pic: Unobtainium 6wd - Cantilevered, Dead-Axle, Slot-Tensioned Drivetrain

CD-Media: Photos
, , ,
1

e7baed3f05d4c16395421f1fdb4866a7_l.jpg
e7baed3f05d4c16395421f1fdb4866a7_l.jpg1280×720 140 KB


No, I didn’t lose a bet… at least not yet :stuck_out_tongue:

Our lead mechanical mentor, Mr. Stehlik, in an effort to devise an easy to build drivetrain, half-jokingly sketched out a concept to me:

Fundamentally it was a 3/8" shoulder bolt screwed into a piece of 1/4" aluminum bar with two nice big washers sandwiching the bar.

“There’s your dead axle.”

I smirked at first, then did a double take. I didn’t think something so simple would be strong enough to support a robot’s weight in motion. But then I thought: “put these shoulder bolts in front of me, and I bet I could do chin-ups on them…”

I thought Mr. Stehlik was on to something…

I still don’t know for sure if this exact setup will be strong enough, but I’ll leave that for someone much smarter than I to figure out.

I’ve been involved in FIRST since 1995 and this is the first drivetrain I’ve tried to design. I am an electronics, programming mentor first, and I occasionally throw my two cents in when it comes to strategy. Mechanical work is not my thing, but I’m glad I went through something like this. It gives me a new appreciation for all you crazy FRC folk who post drivetrains here on CD every other day (Kajeevan?).

It also has renewed my belief that every semi-serious FRC student should design at least ONE simple drivetrain by the time they are in grade 11. Superstar students should have already designed multiple by that point.

Technical Specifications:

  • 6x 4" AndyMark Plaction Wheels
  • 4 CIM, 2 per CIMpleBox (stock)
  • 22 tooth to 36 tooth final gear ratio
  • 25 chain driven, 3 chains per side
  • Cantilevered, dead axle design
  • Front and back wheels have 1 sprocket bolted directly on
  • Middle wheels have 2 sprockets bolted directly on
  • Front and back wheel axles mounted in slots for chain tensioning
  • Middle wheel axles are not in slots, and dropped 1/8"
  • CIMpleBoxes mounted in slots for chain tensioning
  • 80/20 extrusion attached to angle with 3 bolts at each corner (1 axial, 2 from underneath via t-nuts)

An exploded view and a close-up of the shaft assembly are posted as well.

2

Great! What are the drive’s advantages, in your opinion?

3

I do have to ask: Wouldn’t an 1/8 wall square extrusion weigh just as much as a quarter inch L, with a much stronger profile?

4

Is the idea here that the bolts are threaded into the frame member or are they through-bolted with a nut on each side?

My gut reaction is one of concern that the bolts will unscrew themselves, but maybe that’s unwarranted.

Edit: Nevermind; missed that this is a shoulder bolt.

5

This is very nice and simple but I’d be worried about the forces that cantilevered load is putting on a very small amount of aluminum. Id guess that over time either the hole in the aluminum would oval out or the angle itself would want to bend below 90 degrees. You may want to run some CAD analysis, but with me this doesn’t pass the sniff test.

6

The shoulder joint will definitely hold, but it has to be supported by the frame. With a single support point in just 1/4" of aluminum, it will likely just bend the angle aluminum and not stay horizontal. A box extrusion will solve this problem by providing two support points for the bolt and not deforming under the loads we’re talking about.

7

I’m pretty sure this would hold, over time you would probably see the slots warp a bit after repeated use. My main concern would be the flex in the main angle bar. The middle wheel would likely flex a decent amount when it had weight on it, if there wasn’t any extra cross support in the middle of that system. I could see it flex enough in this configuration that it could almost negate a 1/8" center drop. But that flexibility could be interesting for a mecanum drive…

8

Interesting design, it might work for some games, but not for others.

We built a cantilevered dead axle drivetrain on our 2008 robot, using 1/2" threaded rod for the axles, and 1/8" thick 1x4" pultruded fiberglass for the frame members. We had an adjustable chain tensioner instead of slotted holes. It worked fine for the Overdrive game, since the playing field was flat and most of the action was driving laps.

http://www.chiefdelphi.com/media/photos/29697

9

What are the drive’s advantages, in your opinion?

  • Simple base components (80/20, L-angle, shoulder bolts)

  • Reduced machining operations (no custom shafts, non-CNC milling)

  • Simple chain tensioning method

  • Quick assembly time

  • Easy servicing (one bolt wheel removal)

  • Lightweight

  • Large baseboard area for electronics

  • Less gear reduction required for 4" wheels (more efficient)

  • I could see teams with some basic machining capabilities knock two of these out within the first few days a build season, then move on to bigger and better things

Wouldn’t an 1/8 wall square extrusion weigh just as much as a quarter inch L, with a much stronger profile?

