Belts in a Drivetrain

So I have been working on a WCD style drive train, and I was wondering what peoples experiences with belts have been. I have been looking at HTD timing belts, and from the reading I have done, some people have had issues with the 9mm wide belts. I was wondering if this was a common issue or the result of improper tensioning. Are 9mm belts sufficient in an FRC drivetrain or do I need to look at 15mm wide belts?

Our team has never built a drivetrain with belts before so I want to do as much research as I can before we do this in the offseason. If you have any other tips you learned from your experiences, please share them as they will be much appreciated.

Thanks

1 Like

We’ve used belts a few times and we’ve had no issues (apart from actually sourcing the parts one time). Some of the key things to remember are, of course, to tension properly, but also make sure that there is nothing caught in the sprockets. This past competition (district event) we had some tape and cables get caught in the sprockets for our drive train and it was not driving properly.

Kitbot has a really nice sprocket system built into it (well, the holes at least), and we’ve not had any issues with tensioning. If that’s an option for you you may want to consider it.

We run 9 mm belts on our 6 CIM 6WD Versa-chassis. We often play a strong defensive position and so far they have held up reasonably well. We’ve had a few break after hard pushing matches but we believe it was because of them going out of tension. If our center wheel wasn’t direct driven we’d probably be more comfortable running 25 chain or a 15 mm belt that can better handle being out of tension.

We used 9mm belts this season with 30 tooth pulleys. Exact Center-Center distance, we added in some delrin blocks because of some rubbing issues on our bellypan but they weren’t needed at all for tension. We’re going to try 24 tooth pulleys next, as it will fix the rubbing.

Here’s a Google Doc you may find interesting.





That is a pretty awesome spreadsheet! It looks to me like proper tensioning and/or proper C-C design is more important than belt width. On a side note, where do people buy colson wheels? I have looked online and found a few combat robotics websites that sell them, as well as regional distributors. Belts and colsons seem to go hand in hand for some teams and I’m curious.

9 or 15mm belts should do it - 15mm is the ‘safer’ bet.

You can get colsons from various robotics sites or McMaster. McMaster is, of course, ridiculously overpriced though. When we bought ours for this season, they were out of stock and I found them on some sketchy website. They came with bearings, but we just popped them out.

We went with 9mm belts and 24t pulleys tensioned by cams on VersaBlocks. It’s amazing, we’ve never had problems with ratcheting, and the drivetrain is noticeably more efficient. If you tension properly, 9mm shouldn’t be a problem (given reasonably sized pulleys).

Yep, according to the google doc I listed above (I believe I got that through Picone, so credit there), I wouldn’t go under 24t.

Could you put your information in the spreadsheet? With the new versachassis + belts + pulleys from VEXpro your specific setup will become more common, and your data point would be useful.

So how do you know what the correct tension is for a belt? I have always just gone by feel but wouldnt an over tensioned belt be bad?

I know there are official force measurements on Gates’ website but is there an easier way to measure this like over a certain length it should deflect this much with this much force?

Done. Thanks for reminding me, I’ve found looking at that spreadsheet very useful in the past.

This year we went with a 6 CIM 2 speed 6 wheel drive using colsons and geared with a high speed of ~15.5fps and low of ~5.5fps. We used 15mm belts and pulleys with nearly everything purchased from VexPro.

We were the single strongest drivetrain in terms of pushing power of any other robot in both Dallas and Oklahoma. We faced a self-proclaimed “unpushable robot” and bulldozed them from one side of the field to the other.

While all of the above is true, and the belts did well, I personally would rather recommend using chain for the following reasons:

-Breaking:
At Oklahoma (2nd regional), we broke 3 different belts. I think (I am not 100%) that they were a set used either partway in Dallas, or on our practice bot, so they were not brand new. The cause of the breaking is known. In Dallas and on our practice bot, we used encoders as servos (Im not a coder dont get mad at me). The class in the code also contained a system that would quickly ramp up the power from 0 to 100%, even if full yoke was suddenly applied. In early Oklahoma, one of our encoders broke (the mechanical linkage broke), so the servo class was temporarily removed, and thus the power ramp function. With 3 CIMS geared to 5.5fps, there is an utterly massive amount of torque. Without the power ramp, going from 100 to -100 instantaneously is little problem for the CIMs, but a major problem for the momentum, so somewhere something had to give. The belts did. I would expect the larger gauge of chain would be capable of handling that torque, even if it meant we were spinning the tires.

-Replacing:
In many cases, replacing a chain is much easier than a belt. In our drive you had to remove both pulleys to slip a new belt on, which took a fair amount of time. With chain you can replace one with everything else still fixed. If you break a belt, you throw it away. You break a chain, it is usually repairable.

-Buying:
On our robot, each pair of belts (there were 3 on each side) was a different length. This was mostly due to the fact that with our compact robot, the gearboxes had to go where there was room for everything else. We bought 4 pair of belts (8 of each size). The problem was, when we broke belts, we kept breaking the same one (even with our redundant system: if any one breaks, at least 1 wheel would still turn). When we ran out of that one kind of belt, there were no more, even though we had 2 extra of a different belt. With chain, you just chop it to length. Bring 20’ of chain and you could replace each loop once, or one single loop 10 times, depending on your needs.

