pic: Team 192's 2014 Gearbox

This gearbox is simply gorgeous! Really love how it looks, obviously… but it certainly looks like it’s filled with so much design!

I assume it’s just the two stages, given the space on the gearbox and your speeds (19.5 and 8.5 fps)? Are those numbers theoretical, adjusted, or actual? I’m curious how you find the 3 CIMs combined with such high gearing… Seems like - especially for a full weight robot - that could be very battery-intensive.

Those are the speeds adjusted by a .85 speed loss constant, as for the battery drain: our robot is looking to be 75 to 90 pounds so the loading is appropriately scaled down.

This is beautiful, and it is a great example of my biggest problem with FIRST material usage rules. You can buy a gearbox that is developed and built specifically for FRC in the offseason, test it any way you want, and then use it without restriction. If you design and build one yourself, you can’t (without restrictions). It definitely does not inspire teams to do what you did. Excellent work!

Hey, I really like how you used the pulleys as the first reduction off the cims. I found what I think are the same pulleys and they seem to have 2 8-32 set screws. Did you guys just crank one of the set screws into the keyway of the cim, and if so how has that been working out for you?

The pulleys are manufactured with two M3x.5 set screws, we then drill the bore out to 8mm and broach a 2mm square key way and use that to transfer torque.

Of course it’s fabulous. I remember strolling over to your pit with our captain at Champs last year while you guys were gone to a match. We ended up marveling at the spare transmission you had on display for about a half hour and talking to a pit crew member about it. That’s still the most amazed I’ve ever been by seeing an FRC robot part in person. This year, it looks like you’ve stepped it up again, and I commend your efforts. There might not be another team that condenses so much art into a transmission year in and year out.

That sure is purty. Did you consider using a double sided belt (http://sdp-si.com/ss/pdf/80502117.pdf) and running it around all three motors in a serpentine path? (You would also have to relocate the motors a bit). That would let you get rid of some of the overhang, and flatten the gearbox width a bit. That said, it’s pretty compact as-is. How did it perform?

When I was creating initial sketches of the gearbox, I played with using double sided belts but found that I would need multiple idlers to achieve proper tooth engagement on the cims and that a double sided belt would actually lead to a larger gearbox with more empty space inside it. I decided that the marginally lower CoG due to that type of motor placement did not outweigh a smaller gearbox with little to no dead space.

As for the performance of the gearbox, we have not had to replace anything on them after 2 regionals and the drive base (the robot with the upper structure unbolted), as a whole, is operating around 53-58 dB.

Our gearbox won the Excellence in Engineering Award at the three Regional’s that we went to this year.:slight_smile:

I was wondering what the weight of this gearbox was? Did you cheesehole any of the gears?

The gearbox weighs 1.5 lbs without motors and 9.3 lbs including them.
As for cheese holing, the pinions were too small to be cheese holed and most the aluminum on the bull gears was bored out to make cavities for the bearings to press into.

I’d like to hear more about the design process that you went through. What design tools did you use? How were various iterations evaluated? What recommendations would you give other teams interested in designing their own gearboxes?


I’ll start with the last 3 questions then get into the design process.

What design tools did you use?

The whole gearbox was completely modeled in Inventor and then individual aspects were analysed in Solidworks. The reason for the two programs is because the team models in Inventor, however Inventor struggles with FEA while Solidworks does not. Additionally I used the gates technical manuals a lot as well as their belt theory manuals to get a strong understanding of how the belts preform.

How were various iterations evaluated?

The competition gearbox is actually about the 8th iteration of the gearbox, the 7th iteration was the one that was built in the fall as the prototype. Several of the earlier iterations were improved on so quickly in CAD that they weren’t fully completed before moving onto a updated design. Most of the evaluation of these early iterations was done almost entirely by myself with a little input from Joey Milia. Once the 6th iteration was completed in CAD it was more formally reviewed by Joey and a few other mentors and members of the team. The design was heavily assessed for, manufacturability, ease of assembly, ease of maintenance, reliability, cost, size, and weight.

After the corrections, parts of this 7th iteration were tested in solidworks using FEA and the gearbox was manufactured to do physical testing. The changes between the prototype and final are subtle, mainly changes in motor placement, gearing, and the shifting shaft profile. The shifting shaft continued to be tweaked nearing the end of build as I saw how they performed.

What recommendations would you give other teams interested in designing their own gearboxes?

My main piece of advice would be to have a clear goal of what you want the gearbox to do and know what having this custom gearbox would let you achieve. If the taxing on the team’s resources outweighs the benefits, don’t waste your time; there are a lot of good gearboxes you can just buy and I’d suggest just buying one of those.

If you do decide that a custom gearbox is beneficial I’d recommend you make really a good layout sketch that has all the parts in your gearbox. Having a sketch with every element of the gearbox on it makes transitioning into 3D, and quick adjustments to the entire design, very easy, speeding up the iterative process.

From here you can base all of your parts off of these one or two layout sketches. That way, if you make any changes in the sketch, all the parts update so you don’t have to remake the entire part.

For example, here is the layout sketch of the build gearbox:

I’d like to hear more about the design process that you went through.

