Anyone having problems viewing brain’s last video and using Internet Explorer?

DJ

Yep but the motionbox link fixed it for me. (I use Firefox BTW.)

Dan, this post won't load the player?

http://www.lumenlab.com/forums/index.php?s...st&p=287750

Thing is I usually use Firefox. When DAZZ mentioned he was having trouble I fired up my IE browser and the video didn't load up. I went to the motionbox link and that fixed things. Could be since I rarely use IE that it was kinda out of date or sumpin'???

I'm pretty sure you just need the latest Flash:

http://www.adobe.com/shockwave/download/do...=shockwaveFlash

DAZZER: What problem were you having???

Well you asked.

http://www.cnczone.com/forums/showthread.p...3232#post463232

Steppers are normally a lot slower than servos but they do perform well if you want low speed torque. The belt idea in the video is ingenuous. For longer runs with a machine mainly used for wood cutting it's a cool efficient system. The belts can be had for a fairly low cost too making it a good system for DIY.

Cool vid; fast as hell! That long rack of grooves looks expensive though.

I'm thinking of using belts in my eventual machine. The cost for good belts is a little higher than cheap leadscrews but "negligible" backlash and ballscrew-like positional accuracy make them a great choice... on paper at least.

If anyone would like price some belts and pulleys this is a good place to start: http://www.sdp-si.com/web/html/products.htm
Looks like the GT2 - 3 mm or 5 mm pitch belts and pulleys are the best they offer.

I like the lead screws fine for short runs but as the run get longer you start to need a fairly large and heavy lead screw. Short belts are a lot more efficient but as they get long the stretch will be greater than any properly sized lead screw. For a wood cutting 4 x 8 machine a rack system is a good solution and the belt system I linked to would be way cheaper than a steel rack.

The belt rack system is just using a belt glued to the rail to act as a rack. The bottom belt does not move it is just stuck to the rail. The top belt is just a strip of belt that gets pulled on by the stepper.

For smaller machines the lead screw system is just fine and fairly cheep at around $55.00 for a 6 foot 1/2" 5 TPI ACME precision lead screw. The antibacklash nuts are a greater cost at $18 each.

It looks like SDP charges a lot for everything, I would keep shopping.

Is it a fact that belts stretch? From everything I've seen people say they don't. Here's a quote from a log on CNCZone...

New Machine Build - CNC Design from Scratch, w/Belt Drive
Vid of the first belt

The SDP stuff isn't bargain-basement cheap, but I haven't been able to find any other reseller of 3 or 5 mm GT2 belts. Grainger sells 8 mm pitch GT2, and I can get mystery belts on eBay, but neither of those interest me. If you have any suggestions I'm all ears.

Is it a fact that belts stretch? From everything I've seen people say they don't. Here's a quote from a log on CNCZone...

New Machine Build - CNC Design from Scratch, w/Belt Drive
Vid of the first belt

The SDP stuff isn't bargain-basement cheap, but I haven't been able to find any other reseller of 3 or 5 mm GT2 belts. Grainger sells 8 mm pitch GT2, and I can get mystery belts on eBay, but neither of those interest me. If you have any suggestions I'm all ears.

A nice quote from a source on the Wikipedia timing belt page...

I think I can handle replacing a $15 belt every million runs so that I can have a system with zero backlash. Heck, that sounds like something that I could replace every year... or if I'm feeling especially paranoid maybe every 150 hours of use (10¢ per hour).

The original big RoGR ran on timing belts. To solve the price problem, I spec'd automotive timing belts and gear-pulleys. The crank pulley makes a good driver. The guys at the local auto supply let me wander around the back.

Screws won the day however...easier, cheaper, choices etc. Belts definitely stretch, which is why 'tensioners' are used on cars. Hell, even timing chains stretch.

Belts must be enclosed too...any chips between the belt and cog will ruin your job.

Everything loads up except for the last two videos brain posted. They just appear as black boxes. After about a minute or so internet explorer shuts down and the report error message box pops up.

DJ

I've been looking for a while now and I cannot find anyone claiming that their belt driven CNC machine has lost accuracy because of stretching... I haven't even seen anyone report that they suspect that they're loosing accuracy. If y'all find some I'd love to read it!

Timing chains are (apparently) notorious for stretching... I'm not a "car guy" but it seems that one advantage timing belts have over chains is the fact (or at least belief) that they do not stretch.

Tension is required to prevent the belt from jumping teeth, but it doesn't take much. It looks like a 6 mm GT2 3 mm pitch belt needs about 0.5 lbs of tension minimum. At 5 lbs of tension it should be able to take up to ~35 inch pounds of torque (~560 oz-in) at 750 RPM before it jumps a tooth (on a 30 tooth pulley). Seems logical that a 9 mm belt could take 50% more torque (lets say 800 oz in). That seems like plenty for a 300 oz in stepper. For those servos you're selling I think I'd want at least two 9 mm GT2 belts or a single 15 mm wide 5 mm pitch GT2.
A 15 mm wide GT2 5 mm pitch belt can handle ~350 lb-in / ~5600 oz-in (10 times more) of torque with ~20 lbs of tension, this is with a 20 groove pulley moving at 2300 RPM.

