VEX Differential Guide

VEX Differential Guide

After I posted my video on ratchets, many people asked me to make a video on differentials as well. In this guide I go over how a differential works, and 4 different examples of how you could use one.


credit to tilden obvs

The Core Design

I’m calling it this because the only implementations of this design have involved 2 mirrored copies, causing 2 spinning “cores” to be formed in the chassis. Last year, one of the VEXmen teams used a design like this to share power between their drive and their tilter. They used 393 motor bearings to decrease friction. While this does make the system slightly smoother, it’s not necessary in my experience.

Either the shafts connected to the motors, or the cores themselves, spin depending on the difference in speed between the two motors.

If both motors spin the same way, the shafts will spin, spinning the wheels. The chain connected to the shafts force both shafts to always be locked in their rotation, regardless of how the motors spin.

If the motors both spin in opposite ways, however, the chain that locks the shafts together prevents the shafts from spinning at all, as equal forces are acting trying to spin the shaft either way. Instead, this forces the cores themselves to spin, and the meshing gears close the differential.

Here’s an example of this kind of differential:

DiffyLift

It uses one of the outputs to power the drive, and the other output is diverted to power the lift. We can see that the lift uses the power of all 4 motors, as does the drive.


The 4-Bar Design

This design includes a 4-bar linkage and a ton of gears.

Systems similar to this have been useful in both Tower Takeover, to share power with a tilter, and In The Zone to share power with a Mobile Goal lift. To the best of my knowledge, a 5-digit team from Colorado that I’ll leave unnamed pioneered this design.

On a system with 4 gears, the outside gears spin opposite ways. When the outside gears are forced to spin the same way, the gears lock up, and don’t spin.

Because the 4-bar is allowed to pivot, instead of snapping the gears, as a locked-up system like this usually would, the bars simply pivot, and the pinions crawl across the big gears.

When the pinions spin in opposite ways, however, the gears spin smoothly, and the wheels on the chassis spin.

Here’s the example that I made for this style of differential:

Diffy Mogo

The 4-Bar could act as a Mobile Goal lift, effectively allowing the same 4 motors to power both the drive as well as another necessary robot function. Due to the absurd amount of gears on this drivetrain, it’s not the smoothest to drive. If I were to rebuild this, I would shorten the chassis, and reduce the amount of gears.


The Hood Design

I am not exactly sure what to call this design, as I have seen it used for things other than hoods, but no other name really applies.

I first saw this type of differential on some early designs this season.

This differential works by having two motors always spinning in the same direction, but at slightly fluctuating speeds.

One motor is on a bar that pivots around a shaft with the main flywheel, while the other is powering said flywheel.

When both motors spin at the same speed, the flywheel is powered by 2 motors. When one is slightly faster or slower than the other, the power of the flywheel decreases slightly, but the bar can be moved to a new position.

Here’s a basic example:

Diffy Hood

This type of differential could be used for a moving hood, such as one where you want to angle the shot from a flywheel or one where you want to let a ball out of the system entirely.

Here’s an example of a differential of this type more relevant to this season:

60 Frames


Differential Swerve Drives

hese are complex gear trains and chassis that allow for full 360-degree movement with full power in every direction.

The wheels are in “pods” that rotate, similar to the wheels on a shopping cart. The difference in the speed of the two motors determines if the wheels will be moving, or changing orientation.

Swerve drives in general aren’t used much in VEX, as they often pose more challenges than they create due to their complexity.

On the other hand, swerve drives are used quite frequently in FRC, as their robot’s requirements and constraints are very different than the ones in VEX.

Here are some examples made by Kyle from team 81818X of some differential swerve chassis. I wasn’t able to build an example for these, as I do not own enough of the necessary gears to build it.

eb56aa1f-1abb-40f1-82e6-5455647ffa78 e1a90b7d-6ce3-4a55-a0bc-04f67f6696c3

You can see the humongous amounts of gears in the drive, and how much space they take up. Huge thanks to team 81818X for letting me use these renders in this guide.


Conclusion

Using differentials can be an effective way of sharing motor power between multiple different subsystems on your robot.

