Belt Drive Design Problem

[yarden.saa]

Hi CD,
For the first time I am trying to design a drivetrain with belts. I would like to keep everything in the minimum size.
**I will attach a photo of the drivetrain I want to convet to belts.
We are using a direct drive with hex shaft. the problem is that Vex doesn't offer pulleys it the same diameter as the hex pulleys. adding a hex hub over the pulleys will make the thickness of the wheel+pulley+hexhub+bolts+nuts too thick. I looked at andymark solution in their AM14U, they made holes in their pulley so the hub will fit in the pulley. I am looking for a solution that will work with vex wheels because andymark pulleys doesn't have the versa key pattern. Andymark variety of pulleys and belts is pretty small so I can allow my self using it.

[Chris is me]

Hi CD,
For the first time I am trying to design a drivetrain with belts. I would like to keep everything in the minimum size.
**I will attach a photo of the drivetrain I want to convet to belts.
We are using a direct drive with hex shaft. the problem is that Vex doesn't offer pulleys it the same diameter as the hex pulleys. adding a hex hub over the pulleys will make the thickness of the wheel+pulley+hexhub+bolts+nuts too thick. I looked at andymark solution in their AM14U, they made holes in their pulley so the hub will fit in the pulley. I am looking for a solution that will work with vex wheels because andymark pulleys doesn't have the versa key pattern. Andymark variety of pulleys and belts is pretty small so I can allow my self using it.

If you are using VersaWheels with Vex pulleys, you may not need to add a hex hub to both the wheel AND the pulley. Just use a hex hub on the wheel and maybe put a hex bearing in the pulley itself just to support the pulley on the shaft. You can rely on the screws and the VersaKey pattern to transmit torque between the pulley and the wheel. Just like the sprocket your old design has doesn't have a hex bore but works just fine, the same is likely true for pulleys.

[Travis Schuh]

Why are you looking to switch from chain to belts? I caution against using belts in situations where you do not have active tensioning. Chain (even #25) is more forgiving for a design like this.

[Monochron]

Why are you looking to switch from chain to belts? I caution against using belts in situations where you do not have active tensioning. Chain (even #25) is more forgiving for a design like this.

Can you explain why this is the case? As I understand it, belts are great if you have perfect C-C distance with no need for tensioning. Chain however, due to their slight stretching, does require tensioning.

[pfreivald]

1. We switched from chain to belts years ago, and are so glad we did. I would recommend that everybody do it!

2. If the wheel and the pulley are on the same hex shaft, the pulley shouldn't need to be attached to the wheel in any manner--the pulley turns the shaft, which turns the wheel.

[gurellia53]

Hi CD,
For the first time I am trying to design a drivetrain with belts. I would like to keep everything in the minimum size.
**I will attach a photo of the drivetrain I want to convet to belts.
We are using a direct drive with hex shaft. the problem is that Vex doesn't offer pulleys it the same diameter as the hex pulleys. adding a hex hub over the pulleys will make the thickness of the wheel+pulley+hexhub+bolts+nuts too thick. I looked at andymark solution in their AM14U, they made holes in their pulley so the hub will fit in the pulley. I am looking for a solution that will work with vex wheels because andymark pulleys doesn't have the versa key pattern. Andymark variety of pulleys and belts is pretty small so I can allow my self using it.

I think you want VersaPulleys (http://www.vexrobotics.com/vexpro/belts-and-pulleys/htdversapulley.html)

[sdcantrell56]

Can you explain why this is the case? As I understand it, belts are great if you have perfect C-C distance with no need for tensioning. Chain however, due to their slight stretching, does require tensioning.

Due to belt tolerances, the only way to get perfect c-c distances would be to lay out and measure the actual belt and pullies. This solves it for that particular belt but what happens when you have to replace it? Last time I checked the belt tolerance is +-.010 on the circumference so the chances of it being perfect with another belt are slim. With a means of tensioning the belt this problem disappears.