Sure...let me say I don't believe stretching matters, it's just that I know it's possible from my days as a mechanic. I actually like belts for robots. If chips in the cogs don't apply, or you can enclose the belt then sure.

Oh yeah... chips... I guess chips are always bad for belts (cheesy pun I know).

Keeping chips out of the teeth would be another advantage of that rack/belt system AV posted. I think I'll try to emulate that system if I can...

Keep in mind you’ll probably need to have some sort of reduction if you were to use belts.

DJ

I've been playing with the numbers and it looks like a chain of pulleys with 32 to 60 to 25 grooves and a 3 mm pitch belt should give me 0.2 mm (~0.00787") per step accuracy (127 steps per inch) which is roughly equivalent to a screw with 5/8 of a turn per inch (125 steps per inch) 1.25" pitch screw (288º rotation per inch). This seems high, but I'm thinking it's a realistic goal to use with a router that'll probably have a fair amount of runout. Seems a little silly to shoot for half-a-thousandth per step when the router might have 20 thousandths of runout. I'm thinking I'll run the machine at half-step most of the time (0.1 mm per step / 0.00394 in) and quarter (0.05 mm/0.00197 in) or eighth step (0.025 mm / 0.00098 in) if I'm making something that requires extreme accuracy. At half a step, each step will cut a little more than the thickness of a piece of paper... sounds pretty damn good to me! If I feel like this isn't enough it wont be too hard to add a couple 2:1 or 3:1 reductions.

Hirudin, Did you do the cad drawing?

Yeah The motor is a NEMA 34 and the pulleys are the right diameter (the pitch diameter) and width for the belts, but I didn't feel like modeling the flanges/hubs.

I can e-mail the 3dm (Rhino's format) or DXF if anyone wants em... I guess just PM me with your e-mail address.
Actually, this iz better...
3dm
dxf

By the way, I just used whatever the default settings are to export the DXF. The model units are inches. Let me know if it doesn't work or something.

Cover the belt with another belt and the chip issue is largely mitigated. Simply enclose/shield the pinion:

http://www.bell-everman.com/ServoBelt.wmv

Sure is a lot of parts. Aside from mechanical losses, the complexity will add significant variance to your tolerances, especially as it wears. The cost will be higher and the maintenance more time consuming. These issues along with real-time usage issues lead me to stop development in favor of the screw.

'course none of that matters if you're doing what you want! What advantage do you see?




Experience teaches; the rest make speeches.

Yeah, it is definitely a couple more parts... Motor, driven pulley, endless belt, reducing pulley, actuator pulley, long belt. Apposed to screws: motor, coupler, screw, nut.

According to the literature (not experience) wear should be low. There is supposedly very little friction between the belts and the pulleys. The only wear area I can see is possibly as the teeth get seated in the grooves. Also, since tension will be low, there shouldn't be much wear on the bearings.


The advantages I'm looking forward to are...

• Should be zero backlash
• Positional accuracy should be very high (~0.0013 per foot) pretty much up there with the best rolled ballscrews that I've seen ($400 ground ballscrews are of course better) and about 3 times better than the leadscrews that I've seen
• They're fast, like I said, the system I'm looking at is about the same as a 5/8 TPI screw.
• Very low inertia
• Very low friction - I think they say they're like 97% efficient
• Length is a very insignificant factor in the cost
• Initial cost is low (I'm going to go with polycarbonate pulleys which will be about $20 per motor for the set, add about $9 for the endless belt, and about $12 per meter of long belt. 2 complete systems per X and Y axis make for ~$90 axises, close to leadscrews.)
• Replacement parts will be cheap (a replacement belt should be about the same as a replacement dumpster CNC nut)
• Probably more

It's not all roses, here's some disadvantages that I'm aware of or worried about

• Distance per step is much greater than screws (depending on pitch of course) I'm shooting for 0.1 mm per half-step, with a 10 TPI screw it would be easy to get 0.00635 mm per half-step which would be almost 16 times more precise. I think that level of precision is an... unrealistic expectation in a home-made, router spindled machine.
• Chips in the teeth are bad (I can't imagine chips are good for screws though)
• Concerns about stretching
• Initial setup might be difficult
• Probably more

Excellent list-format posting: You are "committed".

Say please;You have already built this drive and "this is the result":

I've worked on several pieces of equipment using the belt drive / pulley mechanisms and there are a few other drawbacks that I envision:

• Lack of torque : Yes, they are high speed, but they often have trouble holding their position with any sort of load working against them. Some of this can be mitigated with a proper motor controller, but not all of it. Conversely, you're never going to turn a screw by pushing back on that axis.
• Tensioner mechanism: Without a spring loaded tensioner you will be hard pressed to have consistent drive results using a belt. This adds on a set of slider mounts for the motor (or idler pulley) and a tensioner. Again, any sort of resistance on the cutting head needs to be less than the spring tension so you don't jump teeth.
• Less resolution per axis: Related to the lack of torque, you will get more speed but less step resolution with direct drive.