As some people have requested I do, I’m sharing all the CAD files for all the robots and mechanisms built in this video. These include Pack and Go zip files for inventor assemblies, Blender project files (where applicable), and some renders. Here they are:

https://drive.google.com/drive/folders/1bSPJzHFWN7Fuv-N42cNEoI7Vg4V3sh3f?usp=sharing

Remember that there are many many more ways to use a differential on your robot than the designs that I mentioned. These are only a few of the most common examples.

I hope this guide helped you, and have a great Change Up season.

42 Likes

coolio

5 Likes

For those keeping track at home cough @RobotMesh_Support cough I did in fact credit all of my sources.

1 Like

How do you do the line separation thing? Wish there was some kind of documentation for the text formatting.

1 Like
2 Likes

High-quality video and explanations. Keep it up!

3 Likes

Very nice guide, Taran. This was well thought out and put together nicely. Thanks for your contributions to the community!

Thanks! This is a new style of video I’m trying. Using more animations and still images rather than a video of me talking. Google slides really pulled through here.

On the flip side, when I do this it takes forever to make a video. On top of general computer issues I’ve been having, my laptop really slowed down these past months, and it took weeks to render a simple animation. I also kinda took a break from things over the summer.

1 Like

And for those of you that DMed me asking why I wanted to know this, now you have your answer.

4 Likes

Edit: I got a feeling that the other post will be flagged as off-topic, so here’s this.

Thanks Taran for your amazing contribution to the community with your extensive CADding, researching, and youtube-video-making. Man really took time away from making his Change Up robot the best it could be to help some people out. Real cool man.

1 Like

Taran,

The fact that you CADded, built, and spent time filming and explaining various types of differentials - very nice!

The fact that you didn’t link to any of the previously posted tutorials, videos, or robot reveals - boo!

393 gears are the most important part of this design that prevents additional friction losses and makes it feasible. Without them friction losses are going to be large and you will be better off not doing differential at all. But when cores and motor outputs are mechanically isolated with 393s - you get almost no additional friction over a regular drivetrain. Also, using chain to close the differential loop makes it tricky to tune - in this case all gears design is much more smooth and robust.

I’ve seen a good number of teams at ITZ Worlds that used less gears and were very successful with 4-bar differential design.

I believe, this half-differential idea had originated in Utah - many 2131 teams used this type of differential for their MoGo lifts. By the way they did very well in rankings and their club won Excellence at 2018 VEX Worlds.

The advantage of such design is that only half the motors motors (2 out of 4) have to ride on the 4-bar thus reducing friction losses. The flip side is that those two motors will only contribute power in one direction, depending on which end of the range they are parked. They cannot contribute any power while not at the end of the range.

Therefore this is not a viable design for the variable position hood.

This is very unfortunate - would have been epic if someone actually built it. Friction losses would be humongous because of the limitations of available vex parts though.

However, @Kyle1’s innovative way to build differential core without bevel gears offers some very interesting design opportunities, if friction losses could be reduced and input torques are properly balanced.

Taran, you could improve your topic by a lot, if you offer readers who are interested to further study this subject, some easy to follow links to: vexforum, chiefdelphi, or external resources like this one:

Conclusion

Please, try to avoid making general conclusions based on the limited amount of testing of unoptimized prototypes.

There is a number of teams who have qualified to Worlds with their optimized version of differentials and shared in-depth analysis and lessons learned about performance of different designs:

Plus many more that were shared on VTOW but didn’t make it to vexforum.

Based on those reports, I could say that when built well with understanding of underlying physics and paying attentions to details, 4-bar differential could be very efficient and rotating core design using 393 gears could have almost no additional friction losses over the regular drivetrains.

Half-differentials could have niche applications like with 2131 MoGo lift, but are not suitable for general application like variable position hood.

@Kyle1’s design for continuous dual motor differential transmission has some intriguing potential but is, probably, too complicated to be done with vex parts, unless somebody comes up with a clever way to minimize friction.

12 Likes

Well I mean I did shout out your team

Point taken, will keep in mind

Yeah, I get that, and I understand why that should be the case, but when I was playing around with my robot I didn’t find it to be significantly helpful at all. I had similar power on both mechanisms when I tried both. I completely understand the idea behind this, and it’s a nice thing to show off to the judges, but from what I tried there really wasn’t a significant decrease in friction.

