[HomerSimpson]

Interesting find. I wasn't aware of anyone other than Mike who had developed this drive scheme. He does have a patent pending on this though so it would prohibit anyone from building and then selling the concept, unless they had found a way around his claims.

$95 seems like a good buy but in my experience this hardware won't work well. You will need the extra idlers to maintain tooth engagement. You will also need to gear reduce to get to a good spot for torque and accuracy.

I've been working on this concept on my machine off-and-on for a few months. Unfortunately I'm using belts I already had which have a more rounded tooth profile (HTD) and I think these make it very difficult for this concept to work. I also am struggling with belt tension, between the primary idlers and the motor pinion. No matter what I do, the belts become slack after a few revs--again, I think it has to do with the rounded tooth profile but I am close to giving up on the concept and going with rack and pinion.

I hope someone is able to make this work or that Mike eventually makes something available to DIYers.

[PaulRowntree]

@HomerSimpson : Is there a chance you need to support the other end of the pinion?

How much slop did you have with the original 5' long belt? How was it bound to the support? What belt material did you use?

Cheers!

[HomerSimpson]

Paul,
It's possible. I haven't put a dial indicator on it under tension to measure the deflection but visually it isn't moving. I've played some with the distance (or interference) between the idlers and the top belt and that seems to have the most impact, although, doesn't totally eliminate the belt going slack up to the motor pinion.

The belts are made of polyurethane with a fiberglass reinforced inner layer. The fact this isn't steel may cause problems over life with stretch but not my time zero problems.

[ger21]

It's possible. I haven't put a dial indicator on it under tension to measure the deflection but visually it isn't moving. I've played some with the distance (or interference) between the idlers and the top belt and that seems to have the most impact, although, doesn't totally eliminate the belt going slack up to the motor pinion.

Seems to me this can be only two things. Either the motor (pinion) is moving, or the bottom fixed belt is moving. The only other way it could get loose would be for the belt to stretch (a lot), and belt's don't really stretch that much.

The first is from Joe's 4x4 Hybrid forum, where he was testing a belt drive system several years ago, probably about the same time this post was started. He has a video showing this system working on a 10' rail and designed to fit the current mounts for his 4x4 hybrid, running rapids at 1000ipm. Further testing had rapids over 1700ipm

Those are really meaningless numbers, imo. You really need a minimum of 3:1 reduction. And I'm guessing that direct driving off a 1/4" stepper shaft wil realy shorten the life of that shaft.
With direct drive, the resolution is poor, and the accel is way too slow. I haven't seen a rack and pinion with similar accel rates have what I would consider adequate acceleration. I believe that most use accel's in the 30-70 range. 1G = 388. So even at .5G, I'm hoping to accelerate 5x faster than most users of Joe's machines with rack and pinion. But I'm trying to design for 1G.

Rumor has it that the next version of Mach3 may allow different trajectory planners. If this works, than S Curve accel may be available, which should make 1G much more attainable, and a little less violent.

[HomerSimpson]

Well I finally got my "servobelt" drive working last night! I use steppers and originally tried with no gearing and you guys were right, that didn't work. So, I've added a few more parts and have a 3.4:1 gear ratio and that works perfectly. I was having problems with my top belt losing tension between the idlers and the pulley so I had to clamp down the ends of the top belt so they don't pull in and lose tension. I know Mike at one point said not to do this but he is using very wide, trapezoidal tooth belts. I'm using 9mm wide 3mmHTD belts that have a curved tooth profile and I think that is why mine tend to "creep" unless clamped.

it's not very clear in the attached picture but the purple piece on the same center of the large gear is the pulley with the belt that goes down to the idlers, shown in the second picture. I didn't update this bit in solidworks. Anyway, I'll post a youtube link of it in motion soon.

[diyengineer]

For people using belts and wanted to consistently use the same belt tension. You could buy a #20, #30, #40, etc pound spring, and in the end of your belt make a small clamping device (a hole may work but possibly may just rip out). Then put the belt under the tension of your spring at a certain distance and mark it, and clamp it down. Over time as the belt stretches out, you can loosen up your clamp, and check your initial spring tension and distance, it will have obviously changed and re calibrate it. Probably a really consistent way of setting up multiple belts with the same tension for cheap, also its a bit more scientific than "I eyeballed it". haha.

[ger21]

Are you really using helical gears? If so, the thrust load probably isn't good for the stepper.

Let's see that video!!

Here's my redesigned setup, with 1/2" shafts and 32mm wide belt. This should be a lot stronger.