This system is probably more optimized for low torque systems like 3D printing or laser cutting where the resistance at the working head is minimal.

Hopefully that all makes sense.

Precisely. Seen on printers and plotters; high speed, lower precision devices. When ball and high-lead screws were >$1000 each, belts were attractive for high speed, low torque machines. You'll be hard pressed to find a commercial machine with belts now.

Sounds to me like the pieces of equipment you've used were servo driven. I'm willing to concede that belts and servos are not a good matchup.

Lack of torque...
If I'm not mistaken, torque is supplied by the motor and is increased by gear reduction.

hold their position against a load...
In the situation where the machine is having trouble holding it's position, what exactly is happening?
Is the belt stretching? (If this is the case, why is the belief that belts do not stretch so prevalent?)
Is the tensioner not doing it's job? (Sounds like a poor tensioner design to me.)
Are the motors being forced to turn in the wrong direction? (Maybe a problem with servos, but from everything I've read I don't see how this could be a problem with anything but severely under-spec steppers.)
Some leadscrews will turn with only lateral force. I'm of the opinion that the motor should prevent the machine from moving, not the linear actuator.

Spring loaded tensioner and jumping teeth...
It doesn't appear to me that a spring loaded tensioner is required to use timing belts. Sure tension needs to be present, but the flexibility of a spring doesn't seem necessary.
Click to view attachment

Less resolution...
This is definitely true. The only argument I have is that moving a wood workers router (that is likely to have a fair amount of runout) around a table 5-ten-thousandths of an inch at a time is like having a laser guided shovel. Sure you can tell the shovel exactly where you want it to go, but once it slips off the first rock your accuracy is thrown out the window. Or maybe a more apt analogy would be mounting a scope on a BB gun.


Aren't printers and plotters extremely precise? Printing a straight line across a page at 1440 dots per inch without any discernible break or variation seems pretty precise to me.

Yes, EVERYTHING stretches. In mechanical engineering terms, this is known as Young's modulus. Nothing is perfectly rigid. Nothing is unbreakable. Nothing is heat-proof. All matter ultimately relies upon chemical bonds which have finite energy.

-Brian

I wonder if it's a fact that belts stretch under normal conditions. The idea that the windshield of my car is actually stretching because EVERYTHING stretches is laughable.

Laugh all you like. Young's modulus for glass varies depending upon the exact recipe, but I've seen it range from 50-100 N/m^2 (or GPa). By way of comparison, pure Aluminum is 69 (the alloys are all over the place, and there are thousands of different aluminum alloys). Here's a reference: http://en.wikipedia.org/wiki/Young's_modulus

-Brian

Relative to a mill? Hardly! .0014 isn't bad though, if you could repeat that accuracy across the board. Rest assured you won't be "printing" any engine parts with your Epson Photomate, or whatever.

I'm not being contrary to your idea; though you are brazen to cancel your order and begin your new idea in the RoBLOKS thread! Maybe we should move this thread to the DIY section...

You're right, this discussion would be better in the DIY section. I didn't intend this to turn this thread into a belt discussion, but I've certainly contributed to steering the conversation in that direction. If post #7/8 and on suddenly disappear I'll look for it in the DIY section.

As I said, I canceled the order for the bearing kit out of necessity. There was a fair chance I would have gone with belts even if I followed the rest of the RoGR plans/parts.

1" ÷ 1440 = 0.00069
1440 is the lower number of the X and Y resolution, on Epson printers the higher is something like 5700... (~0.00018)

Well, it's not your windshield but you can watch a wineglass stretch if you jump ahead to about 7:04 into this video...

http://www.youtube.com/watch?v=nQTgejm7vf4

Nice vid. I often talk/think about interference, which is very nicely shown in the wineglass example. There is much to be done with this nascent understanding; yet Tesla understood it well enough 100 years ago. Perhaps some of you know about the Tacoma-Narrows Bridge collapse? The same behavior as the wineglass, yet a giant bridge.

Chalk this up to semantics if you want, but that wine glass isn't stretching, it's flexing. If there was a video of a hula-hoop doing the exact same thing I don't think anyone would call it "stretching".

Steel flexes too, as evidenced in brainchild's video above, and this other one I managed to find...
solid steel flexing

Sigh. No, it is not a matter of semantics. Contrary to what you believe, these are rigorously defined mechanical engineering comcepts. "Flexing" (generally a layman's term) is repeatedly putting an object under tension. Stretch is also layman's term for tension http://en.wikipedia.org/wiki/Tension_%28physics%29

Rather than continue to put your foot in your mouth (at this point, you probably have a pretty bad case of 'Athlete's Tongue'), you might want to educate yourself. Try reading any first-year mechanical engineering text, or for a really good qualitative overview (with minimal mathematics), try J.E. Gordon's "The New Science of Strong Materials"

-Brian

ratchet down the argumentative tones all

During the video the host used the words "bending" and "flexing" (I'm on my phone now, so Youtube is a pain so I don't feel like double checking)... I didn't hear him use "stretch".