Team 449 competed in the Battle o’ Baltimore fairly successfully, but in the quarterfinals we snapped a belt, forcing us to compete with three wheels. With mecanum drive. I’m very proud of how well our driver handled the robot in its injured state, but it completely destroyed our hybrid mode which relied on strafing.
My question is, what caused the belt to break? We used PowerGrip GT2 5mm pitch belts from the KOP (Gates Motion), each connecting one AM toughbox to a single 6mm AM mecanum wheel.
The belts endured two regional, including elimination rounds, several community demonstrations, and qualifiers at battle.
We tensioned the belts very well, but no one really knew how much to tension them, so we ended up making them as tight as we could.
I’m really looking for someone who has a lot of experience with belts to explain this. The future of our drive transmissions hangs in the balance.
I am not an expert but our engineers from FMC told us to use chain over belts in the drivetrain. NO EXCEPTIONS. That was just us. This does not mean don’t use belts, but it is a great thing that you guys used belts.
I’d say the most likely cause is a slightly-shorter-version of the phrase “shift happens.” Unexpected conditions can arise. Belts can get unnoticeable torn accidentally when doing work around the belt. Belts suffer from wear and tear.
The lesson learned is to be prepared, bring extra belts, and design to make repairs easier.
I would suggest asking someone on team 125 for help. I know they have experience with belt drives and the tensioning involved. ask Brandon Holley perhaps.
While I am no expert, I did spend a considerable amount of time this past season working with belts. We used .200" pitch, .375" wide kevlar reinforced timing belts from www.sdp-si.com. We finally had a belt go down after the beantown blitz this year (2 regionals and 2 off-season events).
The thing about a belt is even though ours were kevlar reinforced, they are just plastic teeth. Keeping them tight will help a considerable amount, and I am pretty sure you guys did to have the belt(s) last as long as they have.
Fatigue is basically repeated stress over a relatively long period of time. Even though your belts may not have seen any greatly large shock loads, seeing small loads over an entire FIRST season probably fatigued the material in the belt enough that it finally broke.
Congrats on making it this far into a season before having to change them!
So having to replace the belts this often is normal? I would expect the belts to be rated for a lot longer than we used them (I assume they were made for industrial applications)
I haven’t personally looked, but Ben (m3ch4num470r) claims he couldn’t find any specs for the belts. They were the ones that came in the kit of parts.
Changing a belt is not necessarily normal, however in an application like this, there is a decent chance for small defects to add up and create a failure.
The point of my post was to illustrate that even a great design isn’t perfect.
If you went out of your way to lookup and calculate the correct tension specs and you applied them effectively than this is definitely not normal. However I do not find many robotics teams calculating how much tension they are actually putting on the belt. If you did do this, than some more specs on the belt would be nice to start searching for a cause.
To really understand why the belt failed, we’d need a lot more information, and good photos of the broken ends. Incorrect tensioning, both over and under, is my best guess for the root cause of the failure. tdlrali is also correct, the shock loading these belts would see is likely way over the design limit. In an inustrial setting, with no shock loads over the limit and optimal tensioning, a belt like that might last a year of continuous use.
The solution is simple, though: Inspect the belts every 5 or 10 matches and replace what looks worn - or - just replace them every competition.
Don’t forget that chains fail sometimes, too. We use #35 chain in the drive, and while the chain has rarely suffered a physical failure, the sprockets come loose, things (loose carpet strands usually) get wedged between the chain and sprocket and knock the chain off, etc. Of course the nice thing about chain is that it is a bit easier to fix than a belt because you can easily break and join chains and adjust their lengths.
We’ve even had a sprocket break the welds joining it to the hub (in our very first year… and the very first time we got our auto mode running, too!) So chains are not inherently reliable systems… like belts they need to be designed and built correctly to provide reliable service.
FRC is a high performance application… if you build something so strong that it NEVER breaks then you have probably built it too strong and too heavy. Personally I’m willing to sacrifice a bit of weight for reliability in the drive train, but it sounds like you have come very close to optimizing the reliability/weight ratio on your drive system.
Mecanum drive puts added stresses on a belt because the wheels constantly change direction. It’s not the acceleration or speed that’s put on a belt that causes it to fatigue early – usually it’s repeated instantaneous direction changes. If you have robot-centric driving (i.e. the robot goes “forward” relative to its own direction) then the driver has direct control over how much stress it put on the belts. However, with field-centric driving (i.e. “forward” is always the same direction on the field, which is usually away from the driver’s station) the computer calculates the vectors the wheels need and therefore the overall strategy and driving conditions control how much stress is put on the belts.
To help alleviate this in the future, you could make each toughbox transmission slidable so that sliding them with the belt in place will put tension on the belt. This eliminates the need for extra contact on the belt from external tensioners, thus improving the life of the belt. Or you could use a stackerbox setup that does the same thing (note that there have been many improvements made to that first design since its release).
Mechanums, from what I’ve seen, put less stress, not more, on a drive train because they tend to have far less CoF. Skid steer with high traction tires (~1.3 CoF) tends to have far higher shock loads than Mechanums with a CoF (in most directions) less than .8 (consider a 120 lb robot moving full forward and being changed to full reverse without slippage in the wheels).
This is a little advice for tensioning sprocketed belts. When you tension a belt you only want the belt to be taught. It should be a straight line between belt sprockets when under no load and when you push on it with your hands you want to be able to flex it a little. The amount of flex will be determined by the distance between the two belt sprockets. Over tightening it will cause the belt to stretch much faster and eventually lead to a shortened belt length. An under tightened belt will be tossed from sprockets much like chain would. The last thing is to make sure that the two sprockets the belt is running on are inline and parallel. Belts are perfectly feasible for drive applications but you should get all the information on them before using them in application. A good exercise is to set up a test platform using a motor and some belts and play with different tensions and measuring the displacement of the belts and find your most efficient and longest lasting belt tension to increase efficiency and increase your belt life.