scout3 actuator updates: let's try brainstorming for the front section of this box [updates]
So here's what we did last time: https://www.adim.io/post/chip3-actuator-updates-let-s-choose-the-right-filament-to-print-with-and-tolerance-updates basically looked at bearings and came up, again, with more questions than answers. But we did pick a filament we want to print out of.
Today we want to start brainstorming what to do with the front section of the gearbox. Because that's likely the most important panel in the system. Let's talk about the requirements for that section because there's two really big ones.
To clarify, we're talking about the Output Disk of the gearbox opposite the motor side. There are a lot of assumptions we made on the loads inside the planetary gearbox that only hold true if the Output Disk can do the following things:
The output disk is responsible for supporting moment loads reflected to the gearbox from components of the robot.
It's also responsible for supporting any thrust/axial force components supported (as well as radial components).
It also needs to spin efficiently because it is an output gear system.
So basically, the output disk needs to be one large crossed-roller bearing, but here's the catch. Without being a crossed roller bearing because those are too expensive...
So maybe we can look at something like this: https://www.vxb.com/120mm-Lazy-Susan-Aluminum-Bearing-Turntable-p/Kit12876.htm?gclid=Cj0KCQjwjer4BRCZARIsABK4QeXc__ULITaPqYnrklguMjRU2IjVgpWnEz6nVoVOCL6F4LH_sdzTLJgaAlJREALw_wcB ? This is a lazy Susan bearing with a 120mm OD and 70mm ID. There's also this one:
https://www.vxb.com/140mm-Lazy-Susan-Aluminum-Bearing-Turntable-p/kit12877.htm this one is 140mm OD. And it's 10 bucks but can we get this to work. The outer radius of our system is 144mm. So we need a drawing of this thing to see how we'd make it work.
Here's some more info: https://www.silverthin.com/bearings/lazy-susan/sl/detail/ASL140-085/
The outer BOLT circle diameter is 128mm ours is 131mm at the moment. There's a possibility this could be the perfect bearing for us. But we need to look at how a lazy Susan bearing works and what it could be rated for. The inner bolt circle diameter is 101mm. This is really the perfect size. Maybe if we can shave off 4mm somewhere. We can match this system exactly.
This bearing has a 120lbs thrust capacity.
Okay so turns out the bearing info above is NOT for the KIT12877 it's for that bearing's competitor. So let's go find info on the actual bearing. For some reason, for that bearing, all we can find is:
The question is really, would something like this work for what we need it to do? We don't know let's find out. This bearing system would cost 20 dollars though as opposed to the other bearing system which would cost like 700!
So here's what we need from VXB:
The thrust capacity of the bearing/the ratings of the bearing - 95 lbs
Is this a two part construction or one part construction? - One part construction so it doesn't come apart!
What is the bolt hole spacing diameters and what is the size of the bolt holes?
Maybe we should just call them... wow they're so helpful and they're emailing us a drawing!
Okay so the outer hole diameter is 128mm, and the inner hole diameter is 120mm. And the holes themselves are, I think 5mm diameter from what the drawing is showing so that's an M5 bolt which I think is what we already have, and that's a 9mm diameter counter-sink. So we're going to email them back and ask about what the radial/moment capacity ratings of these bearings are - even if it's just an estimate. But let's make an estimate of our own.
Let's say this thing has a 95lbs thrust capacity and we put two back-to-back and this happens at 120mm diameter. That's 422 Newtons at a diameter of 120mm so let's say 50Nm if we put two back-to-back?
Now let's leave the thrust capacity even with two at 95lbs because we're putting the beatings back-to-back. We don't have radial data yet.
Now what we want to do is compare the 50Nm that this holds to what a versa planetary can hold because we know the versa planetary system is working right now on the prototype robot. So this thing: https://www.vexrobotics.com/versaplanetary.html#description if this can hold less than 50Nm then we're ok!
If I had to guess, if we look at the price of the bearings we can likely assume that the bearings are ball bearings. Because they're too cheap to be rollers or angular contact bearings but I don't know: https://content.vexrobotics.com/vexpro/Falcon/VP-LoadRatings-20191031.pdf and in this whole doc there really isn't anything about this either. Okay so back to doing actual math.
If we assume the same geometry as the current robot's leg, and we assume there is no friction (for now), and we assume that the force on each leg (vertical) peaks at around 125N if the robot is 250N in weight. Then the max moment the gearbox would feel is 125N * 0.17m because that's where the foot is from the gearbox face, and that's 22Nm. So this might very well work. Now that doesn't take into account things like friction and etc but straight calculations that's what we have.
So that means the best I can do right now is: this might be fine? We need to wait for the VXB people to get back to us on what the moment rating for this bearing is and what the radial force rating is. That might be the big problem here since this is a thrust bearing. That radial force it needs to withstand of like 125N.
In the meantime let's make a nice CAD file for this bearing just for kicks. Okay let's be real, there's no way this is right. The inner hole diameter cannot be 120mm.
So let's email Vivian back and see what's going on here... I think the 120mm might be the distance to the division line between the two bearings.
