Tank Drive Design

I have some questions regarding drive train design using box tubing. Since my team uses 8020 extrusion everywhere, another team member and I are designing an 8020 free tank drive so that we can fit more features under the weight limit. We would greatly appreciate it if anyone could answer the following questions.

  1. While 2x1 1/16" box tubing is much lighter than 8020, it is not as strong in some ways, especially when holes are drilled into it. What techniques do teams use to design rigid drive trains out of box tubing?
  2. Fastening parts to 8020 is very simple. For box, it isn’t always possible to use bolts and nuts because only one side of the plate is exposed. Rivets come with the downside that they are much harder to remove. I have heard some teams use rivnuts which sound very convenient. What do teams think are the best methods for fastening plates to box?
    Thank you very much in advance.

For a drivebase, I wouldn’t use 1/16 wall tubing, I would use .1 or 1/8. For making it rigid, we just used 90 degree gussets and that worked well for us.

For fastening parts, we use rivets to mount to the box tube.


Rivets are plenty easy to remove, you just can’t use that particular rivet again.


Check out VersaFrame. VersaFrame - VEXpro - VEX Robotics
Plenty of teams use 1/16th wall tubing and it works fine for many games. Maybe not for years like 2016 where the drivetrain took a lot of abuse, but this year we used it no problem. You can also add cross support and a belly pan to increase the strength of drive train. Or just use a thicker tube.

I will note that we’ve used 8020 in the past (for elevators, not drivetrains) and I do not recommend it. It is very strong, but is overkill for most applications I can think of in FRC and is ridiculously heavy.

1 Like

1x2 box tube, 1/8" wall, has an MOI of 0.33 in^4 (in the strong direction), and a cross-sectional area of 0.6875 in^2.
1020 80/20 has an MOI of 0.31 in^4, and a cross-sectional area of 0.787 in^2.

So by those numbers, the box tube is slightly stronger (in bending) and lighter. The story is the same for the other axis of bending.

But you probably don’t need that strength. You have more options in box tube to dial in to the strength you need.


To be clear, this is based on the second moment of area, which is different from the mass moment of inertia that’s usually called “MOI”.

1 Like

I’m no expert but i’ll gladly share my team’s experience making custom box tubing West Coast style drive trains. We use the metric equivalent of a 2 x 1 1/16" box tubing and we the help of some gussets and a 1/8" aluminum belly pan we get a pretty rigid drive train (i don’t recall any issues with our drivetrain’s rigidity).
We use our cnc router to drill holes and cut holes in the tubes, We also use the same router to Create ALL of our gussets, than we rivet it all together and it works great for us. if you have access to a cnc router / mill big enough for your profiles i highly recommend this method even though it takes some time to master this method.

1 Like

Does it not need to come off? Rivets. Make sure you’re using enough (2-3 times as many as you would bolts)

Does it need to come off? Well… use a bolt. But you’re right, it isn’t super straightforward… but there are many ways to skin this cat:

  • Create a clearance hole for a sockethead cap screw head inline with the screw clearance hole on the other side. Not super strong, not super nice to access, but simple.
  • Put bolt through piece of tubing, taking care not to overtighten.
  • PEM nuts.
  • Put through bolts from one side of the tubing to the next, placing the bolts as close to the wall of the tubing as possible. This prevents the bolt from crushing the tubing when tightened.
  • Put a scab plate of thicker material (1/8" metal or 1/4" plastic) to help distribute the clamping load to the edges of the box tube and prevent crushing.
  • Put a buffer insert made of light material (wood, solid plastic, or 3D printed plastic) on the inside of the tubing to prevent the bolts from crushing the tubing
  • Clamp on the outside of the tubing, I.E. with clamping bearing blocks.
  • Get really fancy and make a 3D printed insert with threaded inserts, which effectively produces the same effect as rivnuts or PEM nuts.
1 Like

Seems like an appropriate time to bring this up:

1 Like

So did that contest every happen? And how do you remove a pop rivet faster than a bolt?

a) a sharp drill bit and a drill
b) a sharp chisel and a ball peen hammer


Replace “chisel” with “punch”. If you need to chisel you didn’t use the right size of drill bit.

