Calling all Belt Drive Experts

FZBob

100 W
Joined
May 30, 2011
Messages
181
Location
San Jose
My single reduction mid drive is working very well with 10:1 reduction using #25 chain. However, I can't leave well enough alone, and I'm curious to see if a belt drive might work.

https://endless-sphere.com/forums/viewtopic.php?f=28&t=96769&p=1426882#p1426882

Here are the design parameters:
At least 10:1 reduction ratio
450 Watts at 60 RPM Cadence, which works out to about 175 lb belt tension. (May be able to reduce watts/tension just a bit...)
7.2" max pulley diameter
.575" large pulley width, so 1/2" wide belt (Maybe 15mm with large effort)
Dust for sure, and possibly dirt.
Up to .010" radial runout on large pulley
Prefer not to use an idler/tensioner, or crazy high tension

I would appreciate any suggestions on design guides, sources for pulleys & belts (off the shelf and custom...), and general thoughts.

I took a quick look, and it appears that this is a difficult application for a belt. However, the E-Skateboard guys seem to make belts work with even larger motors. I'll probably also post over on www.electric-skateboard.builders. They have cool stuff...
 
Thanks for posting your single reduction set-up, it has inspired me to chase the lightweight/low-power set-up dream again...

I'm certainly no expert, but I've been tinkering around with 3D printed HTD5M pulleys and #25 sprockets for a few years on various esk8 and other LEV projects. I think 15mm HTD5M belts/pulleys would be the best/easiest way to proceed- you'd probably get away with 12mm belts on this low power stuff, but sourcing large 12mm belts will be an issue.

As to motor/driver pulley, 12t is the smallest being used successfully in the esk8 world. You will almost certainly need to use an idler with such a small pulley, but it's been my experience that a well designed idler doesn't increase system drag much at all, at least on skateboards.

To get your 10:1, you will need a 120t pulley on the cranks. This sprocket is nearly 190mm in diameter (roughly 7.5 inches). Here's what that looks like compared to a 44t (pink) and a 36t (rust/orangey-red) chainring:

120tvsSprocket.jpg

It's a bit of a monster, and will no doubt cost a fair penny to have one machined. If it were me, I'd 3d-print one- I have no doubt that one printed in nylon would stand up to the power levels we're talking here. Though, such a big diameter pulley is begging to get smashed up off-road.

Here's a quick mock-up of how I'd build it:


A right side view:
RightSide.jpg

One from the bottom:
Bottom.jpg

And the other side:
View attachment 2

Frame used in this model is a early 2000's Specialized Big Hit (model from grabcad). It's shown with a 68mm BB shell, 48mm chainline, 135mm rear drop-out width and a 38t chainring.

I've been pondering your set-up for a little while now and I've got another way (or two) to skin this cat. I'll post back here in a day or two with some models to show my idea(s). Thanks again for all the info you've shared with your project!
 
I recommend something between a 2:1 or 3:1 belted reduction on the primary to drive a jackshaft, and a chain-drive for the final reduction.

If you are certain that you want a single-reduction, then I have no suggestions.
 
Qwiksand said:
Thanks for posting your single reduction set-up, it has inspired me to chase the lightweight/low-power set-up dream again...

I'm certainly no expert, but I've been tinkering around with 3D printed HTD5M pulleys and #25 sprockets for a few years on various esk8 and other LEV projects. I think 15mm HTD5M belts/pulleys would be the best/easiest way to proceed- you'd probably get away with 12mm belts on this low power stuff, but sourcing large 12mm belts will be an issue.

As to motor/driver pulley, 12t is the smallest being used successfully in the esk8 world. You will almost certainly need to use an idler with such a small pulley, but it's been my experience that a well designed idler doesn't increase system drag much at all, at least on skateboards.

To get your 10:1, you will need a 120t pulley on the cranks. This sprocket is nearly 190mm in diameter (roughly 7.5 inches). Here's what that looks like compared to a 44t (pink) and a 36t (rust/orangey-red) chainring:

120tvsSprocket.jpg

It's a bit of a monster, and will no doubt cost a fair penny to have one machined. If it were me, I'd 3d-print one- I have no doubt that one printed in nylon would stand up to the power levels we're talking here. Though, such a big diameter pulley is begging to get smashed up off-road.



Frame used in this model is a early 2000's Specialized Big Hit (model from grabcad). It's shown with a 68mm BB shell, 48mm chainline, 135mm rear drop-out width and a 38t chainring.

I've been pondering your set-up for a little while now and I've got another way (or two) to skin this cat. I'll post back here in a day or two with some models to show my idea(s). Thanks again for all the info you've shared with your project!

