Drive Train Design - Weight & Belts

[dellagd]

As I'm sure some of you guys are I have been having some fun lately going through CAD'ing different possible drive train designs. I have a few concerns though.

Three questions:

1) What is a good goal for total weight of the drive train? This would include all frame, wheel, axle, gearing, and motors.

2) I've seen lots of discussion on the smallest you should make your pulleys on a belt driven drive train. Would I be safe to use 18T Pulleys on an exact c-c distance setup with 15mm belts? I know you definitely shouldn't go that small with 9mm.

3) Should I be using the WCP Belt Calc to get me exact c-c distance or do I need to do some more calculations?

Hopefully you drive train guys out there can give me a hand. Thanks!

[Oblarg]

Originally Posted by dellagd

As I'm sure some of you guys are I have been having some fun lately going through CAD'ing different possible drive train designs. I have a few concerns though.

Three questions:

1) What is a good goal for total weight of the drive train? This would include all frame, wheel, axle, gearing, and motors.

2) I've seen lots of discussion on the smallest you should make your pulleys on a belt driven drive train. Would I be safe to use 18T Pulleys on an exact c-c distance setup with 15mm belts? I know you definitely shouldn't go that small with 9mm.

3) Should I be using the WCP Belt Calc to get me exact c-c distance or do I need to do some more calculations?

Hopefully you drive train guys out there can give me a hand. Thanks!

You can find the rated torques for HTD belts here. You should be able to calculate the peak torque you could see in your drive train from your robot weight, wheel size, and wheel COF. Bear in mind that these ratings are a bit conservative - since the service life of belts in FRC is so short, you can likely push it a bit.

The WCP belt calc works fine.

[Arpan]

Originally Posted by dellagd

As I'm sure some of you guys are I have been having some fun lately going through CAD'ing different possible drive train designs. I have a few concerns though.



3) Should I be using the WCP Belt Calc to get me exact c-c distance or do I need to do some more calculations?

Hopefully you drive train guys out there can give me a hand. Thanks!

That URL will work to give you a good center-center distance. That said, if I were you I would also implement some sort of tensioner in my drive to keep my belts from slipping.

Personally, I love #35 chain over belts because chain does not have to be as precisely tensioned and most powerful (4 or 6 CIM) FRC drive applications are actually a little out of spec for either HTD 5mm belts or #25 Chain.

I look for robustness first in a drive, and for that reason chain (with a pair of direct driven wheels) is my favorite torque transmission solution.

A good drive weight depends on the type of drive. For the 6-cim monstrosity that 3061 ran last year, 35 lbs was the target weight not including control electronics.

[Oblarg]

Originally Posted by Arpan

Personally, I love #35 chain over belts because chain does not have to be as precisely tensioned and most powerful (4 or 6 CIM) FRC drive applications are actually a little out of spec for either HTD 5mm belts or #25 Chain.

To provide some counterpoint:

I have seen a grant total of one belt failure in my 7 years in FRC (out of 4 years of belt drives), on a belt that was overtensioned, at an offseason event after several competitions. Both 4464 and 449 ran belt-driven 6 CIM drives last year without incident (modified KOP on 4464, 9mm belts on 36-tooth pulleys in a WCD on 449).

I have seen more chain failures than I can count, mostly due to master links or slightly out-of-alignment sprockets.

Thus, I can't honestly say that I wholly buy the robustness argument for chain over belt. As far as failure rates go, my observations are precisely the opposite. Belt is far more forgiving in terms of alignment slop between the pulleys, and if you don't have the machining capabilities to have exact C-C distance you can do fine by having one end of the belt run be adjustable (versablocks are a great way to do this). The lack of stretch means that you need less adjustability than a comparable length of chain, as well.

The only downside I've ever noticed for belts is that they demand that you pick a length and stick with it, so there's little potential for last-minute design changes (unless you purchase a new belt). In a drive, this isn't so much of a problem - frame dimensions are usually decided on very early and seldom changed.

[Jared]

Originally Posted by Oblarg

To provide some counterpoint:

I have seen a grant total of one belt failure in my 7 years in FRC (out of 4 years of belt drives), on a belt that was overtensioned, at an offseason event after several competitions. Both 4464 and 449 ran belt-driven 6 CIM drives last year without incident (modified KOP on 4464, 9mm belts on 36-tooth pulleys in a WCD on 449).

