10-15-2016 07:02 PM

Ari423

Re: pic: SWERRRRVE

Here are a few more renders:

10-15-2016 07:23 PM

SenorZ

Re: pic: SWERRRRVE

Pretty considering the self-imposed limitations.
Revision 1, I'd put the CIMs vertical closer to the center of the triangle. Give some space for frame attachments.

10-15-2016 07:37 PM

asid61

Re: pic: SWERRRRVE

Looks really good, and I love the packaging more than anything. Looks easy to machine, except the the caster box (what is that made from?).
What made you decide to use a plastic bearing instead of a roller thrust bearing?

10-15-2016 08:20 PM

Cothron Theiss

Re: pic: SWERRRRVE

Love the idea of using the clamping gearboxes for the initial reduction. I do have a question. Can you explain why you seem to have an overdrive from the driving sprocket out of the clamping gearbox to the driven sprocket on the wheel?
Also, how thick is the metal on the caster? Tapping threads into the side of sheet metal typically isn't advised.
One last thing. How much of a reduction to you have on rotating the caster? Depending on what motor that is, you may need to have more reduction.

10-15-2016 08:37 PM

carpedav000

Re: pic: SWERRRRVE

Nice swerve! This reminds me, I still haven't finished mine One thing, 44 lbs sounds kinda heavy. It appears as if you could shed some weight by designing it to fit on a single sheet of alum/metal of choice (removing the need for the box tubing in the corner) and/or moving your motors closer to the shaft on the caster.

10-15-2016 08:58 PM

Roboshant

Re: pic: SWERRRRVE

Did you get any inspiration from the 2468 Versaswerve? I know some of their older designs used similar techniques as yours (clamping box for CIMs, etc.). I think their design would work much better for a more lower resource team. Overall pretty cool design though!
More information here: http://www.frc2468.org/resources, https://www.chiefdelphi.com/forums/s...d.php?t=140942, http://www.vexrobotics.com/vexpro/ex...guides/#Guides

10-16-2016 06:16 AM

Ari423

Re: pic: SWERRRRVE

Thanks! Right now the clamping gearboxes are slid as close to the corner pieces as possible. I would love to be able to put them on the corner pieces, but there are gussets on both ends holding it onto the rest of the frame. I could not think of any other way to attach the corner pieces other than welding, which was outside the limitations I was working with.


Thanks, that means a lot coming from you. The caster box is made out of three pieces of thick sheet metal. The top piece is 3/8" and the side plates are 1/4". For the side plates, you only need one bearing hole and one dead-axle hole (I just realized I can simplify the design by flipping the bearing - v2!). The top plate is a little more complicated, you need a bearing hole, six holes for a 1.875" bolt circle, and holes in the sides of the sheet metal to hold the side plates. It still should be doable (carefully) with a manual mill.

I chose to use a plastic bushing instead of a roller thrust bearing mainly because I have worked a good amount with HDPE bushings and have never worked with roller thrust bearings before. From a quick look at what McMaster has to offer, it looks like they would work too. For the speeds this module is spinning at, I have some pretty good real-world tests that say HDPE bushings and bearings work pretty much the same.


Thanks. I was pretty limited on the gear ratios in all the other locations, so any gearing changes are going to come from that chain reduction. In the clamping gearbox, I needed to use a CIM pinion with 84t spacing, so my lowest option was 14:70. For the bevel gears, I could have used a larger 18t bevel gear from McMaster, but that would have made the design bigger and more complex, which I rather not do. For the final spur gear stage, I wanted the wheel to be dead axle so I needed a VersaGear on the bottom (the smallest of which is 48t). I wanted to keep the module short (to minimize torque on the bushing), so I used the smallest driving gear I could (30t). Since all the other stages were decided, I needed to overdrive the chain reduction to get it to the right speed.

I copied the side-drilled sheet metal design from the Revolution Pro Swerve. The top piece of sheet metal is 3/8" with #10 holes drilled and tapped in the side. That leaves ~.1" of wall on either side of the whole. I suppose I could use #8 instead of #10 screws, if that's not enough.

The final reduction on the rotating mechanism is 16.8:1 (7:1 VP and 15:36 chain reduction). With a 775pro, that gives a 90deg rotation in .03s at ~50A. (Using 154lb robot / 4 wheels at 1" arm length in the JVN calc) If that isn't enough, I could easily change the VP reduction to 10:1 or change the driving sprocket from 15t to 12t


You should get on that lol. I looked into using a piece of sheet metal like the Wild Swerve, but I ran into a few obstacles. My team didn't have a waterjet/sheet metal sponsor, so any machining would have to be in house. I didn't think that a single flat piece of sheet metal would be strong enough to support the torsion on the module. There were two solutions I could think of. The first is to use two pieces of sheet metal, one on the top and one on the bottom of the frame. But at that point, it wasn't much lighter than a tube and it was a lot weaker and harder to make. The other option was to use a thicker piece of sheet metal with bends on the sides to provide torsional stability. I would either need to find a wide piece of COTS U bracket (~3" wide) or a brake press to make my own (which I don't have).


I only saw the VersaSwerve after I had almost finished my design, so I didn't get a lot of my ideas from there (I came up with the idea to use the clamping gearbox independently). I think it is a great design, but it is a little more advanced than my team was capable of. It requires a lot of custom and/or modified parts. And I wouldn't trust our mill to make slots in metal. It was really must a drill press with a sliding table.

Really, now that I think about it, my design doesn't even need a mill. It just needs an accurate drill press and printouts of the hole patterns.