Diagnosing Belt Skip Under Load

[Monochron]

We are planning to increase our driving and driven pulley size and also increase their tension slightly based on gpetilli's recommendation. I would also like to look into better programmatic control but so far we haven't gotten a solution that we are happy with.

We need PID control so that we can hold the arm in position to intake balls, but a ramp function tends to cause the I term to accumulate too much and we end up with massive oscillation. We also looked for PID values that would reduce acceleration/deceleration but we weren't able to eliminate steady state error.

Does anyone have a good recommendation for a way to maintain relatively accurate control over the arm but reduce acceleration and deceleration forces? Motion profiling is certainly a possibility, though tuning it without a practice bot is going to be a big challenge for us.

[philso]

It looks like the hex shaft that your arm pivots on is supported by one of the green plates on each side. Is it possible that those plates are flexing? It is not clear from the angle that the video was shot at if there is room to add the brace that Kevin suggested. There appear to be some spacers between the sprocket on the hex shaft and the green plates on each side but I could get a good view of the gearbox end.

We are fighting a similar issue but in our case, the torque on the arm is probably causing the center-to-center distance of our gears to increase, leading to the gears skipping. We have our VP mounted on a rather thin plate that is only supported on one side allowing it to flex under load. We may end up using the same solution that Kevin suggested for you.

[MrForbes]

Providing a few clear shots of the mechanism from a few angles, and a couple overall shots, might help folks help you figure out if there are some mechanical fixes you could make.

I tried watching the video but didn't get too far, the camera was moving to fast for me to follow.

[chapman1]

Agreed with all above who posit that there are not enough teeth in contact on your pinion (smaller pulley).

It's just asking too much of those few teeth. If you could get some slack in the belt and add an idler close to the pinion for more wrap, I think your problem would be solved

[Monochron]

If we decided to switch to #35 chain, how does a pinion sprocket of 15 teeth sound? We don't have a lot of room with the VersaPlanetary right next to the output, and are currently restricted by needing a hex shaft for the pinion, but 15 tooth sprockets are available.

As my mark was off for how many belt teeth would be needed, I wanted to get some opinions on an appropriate number before committing.

[holygrail]

We have almost the exact same setup and we were slipping belts too. We were able to fix it by doing a few things. We haven't slipped a tooth since.

1) We added a 3rd pulley. We needed more teeth in contact with those small 18 tooth pulleys
2) Added shaft support in the center of the driven shaft. It helps minimize any potential bending of the shaft which can change the tension.
3) Improved belt tensioning. We used a Turnbull from the hardware store and it worked great.

[Chris is me]

Originally Posted by Monochron

If we decided to switch to #35 chain, how does a pinion sprocket of 15 teeth sound? We don't have a lot of room with the VersaPlanetary right next to the output, and are currently restricted by needing a hex shaft for the pinion, but 15 tooth sprockets are available.

As my mark was off for how many belt teeth would be needed, I wanted to get some opinions on an appropriate number before committing.

My unqualified, gut opinion not based on any calculations - you should be fine. 35 chain is extremely strong and forgiving. Properly tensioned, it is quite hard to break 35 chain anywhere on an FRC robot. Much, much harder to size incorrectly than belts.

[Monochron]

Originally Posted by Chris is me

My unqualified, gut opinion not based on any calculations - you should be fine. 35 chain is extremely strong and forgiving. Properly tensioned, it is quite hard to break 35 chain anywhere on an FRC robot. Much, much harder to size incorrectly than belts.

I am actually more concerned about skipping chain than breaking it. Last year we had skipping problems as a result of overloading and it took us days to properly diagnose and fix it.