I haven’t uploaded any of my CAD to Chief Delphi in a while, so I figured I’d post this. Recently I’ve been doing a lot of work with replicating West Coast Drives, so I decided to mix it up a little. The 148/217 collaboration really inspired me to learn more about sheetmetal and its fabrication process. This is one of a couple sheetmatal drives I’ve attempted to design.
Specs.
CAD says what’s shown weighs 40 pounds.
2024 aluminum all .09in thick.
Most all flanges are .25in radius
It is riveted together (the holes for the rivets are shown, but the rivets aren’t)
It’s driven by a standard toughbox gearing, I customized the plates.
The middle wheel is direct driven via Hex Shaft, and power is distributed with # 35 chain to the outer wheels, which rest on Dead Axles.
The wheels are Custom 5 inch wheels, they have a leather tread riveted on. The wheels weigh .7 pounds each.
The chains are tensioned by sliding sheetmetal plates. The bolts are loose enough that the plates will slide, but tight enough so that wiggling is minimized ( I may add plastic washers to assist in a smooth slide). The original plan was to have zip ties go around the dead axles and when the chain loosened, one could simply pull on the zip-tie to tighten it. The other option was to attach a spring to the dead axle as well as to the outer frame to apply a constant force in keeping the chain tensioned.
I was under the impression that that temper is what really matters to bending. You want a soft temper to bend ideally. I was planning on using 2024-O which will bend fine. If I were to use 2024-T3 or 2024-T351 (to name common tempers of 2024) I wouldn’t have as much luck. It does say that 2024 welds poorly, which is why I’m riveting the frame.
1/4" Bend Radius? Why so big?
Our sheet metal shop uses the following:
0.125" 5052 Alum = 0.048" Bend Radius
0.090" 5052 Alum = 0.032" Bend Radius
0.063" 5052 Alum = 0.032" Bend Radius
Talk to a local sheet-metal shop and find out what their standards are.
They can probably also give you K-factor or Bend Deduction data for each material. For those that don’t know, these are basically measurements of how much the material will stretch when it is bent and can be programmed into the CAD program so when the material is flattened and exported this is taken into account – but this is all a bit off topic.
For some reason (and I sort of regret it now) I decided it would be a good idea to stay above the recommended bend radii. When I attempt revision 2 I’ll shrink the bend radius down to .188 as recommended. I may also play with making it out of 5052 and using .032 radius bend.
That bend radius table does not remind me at all of what we use at work. I would recommend following JVNs advice and talking to the sheet metal shop you intend on working with. They will have a few parameters that you will have to follow, including minimum flange size (which is a pain sometimes, but hey, part of sheetmetal design).
If you design the sheet metal style properly in inventor, it will calculate the bend allowance and everything else you need automatically. I don’t know how to do this, but I believe there is a tutourial.
You can strengthen the assembly considerably (enough to drop down to 0.062 material) by adding some gusset plates, specifically, where the inner plates meet with the front and rear plates (4 locations)
Triangles (2 or 3" long) over or under the ‘wheel well’ area would be OK, or at the central area if necessary. This will prevent the chassis from racking.
The thinner material would be fine if (and only if) you had bumpers there: Use high-quality plywood and use its strength to your advantage.
If you fear insufficient strength at critical points, like axle mounts, add a metal plate (riveted in place).
Overall an very nice design, looks solid and manufacturable.
I like your choice of 2024, but unless you use a tempered 2024 alloy, 5052-0 will bend easier and be slightly stronger in fatigue, yield, and ultimate strengths compared to 2024-0 (I looked this up on Matweb.com, great resource). Maybe make the bent pieces from thinner 5052 and make your gussets and reinforcement (like Don suggests) out of 2024-T4/T351 to utilize its strength appropriately.
You could definitely use zip-ties to tensions your chain, but a spring would probably be better. A constant-force spring or a gas spring would be great because you could maintain a constant tension on the chain. It might have to be a fairly high load spring depending on your gearing. The gas spring also has the advantage that if anything failed the spring wouldn’t go flying around, they are damped.
Keep rolling with this, it’s looking pretty good so far!
Spring loaded chain tensioners are fairly notorious for not working in more than one direction. There aren’t exactly many successful spring loaded tensioners designs in FRC. Unless you happen to have experience in the area I wouldn’t recommend something as “better” that you haven’t successfully run yourself (and if you have, I’d love to hear about it).
Instead of zip ties or springs, we have used screws to tension dchains like this:
Our axle blocks were made out of 1" wide 1/4" thick aluminum plate, cut to length. We had a hole for the axle (3/8") and a hole for a securing bolt. In the frame, we had two slotted holes, one for the axle bolt and one for the securing bolt. We then made the axle block longer then needed to hold the axle, and drilled/taped a hole down the long end for a 10-24 bolt. When you tighten the bolt, the axle slides and the chain is tight. You could then tighten the securing and axle bolts to make everything nice and tight.
I haven’t done anything like this myself. I was basing my recommendation on this: Invention Awards: Ripsaw Tank Delivers Death at 60MPH | Popular Science which is an incredibly fast tracked vehicle, and it uses a spring/piston to tension the drive tracks, and hasn’t thrown a drive track in 3 years, even with a broken suspension mount. Though perhaps not a great parallel to a FIRST robot.
I would be interested to see these designs that didn’t work well, I bet there were issues where the drive wheel’s tractive forces was also compressing the spring that was tensioning the chain. A workable design might be as simple as an adequately stiffer spring/piston/gas spring. Like I mentioned before, a piston or gas spring with constant force vs deflection (unlike a typical coil spring) would not slacken as long as it’s force rating was not exceeded.
I think it would be a very cool idea to design and implement, but it certainly won’t be as trivial as throwing some springs into your robot. Having thought about it for 10 minutes I think that one could design a cam with a 1-way clutch bearing that could eliminate the need for a spring entirely.