I would be interested to try a 2-motor version of this design with turntable gears to see what would happen, but I only have one set of them, and I’m not about to spend $20 for a one-off experiment.

Had to make mine different from @Codec’s somehow :wink:. Also as I said above, it really didn’t cause any issues. I am frankly shocked by how smooth that robot is.

  1. Could you not set it up to apply power moving the indexer forwards when the hood is fully closed? It might theoretically work a little bit if you did this.
  1. I do agree with this. Simply an example.

If Kyle could share the CAD file for that with me I might give it a go. I didn’t see this one before, but I really can’t see enough from the image to build it. My brain can only visualize so many gear trains at once.

Good point, I’ll take that into consideration whenever I next make a guide like this. This was definitely supposed to be more of an intro than a full-out tutorial, so I see why that could be useful to add.

I would lastly like to add that my CPU quite literally hates me at this point. I think I broke something in the process of all this modeling and rendering because my computer now struggles even running Chrome. I dunno what happened, but it’s kinda annoying. It takes longer to render a single frame on my ASUS laptop with a gen8 i5 CPU (built-in graphics, using the CPU for rendering) than my 80 dollar Chromebook running Linux with Blender installed.

I think I need a new computer. Supposedly building a computer is one of the suggested options for the IB MYP personal project, so maybe I can convince my parents to fund that.

1 Like

First may I say, Taran thank you for this post. So far I believe it to be the best post in relation to differential designs so far that could help newer teams think about more complex designs. I also appreciate your use of CAD and animation to demonstrate many of these designs.

The only problem I have seen with your original post is “The Hood Design.” Theoretically it does work, but not the way I think you think it does (or the designers for that matter). Technik3k has already been over this and I would like to add my 2 cents in the effort to make this differential thread even better than it already is.

This system does work, you could theoretically power both a flywheel/indexer and an angle changing hood at the same time, but there is no real power sharing between motors. I think the easiest way to picture this is to imagine this system is running with both motors running at full speed and a ball gets launched out of the indexer/flywheel whatever it may be. The Motor driving the indexer/flywheel will slow down due to this added weight in the system that it must accelerate, but the swing arm does not share this same load, instead it just counters the spin of the other motor and does not slow down, causing the angle changing hood or whatever device it powers to move and adds no power into the indexer/flywheel system. The only way these two motors would share power is if the motor on the swing arm hits a limit, forcing all of its power to be transferred into the lower gears, instead of into the swing arm to keep it at a certain angle.

Edit: This hood “differential” is functionally equivalent to those two systems being completely separate with no mechanical links.

A similar situation happened to me in turning point in regards to our center powered wheel. I programmed our center wheel (We had a programmer but we had an unfortunate team situation forcing me to program the entire robot, something which I had full ability to do just did not have a programming mindset) I programmed our center wheel to go the speed of the average of the measured speed between the two drive sides. This sounds good in theory but in practice as soon as the drive experiences an unexpected force, causing one or both drive sides to slow down due to the motors having to fight this force, the middle wheel also slows down, but due to the coding, and does not actually add any pushing power to the drivetrain. (this is in addition to the fact that the middle wheel has a much slower response time, which was very noticeable affect)

11 Likes

Yep, it’s an open differential. Cool for playing with, but no real power sharing in most circumstances.

@Kyle1, any chance you could share with me the CAD files for this?

https://www.vexforum.com/uploads/default/original/3X/1/a/1aac668cd0a26cf384c69a1300cd3c405fa1cd45.jpeg

Maybe you should just dm him for this. Also it’s not to complicated to make yourself but you do you.
Edit: Oh wait I forgot there is some weird spacing with the 36 tooth gear but that shouldn’t be to hard to figure out

I don’t have the cad files anymore, they’re on a computer I left for college. You can see everything you need to in the image though, and there’s an exploded view that makes it more clear.

image

I do have CAD files for some of the swerve drives (they’re incomplete in some areas though):
https://drive.google.com/drive/folders/10QsASOg6LgQKjzw-GA5T830QDoo-UXMD?usp=sharing
Low:
image
Skinny:
image
No Crown:
image

Full gallery of what I’ve done:

23 Likes


More on the hood concept here. I haven’t seen any other versions of this except the one Taran posted and this one.

1 Like

This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.