Well, you certainly have my attention! Thanks for drawing up this pulley system, pretty impressive. If the cost is not too bad, I'm tempted to do a risk build. (probably cost less than my motor experiments... :roll: )

I really like the idea of a 3D printed nylon pulley. Would you care to speculate on ballpark pricing? (my experience with 3D printing was 15 years ago...)

My belt would need to be on the order of 32" long, and I would go 15 mm wide. Any suggestions on where to source?

The skateboard guys seem to like the HTD 5 belt, but are concerned about strength/durability in this application. The SDP/SI technical website also indicates the belt might be weak.
http://www.sdp-si.com/D265/PDF/D265T146.pdf
I wonder how much durability/lifetime is baked into those tables. If I got 50-100 hours out of a belt, I think I'd be pretty happy.
 
Following this thread with much interest!
Although I think instead of using a belt drive you might also consider a big 3D printed ring gear (beside the crank chain wheel) and a small metal pinion gear (on the motor) with a fine enough pitch to allow more than the 1:10 reduction. Maybe try to get to 1:20 even. Basically a low power, single stage version of the CYC X1-pro gear drive:
[youtube]4l7unymRrYo[/youtube]
 
Just went down the rabbit hole on Tom Stanton's Ebike build series on YouTube; he's been finding great success with 15mm wide HTD5M belts (up to 1500mm length) and large 3D printed pulleys- his latest V4 configuration shows him running peaks of 4kw through the above mentioned set-up. Here's the first video in the series where he is running a 12t motor pulley to a 128t printed rear wheel pulley (in PLA no less, which is strong but far from heat tolerant):

[youtube]9rIIJbDkpN8[/youtube]

You can see there's a fair amount of wobble/run-out in the rear pulley and it doesn't seem to make much difference. Based on this video alone, I think your idea of running 10:1 directly to the cranks will be no problem at all.

The belts you need are readily available from both Vbeltsupply and VexRobotics- I've ordered from both companies and have had no issues.

As to printing the pulley, going through a printed service like Shapeways will likely be cost prohibitive, so I highly recommend you look into buying yourself one, the costs have come way down as of late.

$180 will get you the Anet A8 which has some short comings, but most of which are already addressed/remedied by the internet hive mind.

A little more than double your money and you can get the highly praised $400 CR-10.

Double your money again with the $750 Prusa Mk3 and you'll have a printer that works straight out the box with a huge internet support network already in place.

Here's a link to the STL and the STEP files for the crank pulley I modelled earlier- the tooth profile I've used has been successful on pulleys up to 60t, but I've not printed anything larger. My current printer is 6 years old and the largest diameter I can do is roughly 165mm. I'm currently building a new printer, but it's been a slow project (though I did finally order the majority of the parts needed this last black friday!).
 
SlowCo said:
... I think instead of using a belt drive you might also consider a big 3D printed ring gear (beside the crank chain wheel) and a small metal pinion gear (on the motor) with a fine enough pitch to allow more than the 1:10 reduction. Maybe try to get to 1:20 even. Basically a low power, single stage version of the CYC X1-pro gear drive...

Thanks for that video, SlowCo- first time I've seen it. Hope they upload some actual product/action videos shortly, It looks pretty fantastic.

As to a crank mounted 3D printed directly driven via motor pinion- that was what I was cryptically referring to when I posted the "another way (or two) to skin this cat" in my first post. I hope to get some free time tonight to model and post my ideas.

The large modulus gears they've used in their design makes me wonder if they've had to go this route because of strength issues or if this is to be more dirt/debris tolerant with the exposed gears...or perhaps a combination of the two.
 
"Crazy high tension" by chain drive standards is just part of the deal with belts. The design has to tolerate those forces. Ideally, you support any shaft that has a belt sprocket on it with bearings on both sides of the sprocket. This isn't always practicable, and it always makes replacing the belt more complicated. Larger diameter shafts and bigger, stronger bearings than would otherwise be necessary are the usual remedy.

This is the key problem that makes bicycle belt drive uncommon.
 
Qwiksand said:
Thanks for that video, SlowCo- first time I've seen it. Hope they upload some actual product/action videos shortly, It looks pretty fantastic.

As to a crank mounted 3D printed directly driven via motor pinion- that was what I was cryptically referring to when I posted the "another way (or two) to skin this cat" in my first post. I hope to get some free time tonight to model and post my ideas.

The large modulus gears they've used in their design makes me wonder if they've had to go this route because of strength issues or if this is to be more dirt/debris tolerant with the exposed gears...or perhaps a combination of the two.