[Mike Marandola]

Why are you looking to switch from chain to belts? I caution against using belts in situations where you do not have active tensioning. Chain (even #25) is more forgiving for a design like this.

I agree that chain may be better in this design, but what about the AM14U? It has no active tensioning.

[Chris is me]

Can you explain why this is the case? As I understand it, belts are great if you have perfect C-C distance with no need for tensioning. Chain however, due to their slight stretching, does require tensioning.

We're now entering our 5th season with belts, and our 4th season using exact centers with no tensioners. As long as you pick a strong enough belt / pulley combination, this setup will have no problems at all. There is no noticeable stretch or wear over an FRC robot's life span. We put the drive together Week 4 and never touch it again. Perhaps for other applications you may want a tensioner, but in a drive we have done this year after year and it's just "set it and forget it".

Due to belt tolerances, the only way to get perfect c-c distances would be to lay out and measure the actual belt and pullies. This solves it for that particular belt but what happens when you have to replace it? Last time I checked the belt tolerance is +-.010 on the circumference so the chances of it being perfect with another belt are slim. With a means of tensioning the belt this problem disappears.

I would actually argue (it sounds totally nuts, but hear me out) that for the majority of teams, adding tensioners to a belt system will make it more likely to fail. For the precise, detail oriented teams (e.g. 971, 254, etc) tensioners are probably a performance boost to account for that tolerance.

However, for the average team, adding tensioners makes it very easy to over or under tension a belt. It is surprisingly easy to over-tension a belt as a "perfect" center distance belt has more slack than you would expect. Overtensioning a belt significantly weakens the system. In some specific cases with under-sized belts and pulleys, this can and absolutely has led to drive failure. Another common problem is include differing tension in two belts on the same driveline. Again, teams that pay a lot of attention to detail and design great tensioners can find success, but it's easier to fail a tensioned system than an untensioned system in my experience.

Despite the wider tolerance in belt length, we've just never had a problem doing it this way. It just works. We've done this to at least 16 individual belts in different drivetrains now.

[Travis Schuh]

I agree that chain may be better in this design, but what about the AM14U? It has no active tensioning.

My statement is colored by our experience, which was with small pulleys (22T I think) with 9mm wide 5mm GT3 belts. We know that we are running these belts out of spec, and it shows because we do have belts break. Previous years we used the same set of belts until practice for off-seasons, but last year we went through some belts during season practice due we think to the higher CG design. I would not feel comfortable running our current setup without a tension. For this year, we are looking at moving to 15mm belts so we are not as under the rated specs (we are currently running ~2X the rated load for the belts based on my memory of the calculations we did).

I see that the AM14U runs 42T pulleys and 15mm belts. This is should be 3X better on the loading than what we are doing (not including the rating difference of HTD vs GT3), but without running the life numbers, I still bet the belts are still close to the rating for this application. If you want to run belts without tensioners, I would follow with this pattern (looking back at your pictures, it appears like this is what you are doing).

Based on quick calculations from AM's listed weights, it looks like a AM14U has under 0.4lb of belts, and would require about 1lb of chain. I don't see that as a huge weight difference, particularly to pay for drivetrain reliability. Last year I saw a few WCD that chose to run belts with small pulleys, and were running competitions without wheels powered because the belts broke and it is very difficult to replace the belts. My opinion is that if you don't have a good plan for how to change a belt mid competition if it breaks (and preferably a way to tension the belts properly to help keep them from breaking), then you are probably better off with chain. This doesn't mean belts aren't working for teams, I just caution the mass movement to put belt drive trains.

