How long (from axis to rotation to COG) and heavy is the arm? 900:1 seems perhaps a little high to me unless the arm is absolutely massive. We did the calculation for a 15 lbs, 36 inch arm and landed on a ratio of 200:1 by using ambcalc and aiming for a ratio that would be at max efficiency at 12V. It’s possible you’ll need a bigger ratio because we are only aiming for the low goal, but also you can get more power if you sacrifice efficiency.
In our case we achieved a 200:1 reduction with a 5:1 VP, the WCP Rotation Gearbox geared at 58:18 and 52:18 and a 15:64 chain reduction using #35 chain.
As others have said, using 10:1s after the first reduction stage is highly suspect (some teams have sworn off 10:1 gears entirely).
Driving your arm with chain helps in two ways. First, you can get extra reduction by using a bigger driven sprocket, allowing you to remove the highest loaded gear stage. Secondly, my understanding is that chain is a bit stretchy, so it helps account for shock loads (extreme but temporary loads that can occur when moving the arm back and forth or running into objects).
I would think switching to a dead axle setup would have less effect in preventing gears from breaking and more of an advantage of preventing your output shaft from twisting.
We got two of these and have been very happy with them. They are heavy, but ours are mounted very low, with chain running up to the two joints. We also have tensioners on the chain situated so they can help take up shock load. We ordered these early on, when the arm prototypes were really heavy. But we don’t need to use the chain for reduction, which helps keep things compact. Overkill, but would probably stay with this choice even with a lightened arm.
The mounting job is poor, but we wanted a quick proof of concept. Even at 900:1, the arm falls under its own weight, so hopefully the springs/cords will help mitigate that.
The competition robot will have this arm mounted on single cascade stage elevator mounted diagonally on the robot. We’re using a mini-CIM, and the output shaft is 1/2". The arm length is still under discussion, but it should be around 36 inches, and if we can redesign our elevator, we might increase the arm length toward 48 inches. I also don’t know the weight, but will estimate it to be between 10 and 15 pounds.
I can get exact measurements on Saturday when I’m able to be with the team.
It looks like you are using 80-20 for your arm, in addition to all the advise about not using higher than speced ratios on you gearboxes and large pulleys or sprockets on dead shafts, I would highly recommend trying to cut weight. Using 1/16" wall tubing could cut out a lot of weight.
Other ideas include worm gears particularly if you want to avoid back driving. (note that many if not most worm gears are back drivable under the right circumstances, they are just hard to back drive.) You could also try using a winch and rope to pull the arm up and down from the top you tower structure.
(1) use MaxPlanetary or Sport gearboxes. They are substantially tougher.
(2) Your final reduction should be something non-planetary – most teams use chain from a smaller sprocket to a bigger sprocket for this.
Why chain? When two gears connect, you only have a few teeth on each gear where the force is applied. Chain wraps around the sprocket, typically engaging with half of the teeth.
I don’t know why Vex still sells 10:1 VP stages. They’re exclusively used for higher torque applications, which they’re completely inadequate for. I can’t think of a time I used one and didn’t regret it.
All this has been mentioned already, but I’ll re-iterate I guess.
If you’re locked into using VPs, stick to 7:1 or below. If you’re willing to buy some different hardware, Andymark’s 57 Sport gearbox is well proven in FRC. Rev’s Max Planetary is quite beefy too, but they’re out of stock so you’d have to find someone who’s willing to sell or trade. Making the final stage chain on a dead axle means that you don’t have to put your full load torque through a planetary gearbox or a relatively thin hex axle (although you will have to figure out a tensioning solution. With a NEO (or any other “CIM Class” motor) you should only need 300:1 or so for any reasonable length/weight arm.
As has also been said already, making your arm shorter and/or lighter will make everything easier.
Someone probably has, but I expect that they would have regretted it!
Motor selection, for all its mystery and politics, is fairly one-dimensional. You primarily just need to determine how much power you need the mechanism to deliver. In 2023, we are challenged to raise a 1 lb cone, plus our robot’s manipulator, by a few feet, in a few seconds. We can just throw all those numbers together and get an order-of-magnitude estimate of our minimum power requirement :
The 2nd piece of information you’d need is this: Whichever motor we pick, we would like to operate it at say 25-40% of its max power. This ensures that we are in the high-efficiency region and are not dumping an excess of energy into heat that might cook the motor. It also ensures that we have some headroom in case we have accidentally underestimated our power requirement in the previous step.
The snowblower motor is only capable of delivering 20W at max power. Our desired operation point is ~25% of that, or ~5W. This is 10x worse than our estimated minimum power requirement, so it’s not an acceptable choice for this application.
 A better model would take into account efficiency losses, acceleration and deceleration behaviors, the varying angle of arm with respect to gravity, etc.