ok. this is going to be a weird one.
I have been using planetary gearboxes for a while and got great results. sometimes so great that I felt that there was no use for conventional spur gear gearboxes in most cases. and yet, I see the top teams rarely use them and rather go of with conventional reductions. why is that? is there something I’m missing? assembly-wise, maybe they are easier to assemble? maybe there is a significant loss of power? more breaks and damage to the system? I haven’t noticed anything, hence I am asking for your help.
thanks in advance!
I am certainly not apart of some of those elite teams that seem to never use planetary gearboxes, but based on what I know it is probably because of a few reasons:
- Planetary gearboxes have smaller, weaker teeth that are more likely to break under extreme load
- Planetary gearboxes have a lower efficiency in mechanical output compared to your typical spur gearboxes due to higher friction losses.
- Planetary gearboxes introduce way more backlash into a system that you cannot fix like you can in a conventional gearbox. In applications where precision really matters such as an elevator or an arm this makes some potentially serious problems in performance.
These reasons might not be enough to justify an average team to spend the extra time to build a custom spur gearbox for a certain application, but for a super-team with a crazy amount of manpower, expertise, and resources the performance benefits of a conventional spur gearbox happen to be something they just cannot pass up.
4272 primarily runs custom gearboxes with 20dp gears from AM/Rev, especially when small ratios are all that’s required.
You can get up to 8:1 with a single stage reduction (on a Neo) with off the shelf gears. For many applications, this ratio (or less) works out perfectly. You could use a planetary in cases like this, but the mechanism would stick out more. Typically we want to avoid this.
If we didn’t have access to a CNC router, we would probably use more planetaries, but because we do, it’s easy to do the gearing as just part of the mechanism.
Where planetary gearboxes shine is there ability to do very large ratios. We did use a planetary as the first stage of our arm reduction. That was the only planetary on the 2023 robot.
Performance wise, no real difference. It’s just easier to design in many cases.
Three reasons I can think of:
-
Many ratios now can be done with single reductions or even just a belt reduction. These are very easy to do and don’t require all the additional pieces in a planetary.
-
Custom allows you the ability to control the packaging and you can fit some complex gear boxes into some very tight spaces.
-
For us, We have gears and a cnc so we can quickly throw together a gearbox without too much hassle. As of right now we only have ultraplanetaries stocked so we would have to order some which means waiting on shipping.
Designing custom gearboxes will offer you slightly more options for ratios and if you have very precise machining capabilities, custom gearboxes can be lower backlash.
In FRC planetary gearboxes are pretty always COTS, so they are usually the easiest gearboxes for lower resource teams to design with, and implement. When you select the right gearbox for the job, COTS planetaries are extremely robust, especially MAX planetaries.
Custom gearboxes can offer flexibility in shaping your gearbox to the space available in a mechanism, and offer the flexibility to have multiple outputs or shifters. This is particularly useful with mechanisms like shooters where you might want a flywheel geared much faster than your shooter wheel and the axial design of a planetary is hard to package. But, a planetary gear box is usually smaller overall than an equivalent custom spur gearbox so I think they are usually easier to package.
Another advantage I’ll note for planetaries, is that the ones from REV and VEX allow you to easily change your ratios even after a design is completed and built just by swapping out stages.
And another advantage of custom spur gearboxes is that you can much more easily use multiple motors.
Galaxia hasn’t had a COTS gearbox on any competition robot since 2018 (except for an AM Sport gearbox that was on the 2019 robot for one event before we removed it and WCP swerve modules last season if that counts) and there were several reasons for this very conscious transition:
-
Reliability- 20DP custom gearboxes generally hold up better in high load applications, and for low load applications you can usually get the ratio with a single stage of either belts or gears which is lighter and takes up less space
-
Repair and maintenance- since planetary gearboxes are COTS, this means when something in them does break down you better have a spare for whatever part it is you need replaced (Which is even harder as an international team that has to pay extra shipping when buying all this stuff). The fact they’re held together with Imperial fasteners when everything else on our robots is metric also doesn’t help. By comparison, all you’d ever need to fix a custom gearbox is spare gears.
