OK, so is this because of the force on the corners of the shaft? How do I find out how much torque a 3/8" hex shaft can transfer? We were using 3/8" because we couldn’t get the gears we needed with a 1/2" hex bore. Guess we should have used round bore with keyed shaft?
The whole arm is about 20 lbs, with most at the far end, and is 1.12m long. There is a small (2" dia) sprocket on the bottom (on a 3/8" steel shaft), connected by chain to a 4" dia sprocket at the ‘shoulder’ (on 1/2" aluminum hex shaft, held to the arm with hex hubs at the ends of the shaft bolted to very solid brackets).
We were considering an arm with around 120 ft. lb. of torque at the pivot. The original design had a large sprocket at the centerline of the chassis and live axles going out to the arms. The axles on each side would be around 8" long.
Neither AndyMark nor VEX could give me a torque rating for the 1/2" hex shaft they sell so we decided to do some testing. We clamped an 8" section of 1/2" aluminum hex shaft in a vise. I used a torque wrench with a 1/2" socket to apply torque to the shaft. With the torque wrench set to 50 ft. lb., we got around 15 degrees of twist without the torque wrench clicking. After removing the torque, the shaft straighten itself. A local team was not so lucky in 2015. Their arm shaft ended up permanently twisted.
How many wheels on the robot? How heavy is it? How long is the shaft, and how far out is the wheel? What type of wheels? How many wheels? Does it seem like the issue is just with one wheel, or with all of them?
When you say “weird noises”, can you quantify that any more? What, exactly, are you hearing? Are there any noticeable changes when you rotate the wheels by hand?
Perhaps get a high-quality bearing with a super thick inner ring and replace or sand that ring into a hex (or replace the inner ring with a hex bearing). This is not intended to be a professional solution.
Hex bearings are an odd ball thing that really only sees use in Robotics competitions. They have recently been appearing in light weight combat robotics, but they are generally not used in industrial applications.
If you want truly high load bearings you are going to need to get round bore bearings and make reducer bushings if you have to use 1/2" Hex. Mcmaster has a large selection of bearings with specifications. If you can wait a week or two for delivery, you can go to ebay. If you are really hurting on time and can be a little flexible, tractor supply sells bearings in store.
If you are doing something like a 4 bar lift this year, you will need around 600ft lbs of torque which is going to be a few hundred pounds of tension in a chain. This means you will probably need a bearing with a static load rating of 1000lbs or greater if you want a reasonable level of reliability. This means you are probably going to have to move away from hex shaft and move up to at least a 5/8in ID bearing to get that load capacity.
ok, here is my simple 1 significant figure calclulations. First, some assumptions. A fully laden robot weighs about 150lbs. Second, the height you need to rise is 24 inches.
With a 24 inch rise, you need a minimum lever arm of 24 inches or two feet. 150 lbs on the end of a 2 foot lever arm needs a peak of 300ft lbs of torque to rotate the load. To make sure you are not stalling the motor or pulling a enough current to blow main breakers or burn up motors or controllers a margin of safety is needed. For FRC applications and duty cycles and expected life span I generally use a 2:1
This adds up to needing to reduce a motor or system of motors to where they will output 600ft lbs of torque to safely and reliably move the load. Lets take a reasonable sprocket size to estimate chain tension force. Lets say we use a 4 inch diameter sprocket on an axle to rotate the 4 bar. This gives us a 2 inch radius or 0.17 of a foot. We divide our needed torque by this lever arm and get a 3500lb tension in the chain.
Now all this leaves in a generous margin of safety and ratings on bearings, shafts, chain, sprockets etc are not intended for FRC applications where if a team is extremely lucky their robot will see 100 matches or so. Instead they are rated for industrial applications where breakage can be extremely costly or even life threatening and a service life measured in 10’s of years of 24 hour a day use is not uncommon. We can up rate these components. How much we can increase their rated loads is up for debate, but my gut feeling is we should try for 1000lb or more static load in our bearings. I would have to dive into some more serious calculations and simulations to determine exactly how much.