There is already two threads on this mid drive set up here on E-S:
https://endless-sphere.com/forums/viewtopic.php?f=28&t=95704 = crank drive (chain or gear)

https://endless-sphere.com/forums/viewtopic.php?f=28&p=1427706 = direct drive project

As it is a 3000W double reduction drive it needs a chain or large modulus reinforced (PEEK) gear to tolerate the generated torque. It's a very promising drive.
 
SlowCo said:
a big 3D printed ring gear (beside the crank chain wheel) and a small metal pinion gear (on the motor) with a fine enough pitch to allow more than the 1:10 reduction. Maybe try to get to 1:20 even. Basically a low power, single stage version of the CYC X1-pro gear drive:

I think this has some real potential. When I get a bit of free time I need to do some gear calcs...

It would require me to build a different motor mount, and possibly some shielding to keep pebbles out of the mesh.
 
Qwiksand said:
Just went down the rabbit hole on Tom Stanton's Ebike build series on YouTube; he's been finding great success with 15mm wide HTD5M belts (up to 1500mm length) and large 3D printed pulleys- his latest V4 configuration shows him running peaks of 4kw through the above mentioned set-up. Here's the first video in the series where he is running a 12t motor pulley to a 128t printed rear wheel pulley (in PLA no less, which is strong but far from heat tolerant):

You can see there's a fair amount of wobble/run-out in the rear pulley and it doesn't seem to make much difference. Based on this video alone, I think your idea of running 10:1 directly to the cranks will be no problem at all.

The belts you need are readily available from both Vbeltsupply and VexRobotics- I've ordered from both companies and have had no issues.


Here's a link to the STL and the STEP files for the crank pulley I modelled earlier- the tooth profile I've used has been successful on pulleys up to 60t, but I've not printed anything larger. My current printer is 6 years old and the largest diameter I can do is roughly 165mm. I'm currently building a new printer, but it's been a slow project (though I did finally order the majority of the parts needed this last black friday!).

Interesting Video. That does seem very promising for what I'm looking at. His motor almost certainly puts more stress on the belt than mine would...

If it's OK with you, I'll send out your step file to a few 3D print houses for budgetary quotes.
 
Qs motors are now selling motors for mid drives that were originally set up as belt drive. I’ve ordered the smallest 1000w version. I’d expect to see more belt drives in the future, at least for the primary reduction as mr magnets states.

In the market for a 3d printer look at the alfawise U10 from gearbest. It’s 400x400x500mm. It blows the cr-10 stuff away. Under $500. I’m printing 15” sign letters with it, and the ebike stuff. It’s actually $399 now and ships free in the USA. Jump on this folks!

https://m.gearbest.com/3d-printers-3d-printer-kits/pp_1596287.html?wid=1859758

To compare $2700 US company in Brooklyn NY, same printer basically no heated bed though. Says American made but doesn’t look entirely true, rebranded with some add ons perhaps. Good old American mark-up!

https://shop.gcreate.com/collections/gmax-3d-printers/products/the-gmax-1-5-xt-16x16x21-the-next-big-thing-in-3d-printing

Get the U10 my friend.
 
Qwiksand said:
As to printing the pulley, going through a printed service like Shapeways will likely be cost prohibitive

Here's a link to the STL and the STEP files for the crank pulley I modelled earlier- the tooth profile I've used has been successful on pulleys up to 60t, but I've not printed anything larger. My current printer is 6 years old and the largest diameter I can do is roughly 165mm. I'm currently building a new printer, but it's been a slow project (though I did finally order the majority of the parts needed this last black friday!).

So I got a few quotes.

Shapeways - $100 in "Versatile Plastic" (some type of Nylon), and $203 in "Professional Plastic" (some other type of nylon)
(Why I hate Marketing guys...)

3D Hubs - $127 in Standard Nylon (SLS) - PA 12 (These guys know how to specify materials...)

Xometry - $127 in Nylon 12 (SLS) (These guys also know how to specify materials...)

2 more quotes on the way. The $100 is getting interesting...

I'm tempted to try removing about 1/3 of the material, and re-quoting. Can anyone suggest cheap/free software to edit STEP/STL files? I have Solidworks from about 2003, many computers ago, but I bet if I spent a day getting it installed onto my current computer, it wouldn't open a new file... I'm actually an old Autocad guy...

Buying a 3D printer is tempting, but my garage is full of tools, and my wife would kill me if I bought another. (Unless they collapse down flat and can be hidden...)
 
FZBob said:
...Can anyone suggest cheap/free software to edit STEP/STL files?