-
Packaging- this might just be because of how we design our robots but we usually find that packaging a planetary gearbox doesn’t work that well with our designs and it usually ends up working better with a custom gearbox where we have more flexibility.
We still use planetary gearboxes for prototyping since you can get one running a lot faster than if you had to design and manufacture a custom, but we’ve now gotten good enough at machining them that it’s worth it to only use custom gearboxes on a season robot.
While I wasn’t on 111 last year (on 112), 111 used four max planetaries on their robot. All the other NEOs on the robot had a gear ratio of less than 4:1. Max planetaries are just easy to design with when you want a pretty high gear ratio and have a pretty nice footprint. I think it’s fair to classify 111 as a “top team” last year.
Everything others have brought up is definitely true.
One more niche reason is the ability to add in little things like brakes, ratchets, and shifters to the gearbox, packaged in interesting ways.
Our 2019 arm shoulder/wrist gearbox featured pneumatically actuated engageable ratchets/breaks to allow the arm to be locked-out so we could climb off of it with our stinger. See the last 2 photos in this post.
2019 elevator also had an engageable ratchet for climbing and staying up at end of match. These types of mechanisms were more important before the simplicity of just putting a brushless motor into Brake-mode with a large enough reduction.
This, 1000x this.
Sometimes you just need something to fit to make the concept work, i.e. passing hatch panels through a robot, jamming a turret on a bot and having room for a climber, etc.
Photo of our robot this year
In our 2023 intake, I think we originally had no reduction (except that the two belt pulleys are different diameters for belt clearance reasons), but found we needed a bit more torque late in the build season. There just wasn’t the width to fit a MAXPlanetary so we used some REV plates and gears to make a reduction in the space. On the other hand, the wrist and elevator had the room for a planetary and were designed that way from the beginning.
Probably not the primary reason for top teams, but another benefit of custom gearboxes is that it’s a good design project for an intermediate mechanical student. It helps sophomores and juniors bridge the gap from “CAD a 3” spacer and mate it here" to “design a whole intake/drivetrain/etc”, developing that pipeline of skilled students.
I agree with the rest of your post, but this is not strictly true. While I’d trust a 20DP gear over a Versaplanetary, the MAX planetary is much stronger than a simple spur gear set. The gears are hard steel and the load spread amongst several of them.
Usually I use a custom gearbox when I need lower ratios and better packaging. It can also be lighter than a planetary. Packaging a planetary can be really difficult in many cases, especially if you need more than 1 motor.
As a noted MaxPlanetary enthusiast, I’d like to elaborate on these points. If you need a simple single reduction, belt + pulley is the right choice a shocking amount of the time for a few reasons:
- Easier to package
- Weighs less
- Costs Less. This is a big one for a lot of teams since a 1 or 2 stage MaxPlanetary is ~$60 - $90. I can do an equivalent single stage reduction with a VBeltGuys belt, pinion pulley, and a 3D printed pulley w/ WCPWest Coast Products hex insert (for those weary of 3D printed hex bores) for ~$10.
You won’t find my team using MPs for things like intake, shooters, and conveyors because of what I laid out above.
Modern planetaries are usually plenty. The AMAndyMark Sport gearboxes and REV MAXPlanetary are well-characterized by now and with no real surprises.
We use custom spur gear stages when we want to flip around mounting on something so it doesn’t stick out, or something like our 2021 robot’s conveyor system where we needed to power two rollers at almost but not quite the same speed. Actually, that demonstrated both well.
We also used a gear stage on our recent rebuilt intake to power two counterrotating rollers on our intake. It’s an important tool in the toolbox once you develop means to roll custom plates.