In the meantime this is what we know we have because we know the outer diameter is 140, the inner is 89, the thickness is 10, and we think the outer holes are at 128mm. That seems reasonable. I also think since: https://www.silverthin.com/bearings/lazy-susan/ this website says they're competitors and they're directly referencing bearings here. We can say that the dimensions should mostly be the same at least for holes and things. So I think we can actually update this cad a little. Now the only thing left to figure out is where does it break like where is the seam for the bearing. There is a marking at 120mm on the other drawing I think we should go with that for now.
Okay so drawing the 120mm - the 120mm seems a little too big let's look at some pictures of the bearing to see if we can't see.
So if we look at this picture. It seems like there is a 2mm gap in-between both parts of the bearing and that the holes are centered. So let's try that until they get back to us.
So that set of dimensions seems to make a lot of sense that leaves us with a bearing that would look something like the their picture above. Now we actually got another response from the bearing company but nothing to do with the drawing for now.
So 22lbsf before bearing locks, that's 97.86087582151981 N before the bearing locks. So if we use one this wouldn't work. But if we use two bearings. Then if we load it the right way we should be able to have the system withstand around 200N of radial force. And given we're estimating 125N of radial force from the leg of the robot. I think that's within reasonable bounds. Now we're just waiting for McMaster to get back to us about the bearings and this is what they said:
So we're waiting on that now too! But here's the thing. I think we actually may have a solution to the problem of our front-plate! We need to re-design the rest of the gearbox but that's okay because we would've had to do that anyways. While we're at it we also want to shrink the gear sizes themselves because we don't think we'll need gears that are like 20mm in thickness. Or we'll design first for the 20mm gears by maybe drop them down to a happy medium of 15mm. That seems like a thick enough gear but what will really verify this is FEA on the gears themselves. We want to make the system well packaged! But we're not worrying about the gearbox today. Today our job is to deal with the very front piece so back to this drawing:
So now we want to actually CAD that space, and that planetary carrier for Stage 2. So let's decide how much thickness we want to put in-between the two bearings - I'm thinking like 5mm. And then get to designing that part in CAD.
So there's three total parts in the assembly. There's the actual two bearings. There's the spacer between the outer rings, and then there's in inner carrier which is shown here but not necessarily finished because we would need to add the things that actually make it a planetary carrier on one side. But I want to FEA that middle part to get some intuition on how plastic would respond to certain loads and where we should put things like the mounting points for the leg. Should we put that on the bearing or should we put that on the carrier face itself.
So the first thing we did was use the bolt holes themselves as the points where the leg would be anchored to the body. Then we places a 90Nm torque on the other face. Now this design does assume that the bearing will take the rest of the loads but - that's kind of what the bearing is for. For now, we'll do a more accurate FEA with the moment loads we are expecting later.
So the minimum factor of safety for Nylon 6 here is 5.694. What we're going to do now is do a more interesting moment load.
So this is what we're trying next with the bolt holes on the sides as anchor points. Now this study isn't strictly accurate because the bearing would be taking a lot of this force if not all. of it. The unit in the middle here would really be shielded if we mounted the leg to the bearings. So I think that's what we want to do - mount the leg to the holes on the bearing and not the face of the carrier. Because then the bearings will take the moment load and the carrier won't have to. Here's some really inaccurate FEA just for fun:
We want to do a much more accurate FEA than this at some point and we shall but first let's outline the plan of assembly of this thing.
I feel like this is a fairly decent plan of action for how to make this front part of the gearbox. So the next steps are:
Lay out the design for the middle and back parts of the gearbox including the motor mount and the bearings and shafts and assemblies.
Do part-by-part and assembly FEA to make sure everything is going to pass the loads we're asking them to sustain even though the hand-calcs tell us a lot already.
Finally, we need to figure out tolerances and sizes of holes for press fits, slip fits, how to size the parts for 3D printing (increasing by 100.31% or whatever we need for Alloy 910). Bearing fits and etc.
So the last thing we want to do today is a slightly interesting calculation and that's this:
What are the forces on the bearing under the actual loading conditions if we went with this design?
So yes, not just an interesting question but a CRITICAL one. And here's the math and the verification for that. This is going to take a little bit of structure analysis that I don't really know how to do but we're going to go with what we know and see where we get:
So let's just say the moment calculation is a little worrying but that being said we can design the robot differently than how it's currently designed. This is for scout2 right? So for scout3 we can design it so that it passes the boundaries of this bearing. I also think that since the bearings are literally ten bucks we can get some and TRY IT. Just BREAK IT and see how much force it takes... I mean really! They're so cheap that it barely matters.
This does concern me though. It concerns me because if we include the moments the actuator will be feeling due to friction and the feet moving left and right, we might have a few more issues. I don't know - if I'm being entirely honest. So that's really why I want to get a few bearings and test this plan out. I mean this bearing was built for a lazy suzan, but it might just surprise us. Can confirm the radial and axial loads on the box will be okay with a safety factor. Another note on the moment load - the 95 lbs assumes that the ring distributes the force evenly across all the balls. I have no idea how they rated/tested this bearing so we really need to get one and see. So this isn't a solution but it is promising. We might even be able to send this design to the people who make the bearings and ask them if they think it would work. *shrug*
To clarify: at the moment we're only worried about the moment loads to these bearings (that's funny). It seems to pass everything else.