[Deke]

Why are you looking to switch from chain to belts? I caution against using belts in situations where you do not have active tensioning. Chain (even #25) is more forgiving for a design like this.
My statement is colored by our experience, which was with small pulleys (22T I think) with 9mm wide 5mm GT3 belts. We know that we are running these belts out of spec, and it shows because we do have belts break. Previous years we used the same set of belts until practice for off-seasons, but last year we went through some belts during season practice due we think to the higher CG design. I would not feel comfortable running our current setup without a tension. For this year, we are looking at moving to 15mm belts so we are not as under the rated specs (we are currently running ~2X the rated load for the belts based on my memory of the calculations we did).

...(snip)...

Based on quick calculations from AM's listed weights, it looks like a AM14U has under 0.4lb of belts, and would require about 1lb of chain. I don't see that as a huge weight difference, particularly to pay for drivetrain reliability. Last year I saw a few WCD that chose to run belts with small pulleys, and were running competitions without wheels powered because the belts broke and it is very difficult to replace the belts. My opinion is that if you don't have a good plan for how to change a belt mid competition if it breaks (and preferably a way to tension the belts properly to help keep them from breaking), then you are probably better off with chain. This doesn't mean belts aren't working for teams, I just caution the mass movement to put belt drive trains.

Just curious if you can explain your team's design decisions on going with belts over chain for drive train. I don't have much experience with belts in a drive application, it would be helpful to understand the thought process.

It seems like there is a lot of caution with your advice on belted drives, and chain seems like the superior approach with this guidance.

[pfreivald]

We're now entering our 5th season with belts, and our 4th season using exact centers with no tensioners. As long as you pick a strong enough belt / pulley combination, this setup will have no problems at all. There is no noticeable stretch or wear over an FRC robot's life span.

Agreed. 15mm belts are plenty durable for the abuse of one FRC season--though you should design your drive for easy swap-outs just in case.

[Chris is me]

I see that the AM14U runs 42T pulleys and 15mm belts. This is should be 3X better on the loading than what we are doing (not including the rating difference of HTD vs GT3), but without running the life numbers, I still bet the belts are still close to the rating for this application. If you want to run belts without tensioners, I would follow with this pattern (looking back at your pictures, it appears like this is what you are doing).

To add another data point, my team has converged on 24T pulleys and 15mm wide belts for 4" and 6" wheels without incident.

These are my rules of thumb from the last few years of drivetrains, what i've seen other teams do, etc. Use at your own risk, your mileage may vary:
- 15mm can run 24T and larger without tensioners in a "standard" traction tank drivetrain
- 9mm can run ~36T and larger without tensioners in a "standard" traction tank drivetrain. (I'm less sure on the number I'd start being okay using 9mm belts with)
- 9mm belts with 24T and smaller pulleys run a high risk of failure in a "standard" traction tank drivetrain - I would try at least 27T if you must use the 9mm profile.

Really, the big takeaway here is that 15mm belt is so much more forgiving than 9mm belt that I would much rather run 15mm than 9mm in any drive.

[Travis Schuh]

Just curious if you can explain your team's design decisions on going with belts over chain for drive train. I don't have much experience with belts in a drive application, it would be helpful to understand the thought process.

It seems like there is a lot of caution with your advice on belted drives, and chain seems like the superior approach with this guidance.

It isn't as dreary as I paint it. We have just been recently thinking about how to get that last bit of reliability and performance out of our drive train, so this has been on my mind.

971 uses belts primarily because the pulleys integrate into our design better than a sprocket (we can bore out and glue modified COTS pulleys into our integrated wheel module, where there isn't a COTS sprocket that I know of that we could make do this). Beyond that, there is a nice benefit that belts are lighter than chain and run pretty quiet. If we ran a WCD, I would run #25 chain like 254 does. It turns out that #25 is also out of spec for a drive application, but it appears to handle it more gracefully.

[Oblarg]

Re: 9mm versus 15mm belts, it's worth noting when calculating the loading that in a 6-wheel drive train, the center wheels are taking far more load than the front and back ones. I'm much more comfortable running 9mm from a center wheel to the outer wheels in a 6WD than I would be going between center wheels on an 8WD.