Two great free programs: Onshape and Fusion 360

Onshape is browser based and works on just about any machine, but the free version requires all of your designs to be visible to other users- no private storage. Onshape was the first parametric cad program I used, I still use it to share designs/ideas easily. Here's the Parametric Pulley Generator I modeled a while back, just change the two variables in the left hand column to make any number tooth HTD5M pulley of any thickness you desire.

Fusion 360 is very impressive, but has some minimum requirements for computer power (though most relatively modern machines should run it). I switched to fusion a little over a year ago and like the joints/components workflow much better than Onshape. Fusion also has an awesome built-in CAM package. They make you download as a "free trial", but when that is up you can use it for free as a hobbyist as long as you're not making too much money off your designs (like less than $100k, or something like that).

Here's the fusion 360 file for the 120t HTD5M 104mm Crank Pulley.

For prototype purposes, you might also look at printing it in PETG (FDM) by 3DHubs, I think it was going to be under $50 when I checked it out. I've used PETG on several skateboard pulleys and even #25 sprockets and it holds up well, just not as impact resistant as the nylons.
 
I use ViaCad Pro 9 for my CAD work.

I sliced the 120 pulley and it came out to 9 hrs and 22 mins. This is with a .3mm layer height and a solid fill for strength and machinability if needed. 60mm/second for all but first layer, PTEG. This would be $140 using my pricing scheme. I'm sure I can do some favors/barter for ES members. Can this pulley be designed larger and perhaps with a different center or solid filled? I can import it into ViaCad but can not convert it to a solid for modeling I'm limited to 3000 facets it seems. I have no immediate need for a pulley right now but I feel I might. I have a new ride I'm putting together and it could use a belt and pulley, we will see, motor waiting to ship.
 

Attachments

  • FEC14CE5-C99A-4247-AFF2-A488E17C602F.jpeg
    FEC14CE5-C99A-4247-AFF2-A488E17C602F.jpeg
    50.7 KB · Views: 4,197
lets try it

It feels very strong. I had to slow down the print infill and put tape on the bed leveling screws as the printer shakes like crazy when printing the teeth on the pulley. A good challange though nonetheless. So one failed print attempt. Actually printed in 7hrs 49 min. Remember larger print layers are stronger and smaller layers look better. Some clean up/detailing at the teeth of the pulley may be required.

FZbob, PM address i'll mail it to you.
Qwicksand, is the sprocket your work? I think the motor I've orderd has a larger size belt size than this one, I'll know for sure when it arrives. Any pointers to designing a pulley like this, figuring out the tooth profile and copy/rotating it perfectly?
 

Attachments

  • IMG_3469.JPG
    IMG_3469.JPG
    42.8 KB · Views: 4,264
  • IMG_3470.JPG
    IMG_3470.JPG
    53.8 KB · Views: 4,265
Skaiwerd said:
lets try it

It feels very strong. I had to slow down the print infill and put tape on the bed leveling screws as the printer shakes like crazy when printing the teeth on the pulley. A good challange though nonetheless. So one failed print attempt. Actually printed in 7hrs 49 min. Remember larger print layers are stronger and smaller layers look better. Some clean up/detailing at the teeth of the pulley may be required.

FZbob, PM address i'll mail it to you.
Qwicksand, is the sprocket your work? I think the motor I've orderd has a larger size belt size than this one, I'll know for sure when it arrives. Any pointers to designing a pulley like this, figuring out the tooth profile and copy/rotating it perfectly?
Wow, that's pretty impressive! PM sent...

I guess I'd better fire up the Autocad, and figure out belt length and Idler design.

Does this look like the proper 12 tooth pulley?
https://www.beltingonline.com/12-tooth-htd5-pulley-12-5m-15f-7708
I'll arrange for an 8mm bore, 3mm keyway, and maybe some grubscrews...

Do I need to arrange for an outer flange on the 120 tooth pulley? The 12 tooth has flanges on both sides, and the idler will not have flanges.
 
FZBob

That does look like the pulley. Can you design your bike motor to not need an idler? I'm planning to avoid them if I can. It would be easier to add the 2nd flange as a seperate part. It would need support printed with the model in the tooth area, that area is too small to work. Could be a bash guard of sorts, printed in a contrasting color perhaps.
 
Skaiwerd said:
FZBob

That does look like the pulley. Can you design your bike motor to not need an idler? I'm planning to avoid them if I can. It would be easier to add the 2nd flange as a seperate part. It would need support printed with the model in the tooth area, that area is too small to work. Could be a bash guard of sorts, printed in a contrasting color perhaps.

