Best belt drive ever! (If I do say so myself)

[isvflorin]

Mike and Mark,

I've been pondering about the drive system on my router and this type of belt system seems like it should do great instead of ballscrews, because of price and precision mounting tolerances for ballscrews.

The router I'm trying to build will have around 1200x1350x150 usable working area and ballscrews price for that envelope is quite high.

Mike - you mentioned this system is mainly desgined for speed but can't compare with the rigidity provided by ballscrews. Can you translate that in a more simpler way for me/us , meaning : good for wood routers but not for milling aluminium moulds lets say ? I'm trying to uderstand the capability of the system regarding light and heavy milling instead of just cutting wood. If it is not apropriate for ALu milling, than I guess I should go for ballscrews.

Mark, what kind of work are you planing to do on that machine ? You think it is ok to do Alu milling with the belt system ?

Another thing - regarding backlash : here is an ideea with the potential to unecessary complicate things but maybe reduce backlash. Instead of using one 25mm belt - why not use 2 parallel 16mm belts with the teeth offseted just a bit so when the 2 pulleys (2 pulleys - 2 belts) turn - you can be sure at least 1 belt is engaging imediately if the other has backlash due to belt shape mating or pulley backlash. When the motor will turn in the other direction - the belt that had slack previously will engage imediately while the other belt might have backlash. Sounds unecessary ?

Regards,
Florin

[Mike Everman]

Hi Florin,
I've been considering this as a prime mover for waterjet, plasma, wood routing, laser cutting; all light duty stuff beyond our normal pick and place type applications. I'd not thought this would be good for cutting metal, but your question sure compels me to make some stiffness measurements vs. belt pre-tension. The main advantage of servobelt being that the stiffness does not change along the length as a ballscrew would, and it's high damping as well.

Hmm, I need to lock the pinion somehow so I can make some force/deflection curves. Or maybe just hit it and determine the stiffness required for the resulting natural frequency. That's easier. I'll make this test soon!

[isvflorin]

Mike,
thanks for clearing things a bit, it would be great if you will be able to make some test to better compare how the belt system stands against ballscrews in Alu milling type application. Getting ballscrews where I live is not just difficult, the cost for shipping&taxes usually is higher than the ballscrew itself, and getting from ebay is also a gamble.

I'm assuming that if I use a system like Mark describes, where the lower belt is actually an epoxy resin cast - that system should be a bit more robust than the 2 belt config., and maybe a bit less backlash.

thanks for all your informative posts.

Florin

[mikie]

Mike,

Have you progressed on the kits or parts that could be purchased through your company?

Also,

I see that the drive is not reduced and has what looks like a 15 tooth pulley driving the belt (servobelt x-sect.jpg) . So the smaller the pulley the better the resolution produced by the system. I expect there is a trade-off between pulley size, surface contact to the belt and torque. Is there an optimum size for the pulley?

[Mike Everman]

Gday mikie, (this will be more than you asked for, since I haven't made an update in a while!)

Yes, that is a 15 tooth pinion, for 75mm linear travel per motor turn. With our 16-20K cpr encoders, it's around .0002" and better linear resolution. The diameter of pinion boils down to a life issue. As you go smaller on this and/or the idler diameters, the life goes down.
Life testing of a unit to destruction had a belt failure at 9 million cycles, or 18 million starts and stops. We've now changed to a higher flex belt with the same tensile strength and are up to 3 million cycles with 150 lb accel/decel loads on this test in progress.

Take all this from the standpoint of an industrial drive maker. I'd like it to have infinite life, which is of course impossible, but I'm going to get as close as I can. Some of my customers will use this to go from A to B all day long, so the life test is set up that way, to hit the same spot on the belt with accel/decel loads. If we were doing random moves to simulate a CNC application, then the original belt life test would still be running now and I wouldn't know what to say to the guy that's using it as an actuator. A DIY'er could use smaller diameter for sure for better force and resolution, and probably won't have an issue with belt life is my point.

I am very pleased with how it's going. I wish I were closer to kit level parts, and will shoot for something less than a shopping cart on our site, like email and paypal, perhaps by the end of the year. For this to be practical, I want to move some of the parts to castings to take the sting out of the current billet machined parts.

We're in the process of making a 90 foot long gantry with dual bridges, a 22' two rail dual carriage and one for NASA that will make a model of the CEV rock madly in a swimming pool to simulate high seas.

I'm also putting together a router at the moment so I can do circle interpolation tests at various cutting forces and speeds, with both stepper and servomotors (low and high end drive elements with roughly the same torque capacity).

[The Fiddler]

How does mhastings design not void your patent. I noticed his designs where in the public domain what appears before yours by a few months.

[Mike Everman]

How does mhastings design not void your patent. I noticed his designs where in the public domain what appears before yours by a few months.

Point me at the post and I'll take a look. Of course my patent application pre-dates my disclosure here by quite a bit...

[mhasting2004]

Now if you want to send me royalty checks thats fine
I'm flatered Fiddler but, to be honest I still havent built mine and Mike has a viable product.

To have a $1 for every idea someone has only to find out later that someone else has already done that... We'd all be rich.

Not sure, but would think something being in the public domain does not negate a patent. Nor even inventing something first (which BTW is not the case here IMHO) but who registers the patent first.



On a different note:

Mike I remember somewhere you posted that you have somewhere around 75mm per rev with the direct drive setup... How can you get enough torque with such a low gearing out of a servo and wouldn't this be way to "fast" for a stepper ie very low resolution per step. I tend not to count anything less than half steps for resolution (micro steps are great for smoothing but doubt they add true resolution)

Since I still want to use a GT2 8mm pitch belt I now think I really need to get a 10:1 low backlash planetary gearbox inline with my steppers to get a resonable resolution. The original 5:1 belt reduction doesn't "feel" right on paper.

