The cost of motion

[Jcadwell]

Building the physical structure of most of the CNC routers I have seen doesn't look like too daunting a task. But how to move each of the axes becomes complicated quickly. I'd like to quantify the cost effectiveness of ways of linear motion. What are the practical speed limits of an ACME screw for a given pitch? What tolerances can you expect to hold for different types of drive systems?

Right now I want to build a 2X4 router that will have repeatability of about .005 (is this reasonable?), and a rapid speed of about 150IPM. I'll be using 2500RPM servos with 400oz and geckodrives. I want to cut wood and light plastics. Maybe PCB routing for power related projects if possible.

I'd like to work primarily in MDF with an aluminum deck. I was looking at 1" ground rail with linear bearings for the axis supports.

The three options I'm aware of involve screw drive, rack and pinion, and timing belt.

At this size and speed screw drives (acme and ballscrew) seem cost prohibitive, and limited by speed.

Belt drive looks great on paper, but I'm thinking the belts may not be as rigid as I'd like. Big belt also costs big money, especially with requisite bearings, pulleys, and shafts to drive it.

Rack and pinion looks easy. For less that 200$ you can buy enough rack to outfit a 2X4 with dual X axis rack and single Y rack. What kind of backlash can I expect?

I'm really looking for the most cost effective solution. It doesn't have to be the easiest. I've looked at the 5/8th inch by 11 allthread down at the hardware store and wondered if it would work... I don't want to build a machine that isn't precise just to save a buck. But I also don't want to hand out money for top shelf parts if I can get the same performance out of something that costs less. I seem to be right at the cusp of speed where linear screws top out (other than ballscrews) and where other drive systems take over.

Any thoughts?

Any leads to great deals? Thanks.

[Apples]

i've been planning a eight foot by four foot plasma table and was thinking of using push bike chain an gears to drive my gantry

[vacpress]

i am planning a rack and pinion. it seems most cost effective - unless i find some 1 or 2tpi ballscrews on ebay right around when i am gonna order the rack, pinion, friction belts, and reducer pulleys.


yep.. for most practical purposes you will find yourself still needing toothed belts and pulleys. pulleys are shockingly priced - $30-$50 each. belts are like $10-15 for the 12" loops i want.

so, really its just that precise linear motion requires money. repeatability and real accuracy are difficult to achieve on a budget.. thats why i expect to spend $300-500 on just belts, pulleys, rack, pinions.

for a 3ftx3ftx10" table designed mostly to cut high density modeling foam, REnshape, and other fairly easily machines materials.. the parts this machine is going to cut are going to be mostly complex 3d shapes though, so i need speed and repeatability..

plan on at leat $1500 for a decent table...unless you have access to cheap or free components...

[ger21]

Look at InventIt's router. He uses 5 start 1/2-10 acme, I think. Gear the servo down with pulleys and a belt, and you only need to spin the leadscrew at 300rpm to get 150 ipm. With 400 oz's, you can probably go even a little faster. Check MSC for the acme, not too expensive. Also, for belts and pulleys, check Stock Drive Products. You shouldn't have to pay $30-$50 per pulley, only about half that. (For servo to leadscrew)

[Jcadwell]

So the pitch divided by the number of starts gives the normalized pitch? So a 10TPI screw with 5 starts would really be like a 2TPI screw? Is that correct?

I'm leaning rack and pinion but the lead screws look much easier to set up.

What kind of repeatability will the solutions exhibit?

[ger21]

So the pitch divided by the number of starts gives the normalized pitch? So a 10TPI screw with 5 starts would really be like a 2TPI screw? Is that correct?

Yes, it's number of threads divided by number of starts.

I think repeatability is a function of your machines construction and tightness.

[Jcadwell]

So would one of the 1/2-10 5 start acme screws hold 5 thousandths with a given nut? Would the limitation be the screw or the rest of the machine at that point? Thanks.

[ger21]

Are you taling about moving 6 inches and being within .005 of the six inches, or moving 6 inches forward, then 6 inches back, and being within .005 of where you started? The cheaper screws I've seen are accurate to plus or minus .009/foot. So, if you move 1 ft, you're only guaranteed to be within .009 of that foot.

But, if your talking about being able to return to the same spot over and over, to within .005, that should not be a problem, provided you have less than .005 backlash in your screw and nut. Or that your machine doesn't twist or flex more than .005 during normal operation.

I think a 20° or 30° temperature change will cause you to be out more than .005.

A motor, nut and leadscrew, should always (generally) be able to return to the same location as long as there is no backlash in the system. And if you need more accuracy than the .009 I listec above, you can get Nook screws from MSC that are .003/ft for about 20% more money. (About $50 for 6ft)

[ger21]

Have you seen this machine here? http://www.axxustech.com/Axxus4x8/axxus_4x8.htm

I like this system, but not sure about the cost.
Rack and pinion can be tricky to get rid of backlash, or so I've heard.

If you didn't look at Invent It's thread, it's here. My first Router / Plasma Table Buildup...

With the 1/2-10 5 start and 400+ oz-in motors, he says he can get up to 350 ipm.

[avsfan733]

has anyone ever used anything similar to adjustable cam gears on cars to adjust out the backlash from systems like the rack and pinion?

[Jcadwell]

I had seen both the axus and the InventIt machines. With repeatability I was referring to returning to a position, not being within .005 over a distance of six inches as you suggested. I want it to be consistent, (precise?) not necessarily excruciatingly accurate If that makes sense. I just don't want to build a system with .01" of backlash so that it is loose enough to run fast.

The rack and pinion looks really really cool. The rack is about the same price as Nook lead screw. The Nook that MSC-Direct has is only single start, correct?

I found precision Acme screw on McMaster Carr that looks like it will be a winner. 12 feet of 4140 1/2-10 5 start is $79.88.

Looks like the most affordable solution yet... Order 3 PET plastic nuts ($26 each) and I have enough rod and nuts for all three axes.

Any comments? I am considering trying to spin the nuts instead of the screws. Thrust bearings are about $5.00 for a 3/4" bore, which could be made to work with the nuts easily. I could then mount a pulley over the top of the nut. Seems like this would be easy, especially on the X-axis, as you would only have to retain the lead screw with a single bolt down the middle to hold it in position. Any thoughts?

[ger21]

When first designing my machine, I was going to spin the nuts. I don't think it would be very hard to do. But, since I can't even find the time to build my router, I'm going to go the standard leadscrew route for now. The nook screws from MSC are 5 start. They're labeled as 1/2-2. http://www.mscdirect.com/PDF.process...Section_Id=451 Number 00277194

I believe the main thing with repeatability is to make sure you have Zero backslash. Everywhere, including no play in any of your axis' movements. If everything is tight, you should be fine.