4464's current preseason design uses 9mm belts, simply because it's extremely convenient to be able to only have one pulley on the center wheel. We're also using 42-tooth pulleys.

[yarden.saa]

Thanks, I found a way that I like.
See attachments

[philso]

In your latest renderings, it looks like the pulleys on the two middle wheels on each side will have two belts on them. Will there be any problems with the edges of the two belts rubbing against each other and causing wear of some sort?

[Chris is me]

In your latest renderings, it looks like the pulleys on the two middle wheels on each side will have two belts on them. Will there be any problems with the edges of the two belts rubbing against each other and causing wear of some sort?

Provided proper tension or exact spacing is used, nope, teams have done this for years without issue. Remember, the belts are not in motion relative to each other when they're on the pulley.

Re: 9mm versus 15mm belts, it's worth noting when calculating the loading that in a 6-wheel drive train, the center wheels are taking far more load than the front and back ones. I'm much more comfortable running 9mm from a center wheel to the outer wheels in a 6WD than I would be going between center wheels on an 8WD.

Many teams with similar setups are doing the exact same thing, actually. I wouldn't use six wheel drop alone as a reason to go to 9mm. 42T pulleys are more than big enough.

4464's current preseason design uses 9mm belts, simply because it's extremely convenient to be able to only have one pulley on the center wheel. We're also using 42-tooth pulleys.

IMO this is a really bad reason to pick a profile - you can just buy two 7mm slices with your pulleys to make them 32mm wide. That said, 9mm belts on 42T pulleys doesn't sound too bad.

[Oblarg]

IMO this is a really bad reason to pick a profile - you can just buy two 7mm slices with your pulleys to make them 32mm wide.

Sure, we could do that, but it would a) cost more money and b) take up about an additional inch of robot width.

Looking at the rated torques, I think it extremely unlikely that we'll ratchet or break a belt with our setup (especially given that the torque ratings Gates gives are for extended use, and are somewhat lower than the effective maximum torque for the short service times they see in an FRC robot). 449 used 9mm belts on 36t pulleys without incident last year.

[Travis Schuh]

Re: 9mm versus 15mm belts, it's worth noting when calculating the loading that in a 6-wheel drive train, the center wheels are taking far more load than the front and back ones. I'm much more comfortable running 9mm from a center wheel to the outer wheels in a 6WD than I would be going between center wheels on an 8WD.

While I would agree that on average this is true, I would argue that there can be cases where the outer wheels will be loaded with all of the robot torque, and thus it is appropriate to run the FOS calcs assuming this as worst case. Depending on the height of your CG, you can get significant load transfer to your outer wheels when accelerating. If you accelerate and the robot goes up on its back wheels, then the outer wheels are taking all of the torque (our robot this past year would do this).

[AdamHeard]

A few more data points on pulley sizes, etc... :rolleyes:

https://docs.google.com/spreadsheets/d/1yPFHwloNqrz5OGq9FsCOX94ca3qBXpoNBaQRHZdX7UU/edit#gid=0


While I would agree that on average this is true, I would argue that there can be cases where the outer wheels will be loaded with all of the robot torque, and thus it is appropriate to run the FOS calcs assuming this as worst case. Depending on the height of your CG, you can get significant load transfer to your outer wheels when accelerating. If you accelerate and the robot goes up on its back wheels, then the outer wheels are taking all of the torque (our robot this past year would do this).

Very much agreed, giant mistake to neglect this case.

[philso]

Provided proper tension or exact spacing is used, nope, teams have done this for years without issue. Remember, the belts are not in motion relative to each other when they're on the pulley.

I am more concerned with the radial motion of the belt towards the pulley when the belt is coming onto the pulley and the radial motion of the belt away from the pulley as it is leaving the pulley. I am interested to know if the total run time in an FRC robot application will cause significant wear.

