We put chain shields on the inside of our drive rails to prevent chain from flying into things I’d rather them not… making access to the inside of the drive axle more of a pain than just rotating the wheel to the shaft collar. In a perfect world I’d have a lathe and we’d just use clips…

1 Like

I believe you’re thinking of the 971 2014 robot, Mammoth. The 2016 971 robot did not have a hexagonal frame.

2 Likes

Well, neither did Mammoth…

2 Likes

You’re right, I’m forgetting two sides.

You’re right, I was thinking of Mammoth. I was actually just watching a video about it yesterday too…

How are you defining “stability”? This is one of those words like “stronger” that has several different meanings, and the solutions for each meaning are different. If you mean to suggest that a wheelbase like this makes you less likely to tip, it won’t do anything to help with that.

As for turning, in theory an omniwheel has basically no sideways friction, so it should not make a difference where they are. In practice they do scrub a little bit - making your wheelbase narrower at the ends would then make it a little harder to turn. This shouldn’t make a big difference because drivetrains with omnis in this configuration with a “normal” wheelbase have no problem turning anyway.

I guess when I mean “Stability”, i am referring to the drive trains ability not to studer, jump, rock side to side, and smoothness of operation. I understand most of that pertains to drive code, but without a stable bed, code does nothing.

I also know that there is a general idea of equivalent exchange. That to gain something, a person must give something of equal value. I believe the exchange here is lesser ability of smooth turning for less “unwanted movement”. I am trying to get around that via the curve of the wheel placement and both a drop center and omnis together

I don’t see why this would stutter, jump, or rock less than a robot with its wheels in a line.

If you want the robot to stutter less, you need to make sure you have a wide enough wheelbase relative to your track length (so the robot turns nicely). The rule of thumb is track length/wheelbase > 1. You also want to make sure that your drivetrain has enough power to be able to accelerate without browning out. The biggest thing you can do to decrease stuttering is good code and tons of driver practice, though.

If you want the robot to jump less, you’ll have to either add some kind of shock absorber or add mass (so any upward force results in less upward acceleration). Adding wheels out of plane won’t do anything to decrease how “jumpy” your robot is.

To decrease side-to-side tilting, you need to increase your track length or lower your CoM. The overall wheelbase is the same width (the CoM is the same distance from the pivot point when tilting sideways) so it should take just as much force to tip the robot sideways. If you move wheels further out from the robot center, your pivot length will increase so it will be harder to tip sideways.

All that being said, there’s nothing wrong with this drivetrain per se. But you likely won’t be getting the benefits you’re looking for from it.

If you are looking to eliminate the rocking fore and aft, the best thing you could do is eliminate the drop center such that the wheels furthest from the center of the robot are always in contact with the ground. As other have said elsewhere, the drop center is used to eliminate the scrubbing of the corner wheels during turning. The omnis have negligible scrubbing and therefore you should not need a drop center if you have omnis on all 4 corners.

Stuttering is usually a response to scrubbing (the wheels skip or stutter to relieve the scrubbing loads). So, with the omnis in all 4 corners, you should be pretty good for stuttering. Your traction wheels are fairly close together giving you an effective tank drive length < width which should be good for low scrubbing during the turns. The one thing you could do to fully eliminate that scrubbing would be to go with a 6WD with 4 omnis (in the corners) and only 2 traction wheels (one on each side) again with no drop center.

Side to side rocking is usually a response to stuttering when turning in place. Again, if you fix the stuttering, you should fix the side to side rocking motion.

If, when you move the corner wheels in closer to center, you used that as an opportunity to move the center wheels further out away from the center (wider track) and still respect the total frame perimeter, it should help all these issues. But otherwise, I would leave the corner wheels with as wide as possible to gain as much tipping margin as possible.

1 Like

So, I have started to turn out some the suggestions on this. I first made the bumpers. Not hexagonal, I know. T-boning is still possible with this set, but at least it protects the frame (had to add L angle to the gaps with the outside wheels to make it legal to span bumpers across it).