1 Like

8020 is a lot weaker under torsional loads, and much much heavier, + it’s harder to design rigid mechanisms with. Also, plenty of high-level teams have used 1/16" wall tubing for their drive rails in the past (I believe 254 does, among others).

RE: bolts vs rivets, I’d recommend either through bolting, or just using rivets, since rivets are really easy to remove (just drill them out). For a general style of construction, I’d look at using something similar to what 1678 does, where they hole pattern their tubing with a 1" grid, and then line up all of their plates to that hole pattern

1 Like

Thanks everyone! Thaddeus, your suggestions were very helpful.
I think my first question has been answered. In the past I’ve used a couple of those tricks such as a 3D printed plug with nuts in them, but they all have awkward downsides (not working in the middle of a tube, weakening the tube, needing clearance on the other side). I think rivets are the best for our team since the drawbacks are not too bad. Rivnuts are also a potential option, but I they are probably much heavier and require more time to assemble.
As for the question about rigidity, it’s true that 16th box has a much better twisting and bending resistance than 8020. I imagine the most difficult part would be related to how parts are fixed to the box such that they don’t get ripped out or crush the tube.
Now, I’ve got a bunch more questions for you guys! :slight_smile:

  1. Is it enough to have 4 2x1 1/16" tubing held with gussets to make a rigid chassis frame? Is it necessary to add a cross beam or two to help absorb impacts?
  2. What material is used for bellypans? They don’t actually provide structural support, do they?
  3. How do I design a bellypan that is easily accessible? I’ve looked at a bunch of robots that have lots of holes, but do the pit crew need to drill out all those rivets to access the electronics?
    Thanks again for all of your comments!
  1. Yes, if you use the bellypan for taking some of the loads
  2. We use aluminum, nit sure what type. And yes, they do provide some support.
  3. Some teams mount electronics on the underside of their bellypans, and just flip their robot to work on them. We mount ours on top of the bellypan and have it accessible. This year, our indexer was completely removable, so that helped.
1 Like

Yes, but if you’re building a WCD for the first time, I’d highly recommend going with 1/8" wall tubing. 1/16" is a weight optimization one can make but it leaves much, much less room for error. This is less to do with rigidity and more with local loads (crushing the tube, tearing out bearing holes, etc).

Most of the bellypans you see are 1/8" aluminum, but as you can see pocketing these can be very machining intensive. One of my favorite options is solid 1/4"/6mm baltic birch plywood. Some have also used garolite in the past.
Bellypans are absolutely structural; having a continuous piece of material connecting the corners of your frame together stop the whole thing from turning into a parallelogram.

Usually electronics are accessed from the top. It can be a pain if the rest of your mechanisms don’t leave enough space to get your hands to the electronics. Alternatively you can put the bellypan on top of the frame (sometimes called a brainpan) and have the electronics mounted underneath it, facing down. Then you can flip the robot over to access them without mechanisms in the way.


Derailing this topic for a second:

Statements like these are probably true. Definitely sound advice but does anyone have data/analysis/failures related to this? Part of the trouble of course is it’s difficult to quantify what the loads on a frame would be… so even field testing results related to this failure mode in particular would be interesting.

Really it’s about triangulating any point (or “node”, to borrow terminology from trusses) where loads would be recieved; i.e. each wheel position. And the pan doesn’t even need to be a “pan” per se, could just be flat bar that provides the triangulation. Using a flat piece of material gives double duty in that electronics can be mounted easily.

(Actually, 1/16" box has about half the twisting/bending strength/rigidity of a 1020 80/20 profile. It’s just that it’s also less than half the weight.)

Oh, with regards to tear-out, here’s some strategies:

  1. Use more fasteners
  2. Use fasteners that clamp over large surface areas (clamping bearing blocks)
  3. Use scab plates
  4. Use fasteners that have smooth shoulders rather than threads in contact (i.e. rivets or long, shouldered bolts)

Can confirm. I actually did a study on this for work. Don’t have data handy unfortunately. But, if you have CAD with a simulation/FEA package, it’s relatively simple to set up a study for 3’ of tube vs 80-20.

1/8" is stronger AND lighter than 80-20, per said study.

I think they’re referring to using the chisel to shear the head clean off.

Tried it once. Fun and scary. Solid 6/10, would recommend doing when the opportunity presents itself (inaccessible head).