[Chris is me]

I am more concerned with the radial motion of the belt towards the pulley when the belt is coming onto the pulley and the radial motion of the belt away from the pulley as it is leaving the pulley. I am interested to know if the total run time in an FRC robot application will cause significant wear.

I was also speaking from experience here. We cram belts right next to each other on pulleys without flanges and we've never had significant wear on the belts, even on the belts with over 100 hours of run time. This is totally fine in FRC. The window for belts not moving in the same direction to contact each other is very very small.

Looking at the rated torques, I think it extremely unlikely that we'll ratchet or break a belt with our setup (especially given that the torque ratings Gates gives are for extended use, and are somewhat lower than the effective maximum torque for the short service times they see in an FRC robot). 449 used 9mm belts on 36t pulleys without incident last year.

I wasn't very clear in my post - I was trying to say 9mm / 42T is likely just fine. I just meant that picking 9mm over 15mm just because of the double pulley feature, independent of any other features, is not great.

[Joey Milia]

In your latest renderings, it looks like the pulleys on the two middle wheels on each side will have two belts on them. Will there be any problems with the edges of the two belts rubbing against each other and causing wear of some sort?

192 hasn't had a problem with this on the 5mm pitch drive belts that share a pulley. However on the prototypes of the 2013 gearbox the higher tension 3mm pitch belts would drift more and rub. This was made worse by their arrangement where they crossed over each other. We saw the sides of the belts wearing down so we added a flange on the final version to prevent the rubbing. Should be noted no damage was done to the tensile members and only to the rubber backing.

[Oblarg]

I wasn't very clear in my post - I was trying to say 9mm / 42T is likely just fine. I just meant that picking 9mm over 15mm just because of the double pulley feature, independent of any other features, is not great.

Well, obviously we didn't make the decision in a vacuum, but it was the primary reason we considered 9mm in the first place.

[snoman]

H
adding a hex hub over the pulleys will make the thickness of the wheel+pulley+hexhub+bolts+nuts .

why can't you make your off sets longer or move frame rail in

[DampRobot]

Last year I saw a few WCD that chose to run belts with small pulleys, and were running competitions without wheels powered because the belts broke and it is very difficult to replace the belts. My opinion is that if you don't have a good plan for how to change a belt mid competition if it breaks (and preferably a way to tension the belts properly to help keep them from breaking), then you are probably better off with chain. This doesn't mean belts aren't working for teams, I just caution the mass movement to put belt drive trains.

Travis, I assume you're referring to 100. For those that don't know, 100 ran 9mm vex belts last year on 4" wheels (on 24t pullies IIRC, it's been a while). They ran fine all through SAC, but at SVR we broke 4 separate belts. At first, it seemed like an overtension failure, but we replaced the belts and I personally made sure they were undertensioned, they failed in just the same way. As it took a while to replace the belts, we did run without driving some wheels in some matches just to make it out onto the field.

At this point, I blame those failures on a bad run of VP belts rather than the inherent disadvantages of belts or non center to center designs. However, the whole experience tempered my enthusiasm for belts. One of the things that most people fail to realize with belts is that you have to do your whole assembly with them in place, unlike chain, which you can slip around and masterlink after everything's assembled. For WCDs, this means you have to take apart your gearboxes if a belt breaks.

Having done both belt and chain WCDs in the past, I would probably advise on using #25 chain. The weight savings of belts are tiny, noise shouldn't be a major issue and for most teams the efficiency gains of belts could be more easily gotten in other places. On the other hand, teams can definitely be successful with belts even their first time around as long as they spec their components conservatively and have spares.

[Chris is me]

At this point, I blame those failures on a bad run of VP belts rather than the inherent disadvantages of belts or non center to center designs.

I really don't think it's either of these - I think 9mm belts on 24T pulleys are just very undersized for your application. You will likely continue to have failures even with other brands of belts. If you can't go above 24T pulleys, you have to go to 15mm wide belts if you want to be safe.