I have seen more chain failures than I can count, mostly due to master links or slightly out-of-alignment sprockets.

Thus, I can't honestly say that I wholly buy the robustness argument for chain over belt. As far as failure rates go, my observations are precisely the opposite. Belt is far more forgiving in terms of alignment slop between the pulleys, and if you don't have the machining capabilities to have exact C-C distance you can do fine by having one end of the belt run be adjustable (versablocks are a great way to do this). The lack of stretch means that you need less adjustability than a comparable length of chain, as well.

The only downside I've ever noticed for belts is that they demand that you pick a length and stick with it, so there's little potential for last-minute design changes (unless you purchase a new belt). In a drive, this isn't so much of a problem - frame dimensions are usually decided on very early and seldom changed.

I don't disagree with anything that you're saying, but I think chain is still an okay choice for drive. Remember, most of the belt drive robots you see are kitbots that are easy to build correctly, and all chain robots you see are custom drives, which are much easier to mess up.

It's possible to be successful with #25, #35, 9mm, and 15mm belts with a 6 CIM drive. Team likes 118 and 254 have been running #25 chain drives for years, and they don't have chain failures.

That said, you do need to be accurate when you're using the chain (especially #25). With all the off the shelf solutions, getting a reliable chain drive shouldn't be impossible.

Also, do you have any data to prove that a 6 CIM drive is out of specification for the #25 chain?

[Chris is me]

Originally Posted by Arpan

Personally, I love #35 chain over belts because chain does not have to be as precisely tensioned and most powerful (4 or 6 CIM) FRC drive applications are actually a little out of spec for either HTD 5mm belts or #25 Chain.

It's really worth noting that specs are pretty conservative numbers based on thousands of hours of runtime. FRC robots run for a tiny fraction of that, hence why hundreds of teams each year run 25 chain or HTD belts without any failures. 35 chain may not be a bad choice, but I think nearly any robotics team who can keep sprockets in line with each other and chain in reasonable tension can handle #25. It's really not as finicky as people make it out to be. If you can mill exact centers, 15mm belts are a good solution; I haven't personally heard of any failures of wide belts yet.

[Oblarg]

Originally Posted by Jared

I don't disagree with anything that you're saying, but I think chain is still an okay choice for drive.

I never meant to imply that it wasn't - only that I have had reliably better experiences with belts.

[Thad House]

Originally Posted by Chris is me

It's really worth noting that specs are pretty conservative numbers based on thousands of hours of runtime. FRC robots run for a tiny fraction of that, hence why hundreds of teams each year run 25 chain or HTD belts without any failures. 35 chain may not be a bad choice, but I think nearly any robotics team who can keep sprockets in line with each other and chain in reasonable tension can handle #25. It's really not as finicky as people make it out to be. If you can mill exact centers, 15mm belts are a good solution; I haven't personally heard of any failures of wide belts yet.

Something else is that wheel sizes, gearing and sprocket/pulley size affects the load number dramatically. Our 2013 robot used 2 CIM's, 6 in wheels, and 22 tooth sprockets with a 12.75:1 reduction. That results is 623lbs of stall force being able to be applied to the chain. Our 2014 robot used 3 CIMs, 4 in wheels and 28 tooth sprockets with a 7.44:1 ratio. This results in 665 lbs of stall force being able to be applied. These numbers are very similar, and shows that forces are based on much more then just # of CIMs.

#25 chain is rated for about 800 lbs, so we were within spec, especially since the wheels slip before you can get close to stalling the motors.

Originally our 2014 robot used 9mm vex belts on 24 tooth pulleys, and we snapped alot of belts because the pulleys were too small and we overloaded the belts.

[dellagd]

Before this turns into a whole belt vs chain debate, I was really looking for a weight target I should be designing for . Right now I'm getting things in the ~45 lb range for an 8WD system. Is that pretty typical?

[Oblarg]

Originally Posted by Thad House

These numbers are very similar, and shows that forces are based on much more then just # of CIMs.

In fact, if you're traction-limited the number of motors don't factor into the maximum belt/chain loading at all.

Originally Posted by dellagd

Before this turns into a whole belt vs chain debate, I was really looking for a weight target I should be designing for . Right now I'm getting things in the ~45 lb range for an 8WD system. Is that pretty typical?

Definitely on the heavier side, but not completely unreasonable.