Yes, but the shoulder bolt shaft needs to be bolted to a plate with sufficient thickness, or else the plate will bend under load. Anecdotally, 1/8" wall seems too thin and will flex. Also, the L-angle leaves more space for the electronics board, and provides nice mounting flange. The L-angle looks to be plenty strong(?), and the bottom flange could be milled down to 1" or less to save weight - but that’s another machining operation.

Is the idea here that the bolts are threaded into the frame member or are they through-bolted with a nut on each side?

It is a shoulder bolt (through-bolted) with a single nylock nut, and two washers sandwiching the plate. The above image best shows the setup. When the shoulder bolt and nut are tightened, it causes the washers to squeeze the plate tightly. The idea is that this is enough squeeze force to a) keep the cantilevered shaft perpendicular, and b) keep the shaft from sliding in the slot (the plate is actually SLOTTED for the front and back wheels to allow you to slide the wheels back and forth to tension the chain)

My gut reaction is one of concern that the bolts will unscrew themselves, but maybe that’s unwarranted.

This is where my lack of mechanical experience shows through. I am concerned about this too, but I don’t know the limitations of Nylock nuts, and threadlocker.

This is very nice and simple but I’d be worried about the forces that cantilevered load is putting on a very small amount of aluminum. Id guess that over time either the hole in the aluminum would oval out or the angle itself would want to bend below 90 degrees. You may want to run some CAD analysis, but with me this doesn’t pass the sniff test.

The concept is that the washers that sandwich the plate distribute the load over a larger area of the plate. If this happens, there should be very little load inside the slot itself. CAD analysis is a good suggestion to see whether this works.

10

You might want to try heavy-duty all-metal locknuts (the ones with the elastically-deformable bore), or high-strength steel nuts and threadlocker. Torque them to the limit specified by the manufacturer. Given the shoulder bolts, you’ll probably need an unusually-strong Allen wrench to make this happen. That’s the only way I could see this holding tightly enough in a slotted hole.

Consider http://www.mcmaster.com/#92501A030 or http://www.mcmaster.com/#93591A200, for example.

11

Stover lock nuts, you mean?

I suggest you might want to find or figure out how to make a more substantial washer. Our design used large hex flanged nuts. You might be able to do this too, and use 1/2" fine thread threaded rod for the axle as we did. We used two jam nuts to retain the wheels, but this required that whoever was working on the robot needed to know not to tighten the nut too much!

12

Actually, my original concept called for 1/2" shoulder bolts in a slightly different configuration. I intended to have the shoulder portion of the bolt pass through the frame, and tighten the entire wheel assembly against the frame with very thick washers.

The thread of a 3/8" shoulder bolt is 5/16"-18, which in my opinion is too small to be supporting this kind of cantilevered load. Thicker washers would help some, but the thread length is only 1/2", so you don’t have much to work with, probably not even enough for a heavy lock nut.

I think what you have shown could work with a larger shoulder bolt, but it looks like squirrel’s method with 1/2" threaded rod would be simpler and stronger. And fine thread really helps to get things tight.

Another issue is bending of the frame itself. Putting a cross bar at the top of the angle next to each wheel would help a lot. It could be a thin wall tube with a small threaded rod running all the way through.

13

There is a lot there to love.

A lot that is cringe worthy as well, but overall, a nice effort.

Got me thinking… …How can I use this? Maybe there is something there…

Thanks!

Joe J.

14

Cantilevered shoulder bolts can take a lot of load. I used them in an FSAE car suspension rocker that saw around 5,000lbf. The trick is to get the shoulder supported and eliminate bending stresses from any threaded portions. The only serious issue I can see in your design is how the shoulder bolts are connected to the rail.

In this picture near the left-hand edge about 1/3 of the way down you can see where we used a 5/8" (1/2"?) shoulder bolt on the suspension rocker. About half of its length passed through a steel mounting point, then went through a needle bearing in the rocker, and was captured with a nut on the inside of the car. We sandwiched thrust bearings on either side of the rocker so we could clamp everything down very tightly yet still spin it by hand.

I hope you find this useful cough needle bearings in wheels cough :wink:

15

Now that I think a little more about this, watch out for the stress distribution in the preloaded part of the bolt (i.e. the part from the threads to the shoulder that is under tension when tightened). Normally you preload a bolt to a signficant fraction of its tensile strength. The large bending moment from the cantilever will also add stress right around that area—meaning that you might end up overstressing a bolt tightened according to the manufacturer’s directions.

16

For those of us that use the various drivetrain designs posted on CD as learning tools, what do you find cringe worthy? And why?

Thanks.

17

Let me start by saying that I am home sick today and probably should not have used the term cringe worthy. I really don’t want to pick on the design. There is a lot to like.