(I know you are probably aware of this by now, but I just wanted to say this publicly for the reference of others reading this thread.)

[AustinSchuh]

I really don't think it's either of these - I think 9mm belts on 24T pulleys are just very undersized for your application. You will likely continue to have failures even with other brands of belts. If you can't go above 24T pulleys, you have to go to 15mm wide belts if you want to be safe.

(I know you are probably aware of this by now, but I just wanted to say this publicly for the reference of others reading this thread.)

971's experience has been that 24 T pulleys are just on the edge, though we have been using 3.5" wheels. The old GT2 belts weren't quite strong enough (failed after a hard season), but the newer GT3 belts are just barely strong enough to last a long season. We haven't had a GT3 belt fail, once we failed a set in the first hour from over-tensioning. If you believe the Gates manuals, GT2 and GT3 belts are quite a bit stronger than HTD, and everything doesn't seem rated for long life with the loads they'll see in FRC.

We have found that proper tension is crucial. If the belts are under-tensioned, they will ratchet. If they are over-tensioned, they will break. The line seems to be finer than we'd like. We run lots of un-tensioned belts, but will never run an un-tensioned drive belt. The reward just isn't there.

That being said, we are going to run 15mm wide next year for that extra peace of mind. We can see the pulleys starting to wear.

[Joe G.]

Curious -- has anyone in this thread experimented with belt pitches other than 5mm? A quick look at the gates datasheet suggests that a 22t (minimum size) 8mm pitch sprocket with a 12mm wide belt will have slightly more carrying capacity than a 15mm wide 5mm pitch belt with a similar sprocket diameter, and significantly more than the equivalent 9mm wide belt. I'm always trying to squeeze some extra width out of drive designs, but have heard just enough 9mm belt horror stories to discourage me from pursuing them.

[AustinSchuh]

Curious -- has anyone in this thread experimented with belt pitches other than 5mm? A quick look at the gates datasheet suggests that a 22t (minimum size) 8mm pitch sprocket with a 12mm wide belt will have slightly more carrying capacity than a 15mm wide 5mm pitch belt with a similar sprocket diameter, and significantly more than the equivalent 9mm wide belt. I'm always trying to squeeze some extra width out of drive designs, but have heard just enough 9mm belt horror stories to discourage me from pursuing them.

5mm is the biggest that is still really easy to source. We've never had a good enough reason to be worth the procurement challenge.

[asid61]

5mm is the biggest that is still really easy to source. We've never had a good enough reason to be worth the procurement challenge.

You can get 0.375" (9.52mm) from SDP-SI, although if they're out of stock you're out of luck. The decision would have to be made very early on.

EDIT: Never mind, the profile is low-load. Not good for drivetrain usually.

[pfreivald]

SDP-SI is awesome. We've used their center-to-center calculator since we started using belts, and have never had a problem. (But we've never used 9mm, either.)

[Travis Schuh]

You can get 0.375" (9.52mm) from SDP-SI, although if they're out of stock you're out of luck. The decision would have to be made very early on.

The .375 profile is a L profile, not GT2 or GT3. This is a trapezoidal profile, so its load rating is significantly less than even HTD for a similar size. (L stands for Light, while HTD stands for High Torque Drive). The strength of this profile is very likely less than 9mm wide 5mm HTD. The reason you would use a trapezoidal profile is if you have small loads but need minimal backlash, as it will perform better than a HTD or GT2 profile.

[asid61]

The .375 profile is a L profile, not GT2 or GT3. This is a trapezoidal profile, so its load rating is significantly less than even HTD for a similar size. (L stands for Light, while HTD stands for High Torque Drive). The strength of this profile is very likely less than 9mm wide 5mm HTD. The reason you would use a trapezoidal profile is if you have small loads but need minimal backlash, as it will perform better than a HTD or GT2 profile.

Oh. I thought it stood for "large". Totally forgot about profile.