The best place to start would be the goals for the Drivetrain. For the past couple years 192’s main goal has been space efficiency, and sometimes that was at the cost of power efficiency. For this year I wanted the gearbox to both, have a smaller footprint than it has had in the past, and not have the inefficient right angle stages we’d used in the past. To achieve this I looked to combine techniques used by others and that we had used previously. I drew on, 971’s gearboxes that place the motors over the wheels, the VEX ball shifter that reduced the size of a two speed gearbox, and the use of belts and placing the motors on top of the gearbox that we used in 2013.

While I was deciding what options to pursue to reach the goals I made the below matrix of the possible ideas. I highlighted the possibilities I liked in red and added +1J to the options that Joey approved of.

I used some of the highlighted/+1J ideas as a goal for the design I wanted to prototype. (Note these were only things I wanted to do, and it was completely acceptable to cut some if they proved to make the design too bulky or raised other problems.)

Once the general aspects of the gearbox had been established, a modified version of JVN’s design calculator was used to determine the final gearing. I was careful to use only gears available in aluminum from WCP or VEX and belts and pulleys available from SDP-SI.

The next step is the actual design of the gearbox. Moving into inventor and laying out all the parts of the gearbox and playing with geometry. This sketch included everything, gears shafts, bearings; everything that would affect geometry. From here the design process moved along like I described in the question about iterations. I found a layout that worked and made lots improvements from there.

1.5lbs? Not 2.5? That’s insane (ly awesome)!

Thank you so much for taking the time and posting these details!!!

I am thinking about designing a gearbox that would have a CIM, a Mini CIM, and a larger pulley belted together and I had a few questions.

Are there any problems with belting the 3 together considering the ~15% difference in free speed between the CIM and the Mini CIM?

Is 20T the smallest you can get away with with this type of belt, and why was it chosen over 9mm wide, 5mm pitch HTD belts?

Did you have any problems with this gearbox over the season and did you end up needing to use the tension adjustment CIM mount holes?


We did not have any problems with the mismatched free speeds. CIMs and MiniCIMs were designed to be interchangeable; Additionally the free speeds of both motors have a huge tolerance range so you really have no idea how mismatched they are.

CIM Free Speed: 5,310 rpm (+/- 10%)
MiniCIM Free Speed: 6,200 rpm (+/- 10%)

Firstly, no; you can put a smaller pulley on the CIMs. GRT has used 16 tooth pulleys in the past and you might be able to go smaller, we haven’t tried.

However, there are a lot of considerations that you have to be aware of when designing belt/pulley system. (here are just a few)

  1. To get the full strength from a Gates Powergrip series belt you need at least 6 teeth in mesh and at least 60 degrees of wrap. When you use a smaller pulley this becomes harder to do.
  2. When you use a smaller pulley you put the belt through a tighter radius, which can lead to a decreased belt life.
  3. The width of a belt (IIRC) linearly correlates with it’s maximum power transfer

HTD vs GT2: HTD belts have substantially lower max power ratings so switching to HTD would have put the belt far out of spec.

5mm vs 3mm Pitch: using 5mm pitch components would have been safer due to their increased load ratings, however the pulleys required to get the reduction (with correct wrap, ect.) I wanted would have been a lot bigger and did not fit with the rest of the design.

We never removed the gearboxes from the robot over the course of the season and they never need any repairs. The holes are to account for the manufacturing tolerance in the belt length; once installed, the belts should not need to be retensioned.

9mm wide, 5mm pitch is more suited for the higher torque, lower speed belt spans of drivetrains. For an initial reduction, smaller belts of smaller pitches can perform just as well.

Whether HTD or GT2 is better is subject to some debate, and it also depends on the application. GT2 is rated for higher loads, but some say HTD’s deeper tooth allows it to handle reversing loads better. HTD is probably a bit closer to optimal at the end of a driveline compared to the beginning.

7000 RPM @ 4 max HP is still well within 5mm pitch GT2 belt limits, so they really could be used on any stage, but as you said they are better suited for end stages, mostly due to their size.

I’m sure many on 971 could have a long discussion with you about the strengths and drawbacks of the HTD vs GT Series belts. IIRC they ran GT3 belts in their drivetrain due to the shock the belts experienced with their high CoG. (Take that with a grain of salt as I did not look at their robot too much this year)

I personally go with GT2 in every application as I have yet to find a place where they don’t ‘just work.’

(here a link on htd vs gt2)

This is something my team has gone back and forth on as well. Our first belt drive had several failures and was done with 9mm HTD. We changed some things including the pulley size and went to 15mm* GT2 for a few years. This year we used 15mm HTD in order to get parts faster and we had no problems. The only reason we don’t go GT2 is that pulleys and belts tend to be less available with longer lead times than HTD. For something like a gearbox like this I would absolutely go with the stronger tooth profile.

*(At this point I’m pretty sure that for the end of a driveline 15mm is a lot safer than 9mm for 4" wheels and pulleys in the 24T range, regardless of tooth profile. If the difference in tooth profile is your factor of safety between failure and success, maybe try going a bit wider.)