Bold statement: if it’s an intake or conveyor and you’re using a MAXPlanetary, you’re probably doing it wrong. UltraPlanetary starter kits are $55 with the 1/2" hex adapter shaft (and you’llLimelight, an integrated vision coprocessor have leftover stages to use on the next build), and it’llLimelight, an integrated vision coprocessor take a NEO 550 up to 30:1. And it’s super small and super light. I did a lineup on gearboxes running 1000-1500 RPMRevolutions Per Minute a couple years back:
With a stock NEO weighing in at 0.938 pounds (per REV) and with a bigger OD (2.36" OD vs about 1.5"), the NEO 550 in an UltraPlanetary should at least be in the conversation for someone designing a mechanism.
Sure, but not for me!
A N550 + Ultralanetary is very barely lighter than a NEO + belted reduction. Additionally, you now have use braincells to ensure your mechanism isn’t underpowered and the motor won’t burn out (less of a concern now), I really like the simplicity of a single motor beyond fringe mechanisms like a 2020 color wheel/traversals mech. Finally, N550 + UP is still more expensive.
I really like what someone here said about design that nowadays you can “overkill it in hardware, reign it in in software”.
Not sure if this was already stated in this thread but one benefit of regular gearboxes sometimes is the packaging. Planetaries can quickly get pretty tall (ex. REV ultraplanetary system) which can make designing quite an issue. Manual gearboxes can be made very flat to fit in different types of spaces, and compounding gears doesn’t add much depth either.
This was our logic on our modified Everybot intake this year, and it worked out great. “Why add more complexity & cost when slapping a NEO on it weighs the same and ensures that power will never be an issue.”
Best photo I have of our power setup on the intake (taken during initial assembly, shout-out to Ryan at ThriftyBot for the rough design/inspiration, and the 11t pulley):
For one, if you’re overtaxing an intake or conveyor with a NEO 550 I’d venture your mechanism needs work to run freer. That’s both in the context of “it’s a 200W motor, not the wimpy stuff us old-timers remember” and “we’re already making batteries scream for mercy in the modern age of drivetrains, why are you pulling amps for the memes?”.
For another thing, the price difference between a NEO 550 + UltraPlanetary and a NEO is presently $38.50*. Make it $29 if you use/print an output pulley compatible with their FTCFIRST Tech Challenge motion pattern instead of using the 1/2" hex adapter. And with more gear reduction in the UP, you can probably run a smaller pulley on the driven end–not only helping you package tighter on that end, but also probably shaving off a bit more money given how REV/WCPWest Coast Products/IFI price their pulleys and belts. Even without that, $30-40 and a modicum of brain cells is peanuts in the modern FRCFIRST Robotics Competition budget if the design benefit is there.
This is not to say that running stuff straight off a NEO is wrong–we’ve done it (or at a minimum designed it in before scrapping for other reasons) in 2020, 2022, and 2023. But there have been just as many instances where a NEO wouldn’t package around the other rollers or gears or whatever in the system but a longer and skinnier UltraPlanetary would slot in just fine. Ignore that potential at your own detriment.
*Starter kit $45, NEO 550 $28, 1/2" hex adapter $9.5, NEO 550 pinion $4 (well, $8 for two). NEO $48. And yes, REV’s price structure makes it cheaper to buy starter kits even if you yeet the HD Hex motor immediately; we found a team at our events that also ran FTCFIRST Tech Challenge and hooked them up. Leftover parts fit neatly in a Plano box until we need them.
This just isn’t true. Intakes for games like 2020 definitely benefit from the power of a NEO vs a NEO550/775Pro/etc. When you need something running at a high speed and having pretty constant loading. In 2020 you were quickly intaking 5 balls every cycle, and cycles were potentially pretty quick if you were playing the close clean up role; this is just one example.
I think a Neo550 w/UP can make sense for certain situations (swerve module rotation, shooter hoods, light duty rollers, etc. but for anything else it really works out to just be easier, simpler, usually cheaper, and usually lighter to just run a NEO/Falcon with a single belt stage reduction as Kremer said.