pic: Bit Buckets 2015 Offseason Drivetrain

Here’s a render of the drivetrain the Bit Buckets are designing and building as an off-season project.

It’s still very much a work in progress (not all joints are drawn, electronics mounting needs work, we still need to add some clearance and lightening holes), but I’m sharing what we have now to see if the Chief Delphi community has any suggestions for improvements.

Some information about the current iteration:
-All structural parts are .25" laser-cut plywood, assembled with 8-32 hardware
-27.5" x 27.5" square frame, top plate is 5" from the ground
-Weighs about 47 lb with everything shown above except the battery
-“Drive pods”, containing wheels, axles and belts, are removable from chassis for easy maintenance (the outside plate of the drive pod is hidden in this render)
-Custom two-stage 3-CIM gearbox with the second reduction stage inside the drive pod for a total ratio of 6.4:1 and an adjusted top speed of about 11 fps
-Elevated electronics board facilitates access to wiring

I’ll post some different views later, as some parts are hard to see in the above image.

Looks great! 47lbs is a very good weight with electronics on there. Electronics all seemt of it well, but you have some wasted space underneath that top plate.

I noticed that your hex axles have a deep groove cut right before the end. This will weaken the axle by a lot. Why is it there?
Pulleys also seema bit large… the belts won’t intersect with anything, right?

EDIT: Out of curiosity, how does this go together? Do you make puzzle-piece edges and glue them? How do you get machine screws into joints?

If you’re talking about what I think you are, those are probably 1/4" hex head bolts going through tube shafts. My guess is he hid the outer drive plate so we could see inside.

We’ll put more of the pneumatic components under the top plate, as well as some of the electronics you don’t need to get to as frequently, like the bridge power supply.

The pulleys are 42t, the same size as on the AM14U. We checked that the belts clear the belly pan.

Ryan Dognaux’s explanation of the axles in the post below yours is accurate.

It’s all put together with mortise and tenon joints and captive nuts, as described in this (excellent) Instructables page.

More renders, as promised:

The drive pod, showing most of the powertrain, as well as some example T-nut joints:

The gearbox, which needs a little more work on mounting holes, as well as an encoder mount (Anyone have ideas on how to attach an encoder? Could it be attached directly to a CIM shaft? The gearbox’s output shaft remains in the drive module when the gearbox is removed, so the encoder can’t be mounted on that.):

This might help with the encoder issue.http://www.chiefdelphi.com/forums/showthread.php?threadid=131034

Thanks for the idea. Here’s the gearbox with that sort of encoder mount:

Since we don’t have much experience milling complex parts, I’d like to 3D print the encoder mount. I think it would be structurally sound; others have printed much more complex motor mounts.

The encoder mount doesn’t need to be round. You can bandsaw out some rought stock, square it up, take a couple cuts and drill a couple holes and be done. Bandsaw after to get it round. You could put it on a lathe if you wanted to, but we did something similar for our offseason project and I would not do it again. Could be good if you don’t have a 3D printer.

We cut about half of a drivetrain today as a prototype.

It generally goes together well for a first prototype, though we found some bugs we want to fix before building a full chassis (including a misalignment in the laser cutter that led to beveled edges).

It looks like it is going good. I am interested in seeing how this turns out because we have gained access to a similar machine. It is a 4’X8’ Shopbot. And being able to confidently believe a drive train out of wood will hold up would be really nice to know.

I do have a few questions

Where did you get the wood/what product is it?

Do you plan to use this for FRC next year?

How do you plan to test it? How will you know it won’t break?

The wood is 1/4" 5-ply Baltic birch from Woodworkers Source. They sell 5x5 foot sheets, which we cut into sixths for the 29x17" laser cutter.

We do want to use this as our competition bot design (more or less), if it works. We plan on building a few iterations now to get the hang of it. We want to make a practice robot that we can bang up to get endurance information.

Since we cut the prototype last week, I’ve been hard at work improving the design. We’ll be fabricating this iteration tomorrow.

Full size here!

Some information and a partial change log:

-Weighs 36.7 lbs with everything shown except bumpers and battery
-Contains 34 plywood parts, which weigh a total of 15.9 lbs
-Gearbox removable from top or bottom with 4 bolts. Drive pod removable from side with 5 bolts
-Improved gearbox with stronger design, easier mounting, more choice in ratios and a slightly increased top speed of ~12fps adjusted
-Modified encoder mount for 3D printing
-Revised powertrain to use wheels and pulleys from AM14U so we don’t have to order more parts
-Designed sturdy and easy to change bumpers
-Reinforced battery mount
-Added removable center belly pan section that allows access to the middle box
-Revised joint layout for improved strength and easier design changes
-Added lightening holes in larger wood parts

We plan on sharing photos and the latest version of the CAD files tomorrow afternoon and our engineering journal at the completion of the project.

Any questions or ideas?

Can you swtich to COTS shifting gearboxes? A good portion of the advantage of a six cim drive is lost when using single speed, because going at any speed where the acceleration difference (between 4 and six cims) is not negligible is just asking for breaker blows.

On another note, that’s a really good weight and design. The way you are fitting wodden plates together is very novel.
How do you make the correct tolerances on everything?

By changing the hole pattern on the inside chassis plate and possibly the size of the clearance holes in the belly pan and top plate, we can switch to 3 CIM ball shifters. For this version, we’re using single speed gearboxes because we already had most of the parts on hand and wanted to test making wood gearbox plates.

Tolerances are largely determined by trial and error. Since laser-cut wood is so fast and cheap to make, we can make multiple iterations of a part in a day. We’re cutting the slots and tabs to nominal dimensions; the laser kerf makes them a tight, but still easy-to-assemble fit. For bearing hole sizes, we’re cutting a test part today with holes of slightly different sizes to see which works the best.

We built it today!

A photo, as promised:
Anyone want some small wood disks?
We’ve been generally satisfied with our results. The chassis is lightweight and extremely strong and stiff; it survived being jumped on, dropped and pounded with no damage. In total, it took about 4 hours of laser time. Hopefully, we’ll get it running over the next couple of weeks, as we replace a few parts on which we made mistakes and receive the stuff we ordered from Vex.

However, I’ll wait a little on releasing the CAD; we caught a few bugs I’d still like to fix.

Edit: It’s also very inexpensive to produce; each chassis takes about $30 of wood and $10 of hardware, not including powertrain components, compared to $189 for an AM14U frame or other COTS or custom options.

What is the cutting area of the laser cutter?

If you do something similar in the future, cut the following letters into each disk:

T, U, I, and T.

Add your team number, and use as giveaways. When you finally get “a round TUIT”…:stuck_out_tongue: :smiley:


The Trotec Speedy 300 at our build space, Xerocraft Hackerspace, has a 29" x 17" bed. We made the top and bottom in 3 pieces to let us make a 28" square robot.

throws 1.78 lbs of wooden circles


Another meeting tomorrow, another iteration ready to fab:

This version includes a significant redesign of the gearbox and its mount, allowing for a stiffer gearbox and more sensible encoder placement. New t-slot design should better retain nuts. The battery mount was enlarged and holes for a velcro strap were added. Small improvements to clearances, press fits and lightening holes were made throughout the model.

This will be the last iteration of the drivetrain before build season. We received our Vex Pro and FIRST Choice orders, so will be assembling the full powertrain and wiring all the electronics on this version.

Here’s all of the wooden parts, laid out for the laser:

We tested our last-week’s robot base to destruction with a car.

The next couple of drive-overs resulted in a flat tire on the Volvo.

This week’s robot is closer to working with real wheels and motors and electronics!