pic: Team 5413 Stellar Robotics Offseason Drivetrain



This is the mechanical part of Stellar Robotics’s 2015 offseason drivetrain.
The frame is laser cut from 1/16th 5052 Aluminum sheet and bent on a CNC press brake. The dimensions are 31 inches long and 27 inches wide.
Uses 6, 4 inch Colson wheels. The center wheels are direct driven from the gearbox and the corner wheels are driven with 25 chain between 32 tooth sprockets. The center wheels have a .150 inch drop.
The gearboxes are custom with 12 tooth CIM pinions driving a 60 tooth gear for a 5:1 reduction and a nominal top speed of 14.79 feet per second.
The whole thing weighs 22-24 pounds according to CAD (we still need to weigh it).

Good work! This looks great!

Have you considered adding a third CIM to the gearboxes? Seems like it could be modified pretty easily. Also, are you doing anything to integrate a sensor like an encoder into the drive train?

Looks really clean good work. I also have a few questions about it.

Is the angle at the bottom of the base when looking at it from a side-view there for a reason? (i.e. going over bumps) If there is no reason Other than aesthetics your frames strength can be increased by extending the flanges out to make them square with little weight added. Right now the bottom plane of the base looks like its only able to get one fastener in each corner. That means that that whole plain wants to fold in on itself if hit correctly. On the top you have the ability to put more than one fastener in which stops this folding motion from happening. I am not saying it will fail, in fact it would probably be fine, but it would make it stronger if you extended them out to complete the box.

If it is there for a reason, there are a few other things you can do that would make it stronger as well. If you added a belly pan to the bottom plane of the base it would perform two functions that i think would improve the overall design. Its nice because it would add a place for you to mount electronics too as well as act as a sheer plate and prevent that folding from wanting to occur.

Just some things to consider. Good work last season! We had fun playing against you guys at buckeye. Hopefully some day we can end up on the same side of the glass. You were excellent especially considering it was your rookie year. Keep it up!

I really admire the minimalist simplicity of this design. Exactly everything a drive base needs and nothing more - nothing overly flashy.

I think your drop is a little severe especially for a 31x27 chassis. The .125" drop-center standard was crafted in the 38x28 era, and on my last robot of similar dimensions with Colson wheels I thought it was just a bit too much rock. 1/16" or even zero drop would probably work well. I don’t think it’s really going to hurt your performance one way or another though.

Do you have a system to tension the chain when / if it stretches?

The integrated gearbox is great for reducing part count but not so great for serviceability and relies on your sheet metal shop’s hole tolerances being as good as your milled plate tolerances. You also have less ability to redesign if something goes wrong. I would switch to a 2-plate gearbox just out of paranoia more than anything else.

Any plans for a belly pan? You could really use one for rigidity? You have very little cross bracing.

Another question did you laser and stack that 60t gear? I would personally buy a vexpro al gear since it’s 7075 AL and the tooth profile will last much longer.

If you’d really like to keep it lasered I’d switch to 6061 and make it .5 to .75 thick aka match the full width of a vp 12t.

While it’s not a real answer, you can see that there are four screw heads around each wheel shaft. This implies that there are actual bearing blocks on the other side of each plate, which may be adjustable.

Every bearing is (or should be) mounted in a block, not a plate. There is no need for a bearing where the main drive shaft passes through the inner plate, as it is supported by a bearing in the motor mount plate at one end and a bearing block screwed to the outer plate at the other end.

At first glance, I thought that you’d built an AM14U, then saw that the top folds were more like the U2, then noticed the gearbox and chain mounted inboard of the wheels rather than belts on both sides of the wheels, then the bearing block mounts. If you’re ready to run a game with a single-speed 5:1 gearbox, this seems to combine the rugged battle tested design of the recent kit with some optimizations to reduce the footprint consumed by the belts/chains, and move the wheels farther out. I also like the complete “skateboard” capability that the recent kitbot lacks - if you put the control system (and perhaps battery) inside the volume defined by the inner rails, you have a level surface upon which to build your manipulators. By perforating every 1", you have plenty of mount points as you make adjustments. It’s also a natural to mount a VersaFrame superstructure.

I would also recommend a belly pan, both for torsion control and for mounting your controls. Considering the belly pan makes me wonder - why are the bottom folds on the inner plate outboard towards the wheels, rather than inboard? If you put them inboard, you could mount your bellypan/control panel there. Likewise, if your top folds bent outboard rather than inboard, they would tend to act as chain guards rather than restricting access to the volume between the inner plates where (if you put your control system low), you are going to want to go. Or perhaps you are planning to mount the control panel/torsion control panel to the TOP of this chassis, and accept that when you need to work on the control system, you’re going to lay the robot on a side/end/top so that you can access it from the side that normally faces the carpet?

Edit: I can imagine a truly elegant battery box that’s accessed from below, in which the battery is held in place at one end by the front or rear plate, and an angle bracket keeps the battery from sliding away from that end plate or dropping onto the carpet.

Except the same four bolt pattern is present at the stationary, center wheel, implying a stationary block of some sort.

