[Oracle_9]

Hello,

Concerning: what is belt drives?

My project background:

I have been reading a lot here in cnczone. There is a lot of info here, too much to handle. I am doing my research before I start buying stuff. I will be making a CNC plasma cutter with an detattachable head to hold OXY-A torch and a small router. The cutting area will be 4ft x 4ft. The framework that hold the hold thing together will be using steel sq. tubes. I would like to get an accuracy of 0.002 - 0.005...but no more than 0.005. the base will be stationary and only the gantry will be moving on rails. I will have I guess 2 steppers on the sides of the gantry for the x-axis, 1 stepper for the y. Each will be 640oz steppers (or 1200) I will use Campbelldesigns breakerboards, steppers, and Mach2 or 3 with Sheetcam.

I like the idea of rolled ballscrew with its accuracy. I have not gotten prices or which to select for it yet, but I do know its cheaper than precision ground. I also have to take into about of whipping, warping, thermal expansion under high speed, end mounts.....etc. I hope this aint too expensive. There is the ACME screw and nut way, but I dont like the lack of efficiency, requires more energy to move it, especially if you had the spring loaded antibacklash nuts to it, and the the wear factor, etc.

There is the rack and pinion. Its where a long straight linear gear bar is stationary and the gantry has a stepper on it with a small gear head which moves it along the axis. People mentioned you should have a reducing device for this application, using pulley-type toothed time belts. But an issue of accrucary comes up. People saying you can achieve 0.004-0.007". (0.004 with teeth adjusting software or something), and backlash...but it is cheaper than ballscrew method. I will leave this and the ballscrew method open, until I get prices to get a better idea which to choose.

Recently someone here mentioned Belt drives. I got a description of it as well from another thread, but I do not quite understand it yet. They said its an alternative to rack and pinon. I will also consider this for my project, but I would like to understand it first.

Evodyne nicely stated:
"Poke about the sight and look at some of the machines using belt drives. Essentially a long toothed belt is fixed to the table at both ends. Your servo (usually with reduction) or stepper is mounted to the moving axis and turns a toothed gear. Guide pulleys ensure that the belt wraps around this drive pulley. Once the belt is pulled taut there is no freeplay. It's basically another play on the rack and pinion. Looks like it could be very cost effective too."

This may be a simple thing, and I will probably be kicking myself once I get it!!, but right now, I just cannot visualize it.

Are there any visual examples of this, perhaps a photo? I tend to understand things more once I see it.

Any help will be appreciated. Thanks.

[ger21]

There are two types of belt drive in this thread http://www.cnczone.com/forums/showth...ght=belt+drive

What Evodyne was saying is that it's set up just like a rack and pinion. The belt is the rack, clamped tight at each end. The pulley is the pinion, with two idler pulleys which wrap the belt halfway or more around the pulley.

[Oracle_9]

Oh, cool, I didnt look in the ROUTER section. That link you listed is awesome. I better understand, and yes, I am kicking myself now...ow. Its like a step by step photo instruction manual with diagrams. I will have to take a look at Ynnebs cnc as well later on today.

Thanks everyone.

[ger21]

There was a website with a really well done belt drive system, but unfortunately it's gone now. But, I think Benny (YNNEB) may have some of the pics from it.

[ynneb]

Your memory of links, and what people have on their hard drive amazes me Gerry
I should upload these pics on the photo gallery since they are a fine example of a belt driven machine.

EDIT: These are not my pictures, nor have I sought permission to distribute them. I am just hoping that Dean doesnt mind.

[ger21]

Benny is the man. I've wished so many times that I would have saved those.

[ger21]

Hey, you even included a map to your house.

[ViperTX]

Since what you're designing is a plasma table and you're looking at using 640 oz-in motors, I would not concern myself with ACME screw's efficiency, plasma heads are pretty light weight and don't have any pushback. Personally I would use the ACME screw and throw the belt drive, rack & pinion, etc. out the door.

The cheapest ACME screw is going to have a lead variation of 0.004 inches per foot...

For guides I would use V-wheels.

