers. the 10' length is the real killer still.. getting something to move 10', with speed, with accuracy is hard. i couldnt even begin to take on such a task without thousands of dollars. i wouldnt want too. i mean.. i also bet you could get pretty good results with 4" iron pipe, the straightest ones available, and lots of rollerblade bearings....

the hard part about saving money by building yourself is building accurately.
i hope/bet noone reads this. :)

i am not sure a garge door system would have the accuracy to use for cnc work at all. i remember tge garage door at my parents house - it rattled, screached, etc. the tracks the bearings ran in probably had 1/32 to 1/16" slop in all directions. all the components were made of bent sheet metal if i remember correctly.

True, but that doesn't mean that the rails weren't straight, and the screw/nut wasn't useful.

Think of it as a starting point.

I have 2 rails sitting out on my deck, right now, and more than a few ideas for it. They're pretty stiff, and should be able to be adjusted for reasonable precision...same thing with the screw.

It's just a fun idea. Considering that the walls of the building would provide the structure to keep it stiff, the problem then becomes the drive system.

-- Chuck Knight

Originally posted by buscht
*snip*
I am not sure I understand +-10 mil. Is that +-10millimeters? (+-.397") If so, you could tie a rope to this and probably be accurate enough. If you mean +- 10mills (+-.003937") over a 10' bed, you had better rethink the whole idea of low cost, DIY.

Routers like this would cost about $150,000+

I think 1 mil = 1/1000 of an inch (1 "milli-inch"). It's a common unit in circuitboard CAD programs. The pins on a PDIP chip is 2.54 mm, or 100 mils, apart. The bigger SMT parts (forgot what they're called) have 50 mil (1.27 mm) pin spacing.

But to keep this accuracy, or even repeatability, over 5'10'3' work space, will be very expensive and/or difficult:

1) If you're only going to cut foam and balsa then maybe the machine frame won't be extremely difficult. I still think it'll be the most difficult part, though.

2) My guess is timing belts would be your best option for driving the thing.

Arvid

Sorry I haven't replied lately. I've be sick. Great responses guys. keep em coming.
My main reason for making the table so big was to do full size automotive panel molds not the actual finished part and also some woodworking of cabinet doors and other artsy fartsy stuff.
Even if this turns out to be greater than my budget and I have to downsize the table I think the discussion we are having is pushing the envelop on ideas.
Let make it 10'X10'X10' now...lol

Regarding using wood. I though of that but the thing would warp quite a bit as the humidity of the wood changes. Maybe some more dimensionally stable type of wood/laminate products? Maybe silent floor product? Not sure if these are much more stable either?

I have seen a large table use raised stationary I-beams as a raised gantry and the spindle moved on top of the I-Beams in the say Y direction, but then you may run into a heavier spindle assemble to contol both the X and z Directions.

The +/- 10mil mentioned means +/- .010"