Next thing to do, get rid of drop center and bring the omnis closer to the corners (This is a long running debate I have with my mentors. I believe designing a drive train to have a drop center instead of having everything on the same line and using omnis is a waste of time. My team mentors are looking at it from a view where the less omnis we have, the more traction we have, and the less wear on the wheels. At least with this design there is still 4 traction wheels on the ground at all times, which I might be able to sell to the main mentor I discuss this debate with(neither of these mentors are mechanical engineers, one is a civil engineer and the main mentor I discuss this with is the programming mentor, I can probably sell to him that no rocking from a drop center means easier time programming auto!)). I can probably move them another inch before the bearing block hits a gusset plate, which I do not want to redraw to be shorter, as I like the gusset plate on this design

Last thing, try to get some chain to ride in the tube. I am going to have to find the correct size sprocket. If anyone knows what sprockets for 25 and #35 ride nice in 2x1 tube, that would be great, especially if they attach to 1/2 inch hex.

also need to design proper mounting for bumpers, but that’s a task for later.

Next, bellypan and electronics (which is about how far I go before I consider it designing an entire bot and not just the drivetrain.)

1 Like

Speaking as a team that ran 1 traction + 2 omni per side this year, I would not advise doing so, especially in a defense-heavy game like 2019 was. Right out of the gate our robot would drift if cornering at high speed, and after some matches we were getting shoved sideways effortlessly by defense. The issue with omni corners and no drop center is that as soon as you get a fraction of a millimeter of wear on the center wheel (which happened fairly quickly as we were using a softer tread than the omnis), most of the robot weight was on the omnis instead of the center and we may as well have been an H-drive. Drop center is very much not a waste of time—it will improve your traction by a good deal especially with wear and tear.

@ijensen

Adding onto this, be aware that 4" Colson Performs wheels are about 3.94" in diameter. You appear to be using versa wheels so that isn’t an issue. If you do switch to Colsons try to measure a few first so you have enough wear room.

1 Like

WCP

sells #25 chain in tube sprockets w/ 1/2" Hex Bore. I don’t think the math checks out for #35 chain tube if you are using 2x1 tubing. Someone correct me if I’m wrong.

AndyMark also has a similar #25 chain-in-tube double sprocket. I don’t have any personal experience with it, but I haven’t heard any negative reviews.

#35 is a bit harder. We used metric tubing with ours (to go along with our metric robot and metric country), so it wasn’t a huge problem. If you’re using inch sizes though, it’s a bit harder. The Vex 12t #35 sprocket is listed as having an outer diameter with chain of 1.786". If you’re using 1/16" tubing, that leaves you with 0.089" of clearance (without any drop center). That’s a pretty tight fit. You’d probably be better off going with 2.5x1" tube so you have room for manufacturing tolerances and don’t have to worry about the chain rubbing.

1 traction wheel and 2 ominis is still a fair bit less traction than 4 traction wheels and 2 omnis

In their driving direction, Vex’s omni wheels have approximately the same coefficient of friction (about 1.1) as a traction wheel, on a standard FRC carpet.

The important part isn’t the driving direction, it’s that omnis have effectively no coefficient of friction sideways. It’ll be fine if you’re in a pushing match forward, but if you get T-boned you’re going to slide like crazy, especially after tread wear starts to happen.

The world is not perfect. The floor of the competition fields are not guaranteed to be flat. In some games, the floor is guaranteed not to be flat. With no drop, your traction wheels can end up with no traction in certain locations. You may want to read about the base of the Pyramids in the 2013 game (see page 19).

You many want to thoroughly analyze your auto routines to find the root cause of why they are not performing to your expectations. Often, the reason the auto routine is not working is something other than the rocking from the drop center. Many of the best teams use drop center drivetrains and are able to achieve good performance in their auto routines.

Somewhat off-topic here, but if you use chain in tube in a 2x1, do you not have issues with trying to secure mechanisms to the side rails? It seems to me that if there’s chain running right next to a wall of a 2x1, you can’t rivet or bolt into those sections of the 2x1 without hitting the chain.

You can be a bit crafty with your chain runs and fastener placement. I’ve seen a few robots that seemed to have a strange rivet pattern where 1/3 of the length of the belly pan had rivets biased towards the outside of the robot and 2/3 towards the inside. It’s apparently still a tight fit, but I don’t see why it couldn’t work. Fastening to the vertical face of the tube is also possible as long as you’re aware of exactly where the chain is.

1 Like