[ger21]

The cheapest ACME screw is going to have a lead variation of 0.004 inches per foot...

The Kestone at MSC is .009. The Nook is .003, but costs twice as much.

Go with multiple start acme if you're worried about efficiency. It's much more efficient. 1/2-10 5 start should get you pretty good speed with 640oz motors I would think.

[ynneb]

Yeh, I cant imaging how that map sneaked in.

[Oracle_9]

Thanks for the replies. Also, thank you for the photos ynneb. I really like your machine. Hmm, I wonder if a plasma cnc can fit into Santa's slide? hehe.

Reason for 640oz steppers, is that Cambelldesigns has them right there with the other stuff I am gonna buy...also for $120 for 640oz...it seems like a good price too.

[ynneb]

Sorry if I didnt make it clear, these are not picture of my machine, but I saved them from a web site that is no longer hosted. I did not seek permission to save them or upload them here. I am hoping that Dean would not mind, since we are holding them up as a good example of a belt driven V bearing machine.

I built my machine around this design, although I took a few cheaper options (Comprimises) in doing so.

[Oracle_9]

So, if I use the belt drive system with both sides belt drive and shafts linking both belts at the ends of the table, with 1 servo belted to one of these shafts, would 640oz (reduced to 3:1 or 5:1) be enough for move the gantry? The cutting area is 4x4 mostly for plasma but I would like to do some routing a few times. I guess my max cutting speed with plasma is 500ipm, so the MAX imp for the cnc would be 800-1000ipm.

Or one 1200oz, but geared down more? <--Ima making those grunting sounds that Tim Allen does when stimulated by more power on the tv series "ToolTime"

[ynneb]

I did 3:1 because of budget, while it works and the resolution is acceptable, I wish I had gone 9:1 . Currently my servos only reach 20% of maximum speed., otherwsie the gantry moves way too fast. Try and get as much reduction as you can afford. Greater touque means better control and faster acceleration / decceleration. These things matter with CNC.

EDIT: Dont be sucked in by those who brag fast rapid movements. This is not the measure of a good machine.

[Evodyne]

Hi guys! Benny & Gerry, thanks for input. I meantioned the belt drive to give our friend another option, but didn't have time that morning to track down the links before work. You guys are great.

I got an e-mail at home from a fellow asking about motor sizing, choosing a reduction amount, etc. My reply used my version of common sense (uncommon sense? uncommonly senseless?) to try and correlate the reduction to the motor size, screw pitch and efficiency, force needed, etc. It would apply to a belt setup or rack and pinion just as well. I don't want to retype my reply, so I'm attaching it as an .rtf file (zipped)... if anyone wants to comment on/critique/ridicule/confirm/expand upon my thoughts I'd like to hear their thoughts.

Oracle_9, if the attached is deemed reasonable, it might lead you to a good estimate on the needed reduction. Please note, however, it is based on servos, not steppers. Good luck!

Evodyne

[ger21]

I guess my max cutting speed with plasma is 500ipm, so the MAX imp for the cnc would be 800-1000ipm.

How'd you come up with 800-1000ipm?

You said servo, but you're talking about steppers, right?
I posted this in another thread:

To get .001 (full steps, not counting microstepping), a 16 tooth GT2 (5mmpitch) is about 1" per rev. That's .005, so You'd need 5:1 to get .001 resolution. 100ipm at 5:1 would be about 500rpm.


You want 500ipm, up to 1000ipm. That's 5x500, or 2500rpm. Up to 5000rpm.

At 1000rpm, your 640oz stepper will probably be down to about 100-150 oz-in of torque, and it won't spin much faster. The bigger the motor, the slower they spin.

So, to get the kind of speed your talking about, you're going to have to sacrifice some resolution, and / or go to servos.

Someone jump in if my math is wrong.

[Oracle_9]

Hi,

Wow, thats a very nice explanation you have provided Evodyne. I will need some time to digest all that formulas and calucations :O

Even though you specified for Servos, but your earlier calculations before you got into the servo encoder stuff, I think those could be applied to steppers as well, yes?