-Space:
We needed 15mm belts for our drive, as already stated we still snapped them. 9mm would have been worse. ANSI 35 chain is 0.1" thinner than 15mm, and both require slight clearances (guides on pulleys!). While you don’t save much, it does add up. Using ANSI 25 chain like we did last year (with no breakages on our 4 CIM 2 Speed 6wd robot) you save 1/8" vs 15mm

-Weight:
While belts are lighter than chain, I think it is not light enough to be worth it.

Overall I was satisfied with the belts this year, and they definitely do work, but as much as I hate to say it, I prefer chain much, much more.

Added our experience with Versa blocks. We like them. Easy to tension. Sprocket centers just where the calculator said they should be. Two speed versa gearbox with 36/30 belt drive to the center wheel which drives the outboard wheels. Outboard wheels are driven by 30 tooth pulleys Started with 9mm belts all around. We changed the primary drive belt to 15mm because the 9mm belts were breaking. We have left the outboard drive 9mm with slightly less tension that allows them to ratchet rather than break.

We are using Cheesy Poof style control which seems to help with ramping the violent direction changes.

If you have the manufacturing capability, there is no reason not to use exact centers for belt drivetrains. Absolutely none. It’s effective, idiot proof, and reliable. It’s not that hard to under tension or over tension a belt, which leads to reduced belt life. I just don’t understand how people can prefer tensioners for a system whose tension never needs to be adjusted (in the context of an FRC robot life span).

15mm is far safer than 9mm for a drivetrain. We had poor experiences with 9mm belts and 18T pulleys our first year using the system. Since we’ve gone to 15mm the problems have gone away for our particular wheel size and pulley tooth counts. This year we used 15mm belts and 24T pulley stock with zero problems.

The bigger the pulley tooth count, the more you can get away with using a 9mm belt in the drive. I’m not entirely convinced yet that 9mm 24T belt drives are a reliable way to do a “West Coast” style drive. Bigger tooth counts should be fine. Any discussion of the strength and validity of belts must include the pulley tooth count or else the anecdote isn’t useful.

One thing belts can do in FRC is pick up tape. We found this wrapped around the drive pulley. Field admin didn’t seem to want it back. :rolleyes:





987 has ran 9mm-wide 5mm HTD belts for the last 5 or so years. Our outside wheels are lifted up slightly so we have a center drop, and the center wheel is driven by the gearbox. There is a 20mm wide pulley on the center wheel axle that has two 9mm belts going to 9mm wide pulleys on the outside wheel axles. I’m rebuilding this years assembly right now so I don’t have pulleys in this photo, but you can see how we keep our belts tensioned properly.

http://i.imgur.com/k3bs57U.png (attached the file for anyone who cant access imgur at work/school)

There was a thread a few days ago that was very similar, and a lot of teams seemed to avoid tensioning their belts like this. I believe one of the specific arguments against it was that you were now relying on the static friction of the bolts on the rails to hold your bearing blocks in place and keep the belts tensioned, but we’ve never seen them come loose (although it is one of the things on our prematch checklist). A little bit of locktite and a washer on each bolt also helps immensely. :smiley:





Agreed, thought I’d go one step further and say it also needs to include wheel size. What we’re really discussing is how much torque can a given belt/pulley setup transmit. If we assume robot weight and wheel CoF are constant, then a pulley on a 6" wheel needs to transmit 50% more torque than a 4" wheel to put an equal force vector to the ground.

This is 3476’s first year using belts on a drivetrain. We are a 6wd “wide” configuration using 4" AM performance wheels (w/ nitrile tread) and 3/32 center drop. We are using VexPro HTD 15mm belts, Ball Shifters with 4 CIMs (plus some runtime with 2 550s) and 24tooth VexPro pulleys on all axles. Geared for 4/22 fps. I do admit our shafts are a bit overbuilt using 5/8 7075 aluminum stock through round bearings (hex bearings…don’t get me started on those). Machined to exact C-C + 0.003.

We have driven this thing through some very tough defense at SDR and many nights of drive practice, and have run it at very high speeds with very quick stops. Not a single ratchet or failure (with everything together, that is :stuck_out_tongue: ), and it would be far from the truth to say that we have perfect manufacturing capabilities.

We have still quite a few official and unofficial events until the season is over so we won’t draw any conclusions yet…but with the way we drive and the nature of this year’s game I am pretty happy with our decisions so far.

Good luck!

I believe one benefit of tensioners is that, because of manufacturing tolerances in the belts, exact centers won’t always provide the same tension. For this reason, if you’re using 9mm belt I think you should have tensioners to dial in the exact tension you need without relying on perfectly accurate belt lengths

That said, we use fixed centers but add a bit to the distance depending on how the belts fit. We’ve been very successful the past two years using 15mm wide belt on 22t pulleys inside our DT tubes driving 3.5in wheels. We have 1/2in hex in the pulleys in the tube, 1/2 in round bearings, and 7/16 hex in the wheels. It’s a set up I hope stays around for a while.

This seemed to be a common problem Friday at Arizona. The field crew was repairing the field with short pieces of tape, which the Kit drivetrain likes to suck off the floor, into the belt. The main cause seems to be that the pulleys and belts are about half an inch off the floor so any loose corner will be caught and pulled up. A mentor on another team brought this to the attention of the head referee, and it appears that on Saturday they repaired the field with long pieces of tape, instead of trying to fix it with short pieces.