Every bearing is (or should be) mounted in a block, not a plate. There is no need for a bearing where the main drive shaft passes through the inner plate, as it is supported by a bearing in the motor mount plate at one end and a bearing block screwed to the outer plate at the other end.

You don’t need a bearing there (and having more than two bearings per shaft can be a pretty bad idea in general), but I don’t see how this has much of anything to do with whether or not an integral versus separable gearbox is a good idea.

You will stiffen up your frame dramatically if you connect the two plates in each drive side. We always connect up the box sides. As others have mentioned a bellypan will add significant rigidity as well.

Looks like a solid start!

Well I cant speak for the team but if you look very closely at the outside blocks there appears to be slots cut in the metal around where the bolt go through. I would guess that this would make those blocks able to slide to tension the chain. But without the original designer posting about it I cant be sure. Just looks like that to me.

Great catch, I couldn’t load a hi res picture earlier and missed that entirely.

In that case, I would maybe be concerned about “toe-in” on the bearing blocks, where human error results in the blocks not being perfectly concentric / aligned, hurting efficiency. But that fear is probably not a huge concern since a lot of teams do this without problems.

I was mostly addressing the point that it:

That is, using adjustable bearing blocks allows compensation for coarser hole placement tolerances in the chassis plates.

First of all, thanks for all the great feedback! This stuff really helps to make a design better on the next iteration.

I’m going to try to answer all of questions. Let me know if I missed any. Answers:

3rd CIM - Yes, we could add a third CIM very easily. There are a lot of threads about the pros and cons on 3-CIM drivetrains. Personally I’m not a fan of them, but the team might choose to do one if the game really required it.

Encoder - We decided not to add an encoder to the offseason drivetrain, but the design would be adapted to place the encoder above the CIMs (where a third motor would go) or out in front of the gearbox. It would be driven on its own shaft driven by a belt or gear.

Front profile - The front bevel is for avoiding obstacles. We decided to try it in order to get a feel for the tolerances on the bends. If it wasn’t needed for the game we would just make it square and move the wheels forward a bit.

Belly pan - Yes, we just haven’t put it in yet.

Center drop - The 2014 kitbot had a .125 and it couldn’t turn worth anything. The 2015 kitbot has increased drop and it didn’t have any problem turning. We decided to go with a large drop to replicate that. We are using different wheels though, so we’ll see how much that plays in (probably a lot).

Tensioning - The corner bearing blocks are mounted on slots.

60 tooth gear - Yes, it’s laser-cut and stacked. The 60 tooth gears are on backorder from WCP so we had to do something in the meantime. Overall they work pretty well. We will swap them out when the gears come in.

Gearbox configuration - I’m not 100% sure how this is less serviceable than a typical gearbox. Four nuts hold the plate on and everything else sides off once that’s off. There is plenty of room for tools where needed.

Flanges - I kinda missed the boat on putting the inside flanges inward for the belly pan. It was pretty much just an oversight. That’ll be on the next iteration I guess.

Electronics placement - On this design we will be mounting the electronics on the belly pan in a typical configuration. I have always liked the idea of an inverted electronics board. This drivetrain is actually based off of a design I did a while back which had an inverted electronics board. If I remember, I’ll post some pictures.

Once again, thanks for the feedback. If I missed anything, let me know.

To clarify, do you mean that electronics are accessed from the side of the robot that normally faces the carpet? If so, you will probably want to arrange your cart so that the robot is normally transported in a horizontal configuration compared to its competition vertical configuration. This was not really an option for many 2015 robots, as they were taller than 45" in competition configuration.

For your first custom drivebase this looks very good. Love the simplicity in the gearbox with the weight and space savings of just putting in what you need.

You have so many options available when doing a custom drivebase. Throw in the mounting holes for a few other COTs gearboxes so if you want to move to an off the shelf shifting or single speed option you have the ability to do so pain free. In 2014 we machined our plates to accommodate three different gearboxes so if we weren’t impressed by our first option we make a mid season swap. All it takes was a few minutes to add the holes.

As for the drop, if you are using Colsons you can always shave them down on a lathe to tune the drop if its too much or too little.

Looks great!

Could you explain your reasoning on this? I can definitely see wanting to have the robot easily moved to an electronics-accessible position in the pit or in queue, but can’t see the benefit of transporting it in that configuration.

Diagnostic lights.
Ethernet ports.
USB ports.

Sometimes, you just gotta download that one patch of code in queue…*

There are a couple of other reasons not to do inverted, though. I think the most basic is: There’s often random pieces of metal on the field, either as structure or as robot droppings. Last thing you want is one of those connecting pins or sides of power. And according to Murphy’s Law… well, let’s just go with insulating every connection on the robot thoroughly as well as covering all open ports and leave it at that. Also see “dirt in port”.

*You THINK you gotta, but you DON’T. Programmers can be persuasive.

Those seem like great reasons to be able to easily move the robot to a position where the theoretical electronics on the bottom of the robot are easily accessible, but not reasons to be transporting it in that configuration.

How many times have you had to move a robot to the field in a big hurry because a repair took longer than planned? Sometimes it’s just easier to roll in whatever configuration it’s in. There’s also on-the-fly battery changes (but that can be designed around).