Ger21,
I read somewhere on cnczone of someone giving a "ballpark" estimate of the how fast should a machine be capable of the running, from the actual speed. I think someone posted they need like 300 or 500 ipm to plasma cut something (the rest would require lower ipm to cut) and so it was suggested to making the maximum speed capable for the machine to cut be 800-1000ipm (Btw, this is not the stepper ipm, but the overal cutting speed movement) There was no formula or anything like that. But I could remember this wrong too.

Anyway, what Evo has provided (assuming I can use some parts of the formulas for steppers), is really helpful, and to get a better idea of things.

Ger21, after seeing all these formulas, I no longer know what is right. I need to really digest all this info. OI!!!

"The bigger the motor, the slower they spin"
Thats is an interesting info. I didn't know that. Which brings up another interesting question, what would the max. RPM a stepper can do!

BTW, I am sticking to steppers, servo costs + encoders + other stuff would be too expensive, maybe in the future in another cnc machine...once I get the hang of things on this current project.

[Evodyne]

Hi,
Wow, thats a very nice explanation you have provided Evodyne. I will need some time to digest all that formulas and calucations :O

Even though you specified for Servos, but your earlier calculations before you got into the servo encoder stuff, I think those could be applied to steppers as well, yes?

Hi Oracle_9. Well, if you can get from a force or a speed of the axis back to a motor rpm, you can figure the drive pulses...say a big stepper directly drives your 5 TPI screw. Takes five motor turns to go one inch, right? So multiply both sides by 200: you need 1000 turns to go 200 inches. Do it one minute and you've got 1000 RPM for 200 IPM. If the motor is a 200 step/revolution unit, then you need 200 (steps) x 1000 (RPM) or 200,000 steps to get there. Use x10 microstepping and you need 2,000,000 steps/minute. Divide by 60 to get 33,333 steps/second or 33,333 Hz. That would be the pulse frequency needed (assuming a pulse and direction drive like a Geckodrive).

The rub is that Windows software (due to multitasking,etc) limits the practical frequency that can be generated. Mach 2 uses it's own driver, but still has a limited "top end". I don't know the exact value, but it keeps you from an 800 IPM machine. I don't think it's too terribly much more than 30,000 hz (again, just from what I've read here). Please verify these values-I'm just quoting from memory.

The new G-Rex (Geckodrive) is a pulse generator that sits between the PC and the drives. It's limit is (I think) in the 2,000,000 pulse/sec range. Mach 4, or a version of it, will support the G-Rex. But Mach 4 isn't out yet and the G-Rex isn't free.

The other rub is that the steppers lose torque as RPM climbs. Low torque at speed can lead to missed steps and, well, problems. So you just ca't run pedal-to-the-metal!

Bipolar setups tend to outperfrom unipolar, and, with all else equal, chopper drives will outperform current limiting drives.

So...lots of factors affect the final result (See the attached pdf formerly found on the Geckodrive site).

Go to the many sites selling "professional" entry level plasma cutters. Jot down the details: motor size, quoted cutting speeds, etc. From the pictures you can see the drive setup. Most will be rack and pinion. After a few sites you'll see a pattern in components and specs. These guys have done the math, so you would be in safe waters if you based a machine on their designs.

Evo

[ger21]

The rub is that Windows software (due to multitasking,etc) limits the practical frequency that can be generated. Mach 2 uses it's own driver, but still has a limited "top end". I don't know the exact value, but it keeps you from an 800 IPM machine. I don't think it's too terribly much more than 30,000 hz (again, just from what I've read here). Please verify these values-I'm just quoting from memory.

45,000 with Mach3. Indexer LPT can do 100K+, but it's about $800

The new G-Rex (Geckodrive) is a pulse generator that sits between the PC and the drives. It's limit is (I think) in the 2,000,000 pulse/sec range. Mach 4, or a version of it, will support the G-Rex. But Mach 4 isn't out yet and the G-Rex isn't free.

4,000,000 per axis x 6 axis. An alpha version of Mach4 is available that will run g-code, with working limit and home switches.