[tjskcnc]

Those are really meaningless numbers, imo. You really need a minimum of 3:1 reduction. And I'm guessing that direct driving off a 1/4" stepper shaft wil realy shorten the life of that shaft.
With direct drive, the resolution is poor, and the accel is way too slow. I haven't seen a rack and pinion with similar accel rates have what I would consider adequate acceleration. I believe that most use accel's in the 30-70 range. 1G = 388. So even at .5G, I'm hoping to accelerate 5x faster than most users of Joe's machines with rack and pinion. But I'm trying to design for 1G.

I just showed a bit of what they are working on that site. Joe abandoned this format a long time ago, but other users have further developed this idea using gear reduction and other modifications, complete with CAD drawings for others to use and improve on.

[HomerSimpson]

Gerry,

Your setup is looking good! I thought you were going to go with a bearing support on both ends of the pinion shaft though. I am indeed using helical gears. I pulled these motors, with helical gears attached, from a heavy duty xerox printer that I've worked on. These are life tested units that were being scrapped but function perfectly and the load they were under in the machine were MANY times what they will be under in my machine.

Unfortunately I'm going out of town for a week so won't be able to get that video up for another week or so--but I will keep you posted. When will your machine be ready?

[ger21]

I thought you were going to go with a bearing support on both ends of the pinion shaft though.

I increased the shaft diameter from 8mm to 1/2" instead. Should be about 3x stronger.

When will your machine be ready?

Probably about 2 years, unfortunately.

[HomerSimpson]

I've finally taken some video of my CNC that uses the belt on belt drive design for the X & Y axis.

Take a look: [ame=http://www.youtube.com/watch?v=6xU7CydnCVc]‪My CNC Router‬‏ - YouTube[/ame]

[HomerSimpson]

I created a better video that provides better views and more explanation here:

[ame=http://www.youtube.com/watch?v=T9QU3YQe-Sc]‪My Homebuilt, Belt-on-Belt drive, CNC Router‬‏ - YouTube[/ame]

[PaulRowntree]

That looks terrific. I like the supported gear shafts, and it looks very solid. And for the $64K question (does anyone say that anymore?), how fast and precise is your belt system? Any sign of the belt loosening in use? And finally, how deep a channel do you have for the belts?

[HomerSimpson]

Paul,

I haven't fully tested speeds yet but I've run it at 1000 in/min with no machining load, and 250 in/min while machining. I don't have my spreadsheet in front of me with the precision calcs but as I recall, at 10X microstepping, I'm getting something like a 0.0001" per step, but as discussed previously, I don't think I can count on that precision during microstepping, so it is probably more like 0.001" per full step.

The depth of the channel is the height of the 2 belts when they are mated together.
Stan

[diyengineer]

I've finally taken some video of my CNC that uses the belt on belt drive design for the X & Y axis.

Take a look: ‪My CNC Router‬‏ - YouTube

is this drive kit available online somewhere or purely custom.

[greybeard]

Hi Stan.
Having built my own smaller set up, it was a bit of a learning curve to discover just how flexible my aluminium mounting plates were.
So I'm a bit concerned by the plates you have at the top and bottom of the z-axis where the bearings for the lead screw are mounted. They appear to be about 5mm thick ?
Is there any chance that the whole lead screw assembly can flex up and down by any small but measurable amount ?
Other than that, I'm more than a little envious of what you have achieved.
Regards
John

[ger21]

is this drive kit available online somewhere or purely custom.

Purely custom. It took me a considerable amount of time to come up with a method to use a belt reduction and stepper, as the original design uses a direct drive servo.

[HomerSimpson]

Hi Stan.
So I'm a bit concerned by the plates you have at the top and bottom of the z-axis where the bearings for the lead screw are mounted. They appear to be about 5mm thick ?
Is there any chance that the whole lead screw assembly can flex up and down by any small but measurable amount ?

John,
I must admit that I haven't vetted that part of the design thoroughly to know exactly how much deflection I'm seeing there. Been focused on the drives scheme for awhile. It is entirely possible though so I will keep a look out for that problem and stiffen it up (somehow!) if I see it. Thanks for the suggestion.
Stan

p.s. I should note that these plates are stainless steel though, not aluminum.

[louieatienza]

I created a better video that provides better views and more explanation here:

‪My Homebuilt, Belt-on-Belt drive, CNC Router‬‏ - YouTube

Great job! Nice to see one of these finally up and running!

[Mike Everman]

Yeah, nice work Homer. Someone else mentioned the thrust happening with your helical gear. That's the first place I would look if you have trouble with precision. Having a spring in the drivetrain (the motor's wave bearing preload wave washer) is not great, but may not show up in what you're doing if it's stiff enough.