With 10:1 it will get me down to 17.5mm per rev which is, i think, manageable for my application.


I'll pm my bank details....

[Mike Everman]

Czech's in the male. (that was some joke about a Czech and a Canadian, campers running into a pair of grizzlies.. ha, though the details elude me)

I still want to see your post of this idea. Is it truly identical?

Anyway, yeah, 75mm per motor turn is interesting, though I never use steppers, and always have 16-20K counts per rev. I can imagine it wouldn't be too snappy or precise with steppers. We offer 3 stack lengths in both nema23 and 34, with peak forces up to 200 lb., though most customers want to use their own motors.

I still hate backlash, so I'd rather do the belt pre-pass. If doing it DIY, instead of custom hobbed pulleys for zero clearance, one can simply turn down the od of the pulleys so the belt teeth fall to the bottom of the grooves.

[Mike Everman]

What a lot of people do not notice is the main element of the ServoBelt, the static lower belt(or rack), which makes it so much better than a simple single belt/moving motor design. There are variants that are like tractor treads, but they do not address the fact that there are tooth clearances, though they usually have a metal rack with special tooth profile.

I just went through your router build thread, and didn't see the static belt there, though nice work, man!

[mhasting2004]

Yeah I don't remember posting a meshed belt thing in that thread either although I had been thinking about it prior to seeing your thread... honest. Thats why I was so happy to see you had been sucessful in your design as it gave me confidence it was a sound design.

Like anybody, I am not crazy about introducing backlash into a system. However the way I look at it as long as its a fraction of my ideal resolution, I can live with it. To get 10:1 using belt reduction meant a couple stages and that adds too much complexity and added sources of backlash.
<= 6 arc second gearbox would mean only 0.049mm which for my machine (plywood sheet shreader and hot wire foam cutter) is more than acceptable... I think.

Cost is a big driver in my design and if I could find 2.6m long 10mm pitch ball screws and rotating nuts for less that what a gearbox belt system costs I would probably go that way.

As far as Canadian bears go ... you don't have to out run them... just out run your camping buddy.

[Kipper]

Hows the patent application going? anything turn up in the search? Prior art? Patents are such fun

[mhasting2004]

OMG
With only 2 posts on the forum you've started a patent storm Fiddler!

Really need to keep this thread on topic.

To wit...

Mike do you have any videos of your big setups that you have described? Be great to see a long axis plasma etc screaming along.

[mhasting2004]

Another question for you Mike

I'm at a tipping point with the idea to use GT2 8mm pitch belts. As much as I would like to use the 8mm pitch because the t5 belt just looks sooooo flimsy to drive a big gantry... I find the greater pitch and therefore larger minimum sprocket... problematic.

So here's the question have you been able to test your drive at cutting speeds to get a realistic amount of force that the t5 setup can handle? Clamp a 25kg weight to your drive and see how she goes as my gantry is 50kg all up so the start stops may be brutal on the poor old belt.

I've been going around in circles the last few days trying to justify the GT2 8mm belts....
direct drive.... way to "fast" for my setup and steppers.
Belt reduction.... need two stage as 5:1 is still too "fast"
Low backlash gearbox.... 10:1 would be good ... but ouch on the $$$ for low backlash.

So I am, at this stage, turning back to the t5 with a 5:1 belt reduction as shown in my build log. But have concerns about ripping the shallow teeth off. Am I worrying for nothing?

Cheers

Mark

[blue_luke]

Hi all,
I am chuging along building my machine and elected on a sorta' Mike Evermann servobelt system.

I just got the pinions and the belting yesterday (20mm T5).

I have re-read most of this thread and then it struck me!

What do you use to glue the static belt on raw aluminium???

Simple contact cement is my first reflex, but I wonder what you guys would use?
Thanks, Luc

[timber_MG]

3M double sided tape might be worth a shot, but spray-can contact adhesive should work a treat.

[Mike Everman]

Yes, we use 3M UHB tape, though as said, contact adhesive would work fine, just make sure it's an even coat.

[calgrdnr]

Howdy Mike ,

After the show my buddy and I decided that your servobelt system was
in the top 3 things that we liked about the show, and right this minute I don't remember the other 2

I went to your site don't see pricing, Is there a pricelist for us DIY'ers to wish upon ?

Guy's the system was most impresive in person .


Kent ( calgrdnr at sbcglobal.net)

[Mike Everman]

Thanks so much, Kent! That sure makes me want to hurry up with the availability to DIY'ers. For a bit longer, this is just available to OEM's. I'll get on it!

[oxbadfood]

Mike,

First off, great design. I've been thinking of building a hobby wood CNC using a rack-and-pinon set-up with stepper motors, but I felt it would be quieter if I went to a belt drive with preload. Since steppers have good low speed torque, your direct drive belt drive would work well for my application.

I have a question though. You look to be using servo motors. How can you direct drive a load without gearing the motors down? Your only gearing comes from the pinion's drive radius in relation to the motor's continuous torque spec.

My problem is if I use a 100 watt servo system, I'm limited to like 60 oz-in of torque, which with a 1/2" pinion would only give like 8 lbs of cutting force? Now, I'd only be operating the servo at 15% of its capacity (60-120 RPM?), so I'd need to gear down to get more torque... but if I add a gear head or another belt and jack-shaft type gearing wouldn't that introduce backlash and/or add springiness to the system?

I don't want to use a huge, heavy, expensive, over-powered, over-priced servo to develop any real usable torque if I'm only running it a pokey 120 RPM verse its continuous power peak of like 3000 RPM.

Just wondering if I've got this correct? No doubt this thing can move fast, but how you are able to develop any usable linear force from your set-up?