Here is the thing. For those who have been doing this for a long time, design is an emotional experience.

I believe that we humans utilize our emotional brains to sort through complex space of possible solutions. Our emotional brains are just really good at searching through complex system interactions.

So… …When I see something obviously awesome, I often laugh (I recall the first time I looked up in the GA Dome I laughed and laughed as I realized how cleverly the designers used hoop strength and tension members together with compression columns to raise the higher and higher tent poles over that beautifully clear interior). When I see something not so awesome, I feel it in my stomach.

So… …what do I think is less than awesome in the design?

I think it boils down to two conflicting functions of the 1/4 angle. When I first see it, it seems both not strong enough and too strong at the same time.

It doesn’t seem strong enough to be the sole mount for the cantilever axles. While at the same time the 80-20 and 1/4 angle seem pretty heavy for what they are doing.

The thing about emotional brains is that they can be wrong. We educate our emotional brains the best we can but they sometimes point us in the wrong direction.

I would have to do the calculations to know (and I haven’t).

So… …that is probably the numb of what I was getting at (that and the connection between them draws my eye as well).

Let me close by saying I was too harsh. There is a lot to like and a lot that I would change. But a very interesting effort.

Joe J.

18

Thank you Joe for your comments.

I agree with you about the 1/4" angle. I see teams driving down to thinner and thinner metals so to see something that thick makes me wonder.

Last year we accidentally ran our .090" 5052-H32 sheet metal frame into a cement column at full speed without bumpers. We ended up with a little ding on the flange, but the structural integrity remained. Made me think we could have gone to .080 or maybe even less.

I always like to know the tradeoffs on various drivetrain designs so I can help our students understand the tradeoffs and make decisions according to what we need.

I am also constantly looking for designs that use less precision machining because I see a lot of teams without that capability. That is one thing interesting about this design.

19

I saw the CAD of 1726’s 2008 robot way back in the day, but back then I didn’t realize it was 1/8" C-channel for the side rails. I thought it was box, and the cantilevered shafts were supported at two points. Missed that one big time.

And here I thought this was an original idea :D.

The flanged Stover lock nuts look pretty nice, and I think they would go a long way to improving the design. Placing one on either side of the plate instead of the washers, upsizing to a 1/2" shaft of fine threaded rod and two jam nuts to keep the wheels from falling off seems to be the direction I’d like to go in… and then an FEA.

Questions:
Did your wheel bearings rest on the threads of your threaded rod? If so, any concerns with this? Excessive play? Still worth the trade-off IMO, but curious nonetheless.

Did you have frame flex issues at competition weight? Did you have to add more support to your C-channel side rails to prevent them from twisting, resulting in negative camber?

Was that a typo, or was your fibreglass C-channel really only 1/8" thick?

Thanks all who posted in this thread who shared their comments and concerns (yes, that includes you Dr. Joe!).

Really, as a programming and electronics guy, if I can’t get the mechanical guys to cringe at least once or twice a day, I’m not doing my job right ;).

I believe that we humans utilize our emotional brains to sort through complex space of possible solutions. Our emotional brains are just really good at searching through complex system interactions.

P.S. If you believe in the quote above as much as I do, you need to read Malcolm Gladwell’s book “Blink.” Not just a blatant plug for a fellow Canadian, but his work really expounds on why some people are so good at what they do. There is so much more behind the “emotion” and “gut feelings” we get when working in our fields of expertise. The vast amount of experience and expertise we subconsciously put into honing those “feelings” allows us to correctly decipher really complex problems in the blink of en eye. Often without even realizing the vast amount of information we actually considered in the process.

20

It’s not a tight fit, there’s a bit of slop, but it’s only a few thousandths of an inch. And the reduced load bearing area of the threaded rod vs a solid shoulder bolt means that after decades of use, it might wear the threads enough that you’d want to think about replacing the axles. It’s really not an issue at all. I’ve worked on enough old cars to have a pretty good feel for how long something like this will last.

Did you have frame flex issues at competition weight? Did you have to add more support to your C-channel side rails to prevent them from twisting, resulting in negative camber?

The frame is surprisingly stout for how little there is to it. The 1/16" aluminum bellypan and top corner gussets really make a strong structure when combined with the channel. Did I mention it’s light? Notice that there really is no material where it doesn’t need to be.

Was that a typo, or was your fibreglass C-channel really only 1/8" thick?

Yeah, only 1/8" thick, that’s plenty strong. As far as weight, the fiberglass has roughly half the density of aluminum.

Again, this robot was designed for a specific game, and worked well for that game. The 2009 robot we made of wood was even more unconventional, and also worked very well for the game it had to play. The steel robot we made the following year didn’t work as well, it had cantilevered axles, but because of the bumps it had to traverse, the axles bent significantly during the competitions.