The 2 Bearings Per Shaft Rule (Swerve Module)

Hey, so I am designing a swerve. Any feedback would be great, but I am wondering how necessary it is to follow the 2 bearings per shaft rule, specifically in this case. Removing the top plate in this picture would save a lot of headache.

See the full cad here:

Why can’t you fit 2 bearings in the rotation gear?
The 2 bearing rule is pretty strict, especially for things like this. I would recommend having a bearing on the end of the bevel shaft as well, embeded into the fork.

There’s a reason behind the “2 bearings per shaft” rule.

To put it rather bluntly, a shaft with only one bearing (or other form of support–there are non-roller bearings, often called “bushings”, that can be useful in the right situations) will bend due to loading. Shafts that bend are likely to fail faster than shafts that don’t.

If a shaft has two bearings, it won’t bend as much as it will with only one, and will end up lasting longer.

Here’s a thought… do some research into other types of bearings. Not all bearings are roller bearings.

And then there’s the kids who slay me by putting three bearings per shaft and over-constrain it. Hey, if two bearings are good, six must be three times better!?!

Sent from my iPhone using Tapatalk

On a very long shaft 3 bearings can work fine, even 4 if you line up the holes right. 1072 did 4 bearings on their climber last year without issues.
That being said, generally if I feel like I need more than 2 bearings on a 1/2" shaft, I just upgrade to 2 thin bearings on a 1" or 20mm shaft.

Also, note that aluminum is more flexible, and therefore more forgiving than steel for a third or fourth bearing. Of course, if you use steel, you may find that two are sufficient, where an aluminum shaft of the same length and diameter would require center support.

Here’s a situation where 1 was OK.](

Do you think that bushings could be used in this case? (One bushing on top, then roller bearing on bottom) Another team member thought that the bushing might cause too much friction. Do you think that’s true? We’ve never used bushings before.

The added friction that an improperly loaded bearing and improperly meshing gears would induce would likely be larger than using a bushing and bearing.

That might be somewhat application-dependent. I’d be somewhat hesitant to use a bushing on a very high-speed shaft, or a very high-load shaft, but if it’s properly sized I think you could easily make it work. Check the ratings before you buy (McMaster and Igus both have bushings, IIRC).

Bushings do have a little more friction than bearings, but their OD is significantly smaller than a bearing for the same size of shaft. Also, make sure to get a flanged bushing if you have something close to the plate as you’ll get a spacer/flat surface bearing.

The OTHER trick would be to use a UHMW or possibly Delrin plate instead of an aluminum one.

Our team is a little more lenient on the 2 bearings per shaft rule. We ran a swerve drive that is very similar to the one you have there and it didn’t need that top plate that houses your second bearing. You can see some pictures here in these threads, they aren’t the best views but hopefully, you get what you need. If you want specific pictures of the module dm me and I can help you out.

There are several factors to consider in geartrain design that typically trip people up.

  1. Torque is transmitted through a contact load at the gear mesh face.

  2. Gears push away from each other when they are transmitting torque.

You need to account for these two loads on each of the gears when you look at the restraints on the shaft. If you model the shaft system as a beam in a free body diagram and apply these loads to the beam, you will find that, in addition to the radial loads on the bearing there will also be an “tilting” moment applied to the bearing perpendicular to the axis of rotation.

Ball bearings can carry some amount of tilting moment, but, due to clearance between the shaft and the bearing bore, clearances between the balls and the races, and clearances between the outer race and the housing, the shaft will deflect off axis due to these loads. This will cause the mesh of the gears to tend to separate. This can cause the gears to skip and can also lead to excessive teeth wear and breakage. In my experience, this is a bigger problem with under-constrained shafts than the shaft bending.

Going back to the free body diagram, there are several factors that can make this situation worse:

  1. The amount of torque being transmitted - the loads at the contact point, will scale with torque, so in highly loaded systems, properly constraining the shaft is more critical.

  2. The distance between the bearing and the plane of the gear - the tilting moment in the bearing plane increases with the moment arm between the plane of the bearing and the plane of the gear. Note that this also will affect how much the gears deflect at the mesh - the further the gear is away from the bearing, the more deflection you will get for a small amount of rotation and the more likely you will be to have the gears start skipping teeth.

  3. Unbalanced loads - If all the gears on the geartrain are on the same side of the bearing, then the moments will add. If the gears straddle the bearing then the moment due to the loads of the gear on one side will tend to balance the moment due to the loads on the other side. Generally, they will not completely cancel out and in some cases, they may add depending on the location of the mesh point.

The easiest solution to these design problems is usually to add a second bearing on the shaft with enough distance between these bearings to react these tilting moments with radial loading in the bearing. However, if the geartrain is fairly lightly loaded and the gears are close to the bearing and have some balancing from forces on a gear on the other side of the bearing, then a single bearing may work.

For example, the output shaft on the versa planetary gearbox only has a single bearing. The planetary gearset inside the gearbox does act sort of as a secondary bearing because of the way that the radial forces are balanced in a planetary geartrain which helps to support the shaft. So generally, this does not cause any problems. But, I have seen some examples where this output shaft has a large gear and carries a lot of torque and does end up deflecting a lot.

I would be very cautious about having a shaft with only a single bearing. It can work, but in general it will lead to problems at the worst time. And by the time these problems reveal themselves, it will be difficult to re-design the geartrain to address it. It is far safer to design for the second bearing right from the beginning.

One possible solution to consider is a floating bearing plate that holds a second bearings for the two shafts for the meshing gears but is not attached at all to the main housing. This plate will hold the two shafts at a proper center distance to prevent the mesh from pushing apart and will carry the loads from the gear mesh into those two shafts rather than into the single bearing. I have seen this work successfully is some very heavily loaded applications.

Good points in your post, but this particular bit is not true. There are two bearings in the output shaft assembly, which you can see if you take one apart to change the shaft:

I’ve watched the students change the output shaft several times, and never noticed that. I had always assumed that Vex counted on the planetary gear stage to perform the role of the second bearing.

Thanks for the correction.

Given how close together those two bearings are, the issues associated with the clearances between the shaft and the inner bore and the bearing clearances themselves will still produce some level of off-axis deflection. This is probably why I still see issues with gear mesh with highly loaded large diameter spur gears on this output shaft (such as when you use a gear to drive the joint of a mechanism arm). I would still try to minimize the overhang of any gear on these versa planetary output shafts.

The greatest deflection torque comes when the transmitted torque is large and the gear is **small ** - and located away from the bearing.

T=rF   ==>   F=T/r

Thanks for this! We were just looking at using a VP to support a belt-driven intake wheel at the same offset as another wheel on a VP on Thursday. That’s what we thought was going on; great to have confirmation.

Thank you so much for the detailed response! Do you think there is there any advantage to using a floating bearing plate over what I have now?

Even with the two bearings, vexpro has a chart that discourages certain configurations (IE a long, unsupported shaft with no bearing support) and gear ratios that have high torque, my team had 3 retaining clips on the back of the output shaft of a planetary slip off the groove because we used a setup that was not recommended, it if you have this issue, we found if you dremel the groove out about a 32nd of an inch more, it fixes the problem, if you cannot reconfigure it