Retain Belt Drive or convert to Ballcrews?

[steve_p]

Hi, I hope this is in the correct area, it has to do with a commercial router, but I'm doing a DIY rebuild, so I assumed this is the correct spot.

I'll give you a bit of background first- I bought an old Gerber AR400 CNC router (4' x 4' bed, vacuum table with pump, 800W Perske spindle with VFD, proprietary controller) a couple of years ago on eBay, and I have to say it was really easy to get up and running despite my complete lack of CNC/G-code knowledge. I still know hardly anything about G-code, since I just do my drawings, set out the toolpaths, and the machine does its thing. I run it in my general fabrication/modelmaking workshop, and I use it in almost every job for anything from small aluminium brackets to complete assemblies, and often to make alignment jigs to work to when using non-cnc machines.

I have gradually been more and more itchy to move up the ladder a bit- the main things would be (1) a minimum 8x4 bed would be much better, and (2) the speed, particularly moving from point to point between cuts, seems painfully slow.

It's not running in a production environment, since virtually every job it does is unique, but the bed size in particular is becoming a serious restriction. And now that more people know I do cnc stuff, I'm getting asked to do bigger stuff (this week, for instance, I've been asked to part-mill, part cut, a 2.5m x 1.2m 5mm thick ali sheet in a very intricate persian pattern.)

Anyway, I started looking at larger machines, but as I'm sure you know, they're extrememely pricey new. And used machines seem to crop up quite rarely, and any that do go for high prices because the supply is so limited.

Then recently I noticed another machine exactly the same as mine on ebay, which I won really cheaply (probably because the seller didn't use the word 'cnc' in the listing), and it turned out to be in much better condition than mine. The plan was to bolt the two beds together to create a single long one, although I didn't really know how feasible it would be. I thought this would be preferable to making a bed extension, since they are already stress-relieved and milled flat.

The controller currently on the machines doesn't use homing or limit switches, it just uses the current position as the zero point when the programme is dowloaded. And it doesn't know when it's reaching the far end of its travel, it just runs into a mechanical stop and stalls. Which means, I think, that it would just keep running onto the extended bed without any problem.

So I checked the mechanics, and all axes run on pairs of 1/2" fully-supported round precision rails, with recirculating ball bushings. I found a firm in the UK that can supply imperial sized (and drilled+tapped) rails that would be a direct swap for the current ones, in any length. At a far cheaper cost (GBP70 each for 3500mm) than replacing with profile rails or whatever. The supports for the rails are welded to the machine perfectly straight, so it seems reasonable to keep the same rail type.

My main consideration is the drive system. The machine(s) currently use synchro belts for the X and Y axes, which to be honest, I've never noticed a problem with when cutting. Each side of the gantry is driven by its own belt, although both belts are driven by a single stepper, via a solid shaft across the rear of the bed that carries both driving pulleys.

I'm thinking that if I want to increase the speed of the machine, I should probably replace the motor drives with modern microstepping ones, and run it off Mach3 or similar. But would the belt drive be a limitation? Would changing to ballscrews, and perhaps even twin motors on the long axis, produce a benefit worth the (quite considerable for 2x 3000mm lengths) cost? The current belt system is well engineered, with heavy cast and machined pulley/motor/reduction gearing housing blocks, and would only require me to fit longer belts along with the longer rails, and I'd be done. (The controller and drives could be upgraded a bit later, so I could keep running the machine in the meantime.)

Does anyone here have experience of the comparative drawbacks to belt drive as opposed to screws? I know Gerber also made an AR600, which was 8x4, so the length of the belts is not a problem. But I also know that when Gerber then made the next generation Sabre routers, they switched to screws, so there must be some reason for that. Is there a speed limitation to belts(By the way, I dont think the accuracy is a problem with belts, at least for what I do- parts seem to always come out within 0.1mm or better of the drawing size, which seems good enough for a router.)

Oh, and out of interest, do 1/2" round rails seem adequate for a machine of this size?

I'm particularly interested in hearing from people who actually have experience of the various drive systems, rather than just opinions formed from reading on the subject.


One other thing- I'd be interested to hear opinions of those 4kw water-cooled spindles from China you see on eBay. Even if they don't last that long, they cost about the same as a spindle rebuild, so you could chuck them out (or replace their bearings with decent ones..?)

sorry for the long post!

[johnmac]

Nothing wrong with belt drives. You are correct on the controllers. Gecko drives are highly respected here on the Zone. Mach 3 is a good choice, too. i would add limit switches for homing, but since the machine didn't come with them, you probably can't break anything. If the rails are supported, they should be fine.

John

[Kipper]

Why not combine the two?

I'm nearly finished with this build and have to admit to having a larger twin ballscrew in the works.. I find belts get stretchy over around 2m in length.

Have fun

[steve_p]

thanks for the replies, I'll be interested to see whether anyone else has used belts on an 8 foot ot longer machine successfully.

thanks again

[steve_p]

Just thought I'd add another question- I've just checked out the motors/gearing, and they are 150 in-oz 6-wire motors geared 34:1 to the belt pulleys, which seems quite a lot, which is good for torque and resolution, but obviously not as good for speed. If I replaced the motors and drives with microstepping Geckos, would I be able to increase the rapids ok, with this level of gearing?

[CarveOne]

Originally Posted by steve_p thanks for the replies, I'll be interested to see whether anyone else has used belts on an 8 foot ot longer machine successfully. thanks again

With 8' belts I would think that the belt needs idler rollers at a few places (top and bottom) to keep the belt from sagging between the end pulleys. Sagging will cause the belt to jump up and down during reversals and the belt will act as if it has free-play. With the idlers it will be easier to tighten the belt and eliminate the issue, at least until the belt stretches with time.

