Octagonal WCD

[craigboez]

Given the benefits of having an octagonal chassis (covered in other threads), I've been working on trying to incorporate this general shape while keeping our same WCD-style construction methods of 1x2 tubing and gussets. Below are two preliminary concepts.

In the first two pictures, the "traditional" method of running chains between the gearbox and rails is used. Because the outer wheels are now on the inside of the frame, this pushes them further from the 1x2 tube and I'm a bit worried about the loads on those axles.

In the second two pictures, the chains are run inside the tubes using the 221 Robotics double sprocket method. This helps keep the outer wheels close to the tube, but is a solution we've never used before.

These are still preliminary concepts so the models aren't perfect and are missing some pieces, but just intended to get the point across. Please ask questions and submit comments, as I'm hoping to incorporate some peer review feedback in the next iteration of the design.

[Ty Tremblay]

I wouldn't worry too much about the the cantilever on the front and back wheels. My calculations put 7075 hex shafts at more than strong enough to handle the loads you'd typically see on a corner wheel. Running your chain outside your rails also makes maintenance easier.

[tim-tim]

Originally Posted by Ty Tremblay

Running your chain outside your rails also makes maintenance easier.

If setup properly, the chain-in-tube can really be viewed as maintenance free. From conversations with 118 who ran this setup in 2013, once the chain was on the sprockets in the tube, they never did any real maintenance.

My memory is a bit foggy, so I don't recall the exact spacing, but with a 17 tooth sprocket, #25 chain, and 1"x2"x.125" thick box tubing there is very little clearance between the chain and the inside of the extrusion. This greatly reduces the risk of the chain jumping off the sprocket.

[craigboez]

Originally Posted by Ty Tremblay

My calculations put 7075 hex shafts at more than strong enough to handle the loads you'd typically see on a corner wheel.

Care to share your calculations? And how do you determine the magnitude of the load you'll expect to see at a corner wheel?

[craigboez]

Another aspect of this design that is still TBD is the way the bumpers integrate. I'd like to incorporate a latch like 971 and 254 and Spectrum have shown. The Skunks posted another variation of this a few days ago using a different type of latch. Making a solid connection on the "wing" area is my biggest concern. Any good ideas on this?

[Ty Tremblay]

Originally Posted by craigboez

Care to share your calculations? And how do you determine the magnitude of the load you'll expect to see at a corner wheel?

It was quick and dirty, but here's what I did:

A worst case scenario of running into a robot at full speed with a high COG could lift the back 4 wheels off the ground. I chose this as starting point. Assuming a full weight robot with bumpers and a battery (~150lbs), that means we've got about 75lbs on each of the two wheels still contacting the ground.

I happen to be designing a very similar drivetrain, so I took my most cantilevered wheel and did my analysis on that. My colsons are cantilevered such that there's a 1 inch gap. I then threw all of this into SolidWorks SimulationExpress and came up with a minimum factor of safety of 3.074 right next to the bearing where you'd expect them the shaft to shear.

The above was done with 7075-T6 Aluminum. If a FoS of 3 still leaves you uncomfortable, you could easily go with steel shafts and never have to worry about them breaking.

Originally Posted by craigboez

Another aspect of this design that is still TBD is the way the bumpers integrate. I'd like to incorporate a latch like 971 and 254 and Spectrum have shown. The Skunks posted another variation of this a few days ago using a different type of latch. Making a solid connection on the "wing" area is my biggest concern. Any good ideas on this?

I followed the 971 method of bumper attachment, though it's not 100% fleshed out yet. The important part is to have a well constructed, single piece bumper. Here's what I have so far.

[Jared]

Originally Posted by Ty Tremblay

It was quick and dirty, but here's what I did:

A worst case scenario of running into a robot at full speed with a high COG could lift the back 4 wheels off the ground. I chose this as starting point. Assuming a full weight robot with bumpers and a battery (~150lbs), that means we've got about 75lbs on each of the two wheels still contacting the ground.
[/url]

75 lbs is a very conservative worst case. If I were doing a quick simulation, I would put something like 300 pounds and aim for a factor of safety of around 2. Driving over a bump like 2012/2010 may see the entire weight of the robot drop on one wheel. You may see more than 150 lbs of load if your robot falls or roughly tips down onto that wheel. In 2013, we bent an axle when we smashed a wheel into the pyramid corner. I'd also be worried about the hex bearing's inner race in this situation.

SW simulation express isn't the right tool for this. IIRC, it doesn't support bearing fixtures, so you can't simulate torque on the shaft and the load of the robot. Also, it doesn't take into account the effect of strain rate (how quickly the part is being loaded), which will reduce the tensile strength of the material significantly.

If you're concerned that your cantilevered wheels are too weak, you can just add an additional plate on the inside with another bearing. It's likely worth the extra .5 pounds, rather than having the panic of replacing a wheel/bearing between elimination rounds.

[Ty Tremblay]

Originally Posted by Jared

75 lbs is a very conservative worst case.

SW simulation express isn't the right tool for this.

If you're concerned that your cantilevered wheels are too weak, you can just add an additional plate on the inside with another bearing. It's likely worth the extra .5 pounds, rather than having the panic of replacing a wheel/bearing between elimination rounds.

Thanks Jared. I agree with your points 100%. A second plate wouldn't certainly help.

[Joey Milia]

I would take a look at the 148 write up for their X009 base. Personally for a WCD hexagon or octagon base I like the wheels all on the outside. You wouldn't have to change as much and worry about the wheels on the inside.

[Jared]

Originally Posted by Ty Tremblay

Thanks Jared. I agree with your points 100%. A second plate wouldn't certainly help.

Sorry- I totally misread/spaced out when reading your last post-what you said earlier was fine. For some reason, I thought you said it had a FoS of 1 instead of 3, which is why I made the suggestions.

[asid61]

That's a neat way to incorporate the hexagonal design into a traditional WCD! All it would take is a plate and a piece of 2x1 for each corner.

[craigboez]

While not complete, I'm posting the CAD of this in case anybody would like to take a look. We'll likely develop this in to our 2015 drivetrain, but for now it remains a partially developed concept.

We have built 3 robots using the Vex 2-CIM ball shifters and overall we've been happy with them. We have used a method of attaching them to the rails very similar to what is in this CAD model and it has been rock solid. It requires some modifications to the way the gearboxes are assembled, in particular replacing a few of the hex standoffs and bolts with longer versions and adding a few nuts. This season we may migrate to a 3-CIM setup if we think it is worthwhile.

Caveats: the shafts are not included, the bumper attachment is still a work in progress, lots of fasteners are missing, and in general it lacks finish work, but I think it is a pretty decent start.



CAD here: 2811 Octagonal WCD.zip