Good Idea - I think I might have long enough slots in the motor mount so the idler can be optional. I would also prefer not to have one.

No problem adding a flange, I was not sure it was necessary.
Just for fun - these guys only run flanges on the idler pulley - Probably not directly comparable...
blower belt a.jpg
 
Doing a crank drive belt drive is a tougher problem to solve than a rear wheel mount for a given power level. If we want 60-70Nm of crank torque, using a pulley that is about the same diameter of 32T chainring, we end up with 225 lbs of peak drive tension. Static tension for a 9mm wide, 5mm pitch GT2 belt in this config should be about 250lbs according to Gates documentation. This is quite a bit of load on a cantilevered motor shaft and bearings, as Chalo noted above. Making pulleys bigger diameter should help quite a bit, and maybe allow 3mm pitch belt that would require less static preload tension, but I wanted to keep everything small, for ground clearance. Also, 5mm pitch belt should be tolerant of debris.
Both pulleys shown are 3D printed HTPLA. Pretty amazing stuff, and it prints really well.
 
PaulD said:
Doing a crank drive belt drive is a tougher problem to solve than a rear wheel mount for a given power level. If we want 60-70Nm of crank torque, using a pulley that is about the same diameter of 32T chainring, we end up with 225 lbs of peak drive tension. Static tension for a 9mm wide, 5mm pitch GT2 belt in this config should be about 250lbs according to Gates documentation. This is quite a bit of load on a cantilevered motor shaft and bearings, as Chalo noted above. Making pulleys bigger diameter should help quite a bit, and maybe allow 3mm pitch belt that would require less static preload tension, but I wanted to keep everything small, for ground clearance. Also, 5mm pitch belt should be tolerant of debris.
Both pulleys shown are 3D printed HTPLA. Pretty amazing stuff, and it prints really well.
beltdrive.jpg

Nice looking setup!

I'm curious to see if it's possible to run a belt at much lower tension. I wouldn't be surprised if there is some margin in that 250 lb figure. Maybe I can trade tension for belt life, as long as the belt is not skipping (much...)

At 250 lb, probably with a cyclic component, I would be very concerned about fatigue life and cracking my aluminum bike frame. My motor mount is bolted to the frame with Rivet Nuts, so the load is concentrated.

Have you seen any issues with dirt or pebbles getting in the belt mesh?
 
Skaiwerd said:
Can you design your bike motor to not need an idler? I'm planning to avoid them if I can.

It looks like the same 825mm belt can be used both with and without an idler. And the belt is only $10.98 at V belt supply!
Belt Length a.jpg
 
FZBob said:
Nice looking setup!

I'm curious to see if it's possible to run a belt at much lower tension. I wouldn't be surprised if there is some margin in that 250 lb figure. Maybe I can trade tension for belt life, as long as the belt is not skipping (much...)

At 250 lb, probably with a cyclic component, I would be very concerned about fatigue life and cracking my aluminum bike frame. My motor mount is bolted to the frame with Rivet Nuts, so the load is concentrated.

Have you seen any issues with dirt or pebbles getting in the belt mesh?

Thanks! I had some issues with the meshing of the belt into the large pulley properly without decent tension. I adjusted the tooth design to compensate. I'll plan on continuing to decrease tension to see if skipping becomes and issue. I also increased the depth of the grooves to make room for dirt and pebbles. I haven't put on enough miles off-road to know if this is sufficient. I like the pulley design that these guys did, lots of room for mud and gunk: https://www.veercycle.com/products/synchronous-belt-conversion-kit-for-multi-speed-bike-frames

Overall, I think chain is a better solution - lower bearing/shaft loads, less efficiency loss when pedaling without the motor on (there is noticeable belt drag since my motor freewheel is upstream of the belt). It is nice to have less maintenance and less oil and grease (but there's still the bike chain!). The trade-off is tough to justify. It looks cool, though!
 
PaulD said:
Static tension for a 9mm wide, 5mm pitch GT2 belt in this config should be about 250lbs according to Gates documentation.

Just to clarify, 250lbs is the total load on the motor shaft (125 lbs per span) per the equation:
belt tension equation.png
I think this is unnecessarily high... we'll see.
Also, the load ratings for these belts are super conservative. (below are for the smaller pulley)
belt ratings torque.png

Probably intended for machines that run 24/7. Interestingly, Brose uses a 5mm pitch, 15mm wide (as far as I can tell from pictures) and claims 90Nm of torque. I think they have a large-ish front pulley (and lower hence less speed reduction) which helps.
 
Back
Top