CarveOne

[steve_p]

yeah, the belt stretch/sag is the reason that i'm hesitant about keeping the belts. I think the way they're mounted in the machine at the moment is designed to help prevent that, because the upper half of the belt is only about 5mm above the top surface of the main structural steel member, so it cant really sag much anyway. The bottom runs inside the steel box, so it can, although I guess it's always the top section doing the moving. I'll try posting some pics of the motor/gear/belt setups to give people an idea of how it currently runs.

The engineering seems very solid, the motor/gear housings are built from cast and milled aluminium forms, so I feel I should keep as much of it as possible. At the moment I'm tending towards keeping the Y axis (gantry, in case I've got my axes mixed up) intact with the belt setup, since it works fine as it is. I will probably change the motor to a more powerful type.

The Z axis is driven by a screw, so there's no need to change that, although a greater height would be useful, so I am thinking of adding blocks above the gantry pillow blocks, which might then mean I would have to add a longer pair of rails for the extra Z travel.

If I did jack up the gantry, an added side-effect would be that I'd have plenty of room to fit ballscrews where the belts were, and even use the same nice milled mounting area for the ballnut to attach to.


Sorry to ramble, but it all comes down to this-

I am increasing a belt-driven machine from a 4ft to 10ft x-axis travel, and need to know :

1- whether I really should be looking at using ballscrews (or acme) rather than belts. Are there problems with such long screws?
2- whether I'd see a benefit in replacing the 150 in-oz steppers (1.7V, 4.7A, currently geared 34:1) with new motors. I am sure I'll be using Gecko drives and Mach to run it, so the question is whether the existing motors would be fine or not
3- If I did use a screw along each side of the x axis, I presume I'd be using a motor on each- how do I keep the gantry perpendicular to the x axis? (surely when the power is off, one end of the gantry could be pushed a small amount to skew it- do people use a calibrated limit switch on each screw so the controller can set the squareness when it powers up?)

Any advice would be greatly appreciated.

pics: (sorry if I've done this wrong, it's my first time, be gentle!

belt position:
http://www.cnczone.com/forums/attach...1&d=1247917736

belt clamping:
http://www.cnczone.com/forums/attach...1&d=1247917736

gantry mounts:
http://www.cnczone.com/forums/attach...1&d=1247917736

x axis drive:
http://www.cnczone.com/forums/attach...1&d=1247917736

y axis drive:
http://www.cnczone.com/forums/attach...1&d=1247917736

z axis (with 1hp perske 3 phase spindle attached)
http://www.cnczone.com/forums/attach...1&d=1247917736

[ger21]

10 ft screws would need to be pretty large diameter, to prevent whip. And you'd want much larger steppers, or servos with gear reduction.

MAch3 can automatically square the gantry when homing, using a separate switch on each side of the gantry

[steve_p]

thanks for the quick reply Gerry, good to know Mach can do the squaring thing. I take your point about the whip, I did wonder about that- 10ft is a long unsupported length. And the price of very long, very large diameter screws would be very very large!

I do wonder whether it is worth just trying the belts first- it will only cost the price of two belts (21ft long each!) and a bit of time, and see how it works out. If it's no good, I could try the added idler idea mentioned above, before making big and expensive mods to change to screw or rack and pinion.

I just looked at the pulleys on the x axis- it turns out that one rev of the drive pulley moves the gantry 10" exactly (254mm), and since the gearing to the stepper is 34:1, then the gantry moves 7.47mm for every rev on the motor. I'm guessing the current drives are half or quarter step at best, but new drives would be microstepping (a Gecko 2000 step-per-rev would equate to 0.0037mm travel per step, which is a lot better than I need for anything I use the machine for- 0.1mm is adequate for virtually anything I do). I doubt the belt setup would have the resolution to match that anyway.

Do you know the max rpm Mach can reliably drive a motor? I guess this would allow me to work out whether updating the motors and control with the current mechanical setup would be able to give me a faster machine.

thanks again for any advice.

[ger21]

Originally Posted by steve_p Do you know the max rpm Mach can reliably drive a motor? I guess this would allow me to work out whether updating the motors and control with the current mechanical setup would be able to give me a faster machine. thanks again for any advice.

It's more a question of how fast is the motor capable of supplying usable torque at? Mach can get you 35,000 to 45,000 steps/sec from a parallel port. If you need faster, you can use a SmoothStepper with Mach3 and get up to 4,000,000 steps/sec. www.warp9td.com

[hemsworthlad]

Why not look at doing something like the servo belt system as seen here. http://www.cnczone.com/forums/newrep...ote=1&p=636607
shouldn`t be hard to make version of your own to suit your machine. would eliminate your belt problems and you could have it 100ft long if you wanted with out any problems!!

[steve_p]

thanks for that ger- I hadn't heard of the smoothstepper, it sounds really interesting, and might well help me out if I keep the gearing I currently have. It led me to the Mach forum, which I have looked at before, but just registered on- loads of good info (and I see you're a moderator)

re your point about the torque of the motors at speed- would I be correct in assuming that torque shouldn't be too much of a problem considering that my reduction gearing is 34:1? I'd have thought this would allow plenty of torque, but restrict the speed somewhat.

whilst on the subject of torque, I was quite surprised to find that my current steppers are rated at 150 oz-in, whereas similar sized new NEMA34 motors (at about 100mm long) are rated at about 900 oz-in (and up to 1800 oz-in for 150mm long ones). Is this due to improvements in motor design over the years, or is there a different rating system used now?