[Thad House]

Originally Posted by dellagd

Before this turns into a whole belt vs chain debate, I was really looking for a weight target I should be designing for . Right now I'm getting things in the ~45 lb range for an 8WD system. Is that pretty typical?

I'd say thats about right if you go for 3 cims per side. I'd calculate about 15 lbs for just the chassis metal, another 5-8 lbs or so for the wheels and axles, and 15-22 lbs for the gearboxes depending on if you go with 2 or 3 cims. That gives you 35 lbs in the lightest config, and 45 lbs in the heaviest config.

EDIT: Maybe 15 lbs for the metal is too much. The other parts are still correct, but they weight of the metal depends on if you use sheet or box, and if you count supports for bumpers and other structures in that weight. That 15 lbs is about what ours weighs with a box WCD design, and all bumper mounting included.

[Arpan]

Originally Posted by Chris is me

It's really worth noting that specs are pretty conservative numbers based on thousands of hours of runtime. FRC robots run for a tiny fraction of that, hence why hundreds of teams each year run 25 chain or HTD belts without any failures. 35 chain may not be a bad choice, but I think nearly any robotics team who can keep sprockets in line with each other and chain in reasonable tension can handle #25. It's really not as finicky as people make it out to be. If you can mill exact centers, 15mm belts are a good solution; I haven't personally heard of any failures of wide belts yet.

I certainly wouldn't say that either #25 or 15mm belts are bad solutions; 3061 had used both while I was on the team.

That said, I've had much more success running #35 than #25 or belts - every year that we've had belts or #25 we've required milled floating or slot tensioners to keep everything nice and tight.

I would agree that belt failures are very, very rare - I've personally never even heard of a belt snapping. However, belt slippage is something that happens very commonly to younger teams due to a lack of proper tensioning or miscalculated center distances.

Perhaps robust was the wrong word; maybe I was looking for foolproof? IMO anyways, it's much much harder to mess up #35. It could be that with the newer solutions that have come out, like versablocks, belt tensioning is easier than it was in past years.

[Deke]

Originally Posted by dellagd

1) What is a good goal for total weight of the drive train? This would include all frame, wheel, axle, gearing, and motors.

This gets tricky when you combine all of the frame, wheels & axles, gearboxes and motors. Depending on the type of design you go with, each of those can impact the weight in a large way. Some wheels are around 1 pound while others can be .25 pounds. Then you get into gearboxes which can be single or double reduction, single speed/two speed/three speed, 2 cim, 4 cim, 4cim/2min cim, 6 cim, 4cim/4mini cim. Motors weigh a lot (2.8 pounds for one cim), and the amount of motors you put in a drive can change the weight totals quickly.

Generally speaking, drive trains range from 25-40 pounds depending on a team's design.

I would be careful about weight goals with a drive train. When weight becomes a primary factor in drive design, it can lead to other short comings. Of course weight needs to be considered, but it should be low on the list.

[Deke]

Originally Posted by dellagd

Before this turns into a whole belt vs chain debate, I was really looking for a weight target I should be designing for . Right now I'm getting things in the ~45 lb range for an 8WD system. Is that pretty typical?

Ah, I didn't see this. Like I said on my other post, it depends a lot on your configuration choices.

I will echo the other responses, its on the heavy side, but not outrageous.

[dellagd]

Originally Posted by Thad House

I'd say thats about right if you go for 3 cims per side. I'd calculate about 15 lbs for just the chassis metal, another 5-8 lbs or so for the wheels and axles, and 15-22 lbs for the gearboxes depending on if you go with 2 or 3 cims. That gives you 35 lbs in the lightest config, and 45 lbs in the heaviest config.

EDIT: Maybe 15 lbs for the metal is too much. The other parts are still correct, but they weight of the metal depends on if you use sheet or box, and if you count supports for bumpers and other structures in that weight. That 15 lbs is about what ours weighs with a box WCD design, and all bumper mounting included.

Yes, that was including 6 CIM motors. It was designed with rectangular aluminum tubing (we can't do sheet metal yet) and Colson wheels. After entering in all the weights in Inventor I'm getting 41 lbs.

Originally Posted by Infinity2718

I would be careful about weight goals with a drive train. When weight becomes a primary factor in drive design, it can lead to other short comings. Of course weight needs to be considered, but it should be low on the list.

Absolutely. I'm not designing for weight, but I want to make sure I don't spec something that chews up too much of our gaming mechanism budget.