90 Degree Gearbox?

Hello CD Community!

I, along with a few others on 1405, have been working on an off-season project that we would hopefully be able to use on robots for years to come. One of the requirements we set forth was the placement of the CIM motor used to drive wheels. We’re trying to reduce the amount the CIMs hang out into the center portion of the robot’s drive base.

One of the ideas I came up with was using a “90 degree gearbox” on the CIM, so we could move the mounting position out of the center of the robot.

I don’t know enough about this form of power transmission. I’m doing some research, but I’m hoping someone may be able to share some experiences with these gearboxes! Right now, one of my biggest concerns is that these gearboxes typically use worm gears, which, as I understand, are impossible to run in reverse. Which, if this is going to be part of a wheel set-up, tends to be an important feature…

If you have any experience using 90 degree gearboxes, I’d love to hear your thoughts on them!


A worm gear is a commonly used method for a 90 degree gearbox. And they can run in reverse, however they are usually hard to impossible to backdrive. In other words when you stopped applying power the wheels would come to a stop very quickly.

Another option for a 90 degree gearbox would be the gearset used in most angle grinders and angle drills. These use bevel gears and will be easy to backdrive.

Something else to consider is the reduction. Bevel gears will allow reductions similar to normal spur gears like you’re used to working with, while worm gears are usually more suited to higher reductions ratios.

Okay, I was confusing “anti-backdrive” with “can’t be reversed”.

I’m thinking now that bevel gears will be the way to go. We’re going to be looking for normal gear reductions from this box. Definitely nothing high enough to justify using a worm gear.

Worm gears work in both directions. For example, the Window motor transmission uses a worm gear. However, they tend not to backdrive very easily, or at all, depending on their design and efficiency. The biggest problem with a worm gear drivetrain would be the relatively low efficiency. Due to the way the worm gear interacts with the worm wheel, the design is inherently less efficient than a standard spur gearset. For a drivetrain this is particularly important.

A more efficient solution (but still less efficient than regular spur gears) would be to use bevel gears. These come with their own problems with alignment and thrust loading, but they aren’t unsurmountable. Here’s an example by team 192.

Generally I think efficiency is much more important in a drivetrain gearbox than packaging, so I’d use a regular spur gearbox in a drivetrain most of the time. Depending on your drive layout you may be able to arrange the transmission with the CIMs over the wheels to save space.

I thought the same thing. The spacing of the CIM is more of a “Nice to have” than anything else, so we’re not trying to make our designs revolve around that idea. However, it is an interesting challenge, so we’re trying to make this gearbox meet as much of the requirements as possible.

I’m looking into bevel gears right now. Fortunately, I do have some experience with these (as 1126 ran swerve this past season, using bevel gears in the module).

AndyMark sell their WormBox

Take a look

Heres how 2783 solved the issue last year. Standard nano toughbox reoriented and a few modifications to the output shaft to allow for reorientation. (just had to add another snap ring to hold everything in place)



Depending on how your frame is designed, you could mount the CIMs above the drivetrain like 971: https://picasaweb.google.com/lh/photo/77eBELSkPvqRrQl4q_1YyNMTjNZETYmyPJy0liipFm0?feat=directlink

Of course, this isn’t a COTS solution, if that’s what you’re looking for.

If you go the bevel gear route, you will want as much of the gear reduction after the bevel gearset as possible. This is why you see so many coaxial swerve modules with very large sprockets on the wheel. The higher the reduction before the bevel gearset, the more force there is available to push the bevel gears apart (which they will do by default), ergo the thrust load issues Chris mentions.

The other issue with bevels is that they’re relatively less efficient than spur gears at low pitch (bigger, deeper teeth), yet very easily chatter and separate at high pitch when under load (especially on a vibrating platform like a FRC robot or an aerial quadrotor). Spacing & rigid mounting are crucial with higher pitches.

I would be remiss to not remind you that the CIM output bearings are not designed for high side loads. If you intend to use them, be sure to add bearings or couple to another shaft with bearings to take the loads. I have seen several CIMs used as the direct input for a worm drive only to have the motor fail when the bearings become worn.

Those aren’t bevel gears, but rather a 4 start worm gear that is pretty high efficiency.

that design was first run by a student on 1425 in the 2007 season and I was very impressed by it. He argued that the efficiency of a single 4-start worm gear is comparable to the 2-4 stages of reduction of lower quality gears most teams use. It was likely still more efficient than many teams.

A worm gear of that style is also easily backdrivable.

They’re interesting to try, but ultimately I bet you could achieve similar results with CIMs over the wheel/frame. If you did both…that would be one hell of a space saver.

As for performance, the worm gears will break in over the season and driving dynamics will change. A robot that used to stay still on the bridge suddenly begins to roll at champs. The bot wont be able to be pushed by hand easily on week 1, but by champs it should roll…like a champ. Coasting gets easier and the gearboxes get quieter.

I think my team is working on making it removable from the frame without intruding into the center at all. As it is, a gearbox replacement means all the electronics have to come out, which really negates some of the benefits of space saving.

971 this year moved the CIMS over the wheel/frame. On their picasa they have pictures. Its a really really sweet setup!


I was thinking of their bot specifically when I wrote that, but I guess I failed to mention them. This is what I get for trying to post from my phone.

Anyway, 971’s setup is really nice, we got to spend a good amount of time chatting this season at Davis, SVR, and Champs. The fan, the mounting of the cims (and the entire rest of the robot) are all pretty fantastic.

Dust off the old finney photo album and take a look at the 2009 Robot!
The output shaft of the CIM is pointed towards the sky. I may or may not have the detailed drawings somewhere.

Resurrecting a “dead” thread, but between the 192 90 degree gearbox and the 971 simply-put-motors-in-opposite-direction solution, which is more viable? I can pick this up as a project for my team and work on it this semester. I’d like to look into ways to keep the CIM from pointing directly into the middle of the robot. Being a former Head of Electrical, you can imagine I know how cramped it gets in the middle

The method 971 employed is easier as it’s still just spur gears.

Alright. Are there other methods of reducing the footprint of a CIM motor besides the ones mentioned? (90 degree gearbox and flipping the direction the motor points) There are the various swerve modules that cut down on the footprint but I’d rather leave the complexity of making wheel pods out for now. I’m sure I’ll get to that eventually, but not at this moment. One of our mentors mentioned running the motors via shafts, ie the same method cars use for their drive (albeit without the ackerson drive). Any thoughts on something of that nature?

This is certainly an “out there” suggestion, but you could theoretically mount the wheels such that they rotate around the CIMs, as if the CIMs were a dead shaft. I’m sure the next person on the thread will have the link, but I remember seeing pictures a few months back of a swerve drive setup that did just that.

I really don’t think you should worry too much about the footprint of the CIMs inside the robot. It’s a lot more simple to re-arrange electronics to fit around CIMs than to design and spend lots of time machining a fancy transmission to save a bit of space. I saw 192 at SVR, and they really didn’t use the extra space that they got with that type of gearbox. That said, if it’s absolutely critical to not have the motors sticking into the frame, I would recommend the 971 approach.

You know, I was thinking the same thing a few hours ago too but decided it was a little impractical.