I would increase the ratio of your strafing wheel to give it more torque. It’s a single CIM moving around 150lbs at a single point - the 81% rule of thumb kind of falls apart at that power:weight ratio. Look up Ether’s spreadsheets/calculators on CD-Media whitepapers to figure out the gear ratio. You’ll want to evaluate whether it’s faster to turn 90 degrees them move straight versus attempt to strafe. For longer distances - definitely turn and go forward. For shorter distances (‘juke’ moves, or alignment maneuvers) of around 1-2 robot widths, there’s no need to gear that for speed. The good news is that it’s an easy gear ratio change to the internal components of the strafing gearbox.
Am I wrong in suggesting that one of the side drive gearboxes gets moved to the back half of the robot to balance the weight to allow for more linear strafing?
Good point, I wasn’t thinking about that aspect. The lowest speed option with the single speed double reduction is 10.39:1 for 7.25ft/s. Does that speed sound more effective?
Interesting point, but the reason I put both the gearboxes in one side of the chassis was because I thought I could just counter balance it with the battery in the other side. Also, it is difficult to power the shafts on the other side of the robot since the VersaDrop prevents mounting directly onto the side channels.
The center channel (with the strafe wheel) is easily movable to be mounted at a different spot, so if the CoM shifts with the upper structure, the entire channel can be moved to match the location.
I question how the geometry can be almost identical since we never actually CADded our version, but I’ll take your word for it.
Love the use of the pancake cylinders, I wish we could have used those on ours.
2 things:
If you’re planning to run the chain inside the tube between the gearbox and the drop module, you may want to consider how your gusset for the cross rail attaches to your side rails. (looking at you KohKohPuffs;) ) Trying to run a bolt through the tube there would really mess with your chain, and the rivets may not give you enough clearance either. My first thought for a solution would be to use two angle pieces on the sides instead of a gusset on the top.
As an offseason addition to our grasshopper drivetrain, we modified the center module to be able to actuate up and down. This actually reduced the drag and made the whole drivetrain much more smooth. In fact, after a match at our first offseason event, a friend from another team asked if we had added 2 more CIMs. Not a necessity, but we found it to be a big help.
I have never used omni-wheels in real life, but from a theoretically standpoint it does makes sense to either drop the strafing wheel or actuate it. The friction force you get from each wheel is limited by the traction of each wheel, which is dependent on the weight on each wheel. Left flat, the center wheel only gets 1/5 of the robot weight going sideways compared to the 4/5 of the weight going forwards. If you drop the wheel, actuate it with a piston, or put it on a spring, the center wheel will support more of the weight of the robot and (according to what I’ve heard) move smoother. Personally, I like the idea of actuating it with a piston or spring since the total weight supported by the center wheel would be easier to calculate. Again, let me emphasize that I have no experience and I am simply repeating what I’ve read. It sounds reasonable though.
Is there a reason to use the 2 CIM single speed double reduction gearbox when you only have 1 CIM? It looks like a Single Reduction Clamping Gearbox would work just as well, in addition to being smaller, lighter, and simpler (slightly more efficient because less stages! ). The lowest ratio on this gearbox would be 11:72, and Vexpro’s chart gives a speed of 11.47 ft/s with a 4’’ wheel. That speed is slower than your first post but faster than your second post. I suppose one drawback could be the lack of a ready-made encoder mount.
I completely forgot about the rivets.
Your suggestion sounds pretty good (with using angle gussets) but I also like the method you used on the 624 off-season drive with a block inside the channel. I might be able to do something similar here with blocks only in the middle channel, and bolting into them from the side channels.
The main reasoning behind using the double reduction gearbox was because the height of the wheel using a clamping gearbox was way too high to get enough traction going. The current design has a 1/8" drop and to get it there, I had to use a gearbox with an output shaft that could be mounted anywhere. The reason I used the particular gearbox was because I liked the long output shaft (compared to VersaPlanetaries) and it was also one of the cheaper options from Vex. The reduction options were nice too (for when I inevitably screw up :rolleyes:).
You can insert washers or other spacers at the gusset to lower the entire middle bar down 1/8". That might mess up mechanisms placed on top of the middle bar though, as well as make it more irritating to disassemble and reassemble with washers/spacers rolling around everywhere. I don’t think you would ever be required to remove the middle bar, but it’s still a disadvantage.
Cool design! We are planning to use the versadrop modules as well. Can I ask, how did you go about attaching the surgical tubing to retract the wheels?
When we did our h drive in recycle rush, we found the adjustment of down force on the center wheel critical to dial it in. I think another drop module connected to a dedicated pressure regulator would make for quick adjustments and reduce concern over what drop to use for good performance.
First off, thank you for posting the native SolidWorks Files in your GrabCAD. That makes life so much easier.
Looking at your model, the back Colsons are raised up by 1/8". I’m wondering if those wheels are needed at all, considering how little they’ll actually be contacting the floor. For the sake of simplicity, it might be good to just drop that back set all together and push the Omni’s back to where the Colsons were. Also, I would make a slight change to how your drive rails are connected to your front and back framing rails. Since you have a chain-in-tube aspect, having the fronts and backs of your drive rails open makes life MUCH easier if you have to make any changes or fixes with your chain.
Don’t forget that you’ll need bumper mounts; per 2016 rules, you couldn’t have spans of unsupported bumpers over 8", and right now you’ve got a span of about 10" where your drop modules are.
Also, +1 to everyone who’s mentioned actuating or other methods of transferring the weight of the robot to your middle wheel when strafing. My team built a standard H drive during prototyping for 2015, and it was pretty useless. Part of it was our gear ratio (we used the same ratio for all 5 wheels), part of it was how uneven the floor of our shop is, and part of it was just that H drives need some tuning to make that middle wheel worth it. In my opinion, it’s worth to either have 2 center wheels, each with their own CIM powering it, or put two CIMs in the gearbox you have. Either way, figuring out a way to actually get the weight onto the strafing wheels is crucial for effective strafing.
That defeats the whole purpose of the drive. Without four traction wheels, there wouldn’t be a good pushing mode that resists turning. It would just change from all Omni to 2 traction 2 Omni, which isn’t much of an improvement.
But are the back Colsons actually providing traction when driving forwards? Looking at the model, it seems that the Colsons wouldn’t have enough contact on the carpet to make a difference, but I’m not sure how much the wheels would sink into the carpet.
When the drop Colson is engaged, the entire chassis tips backwards onto the back Colson, or at least that’s the idea behind 624’s system. It uses the chassis geometry to avoid adding 2 more drop modules.
). The lowest ratio on this gearbox would be 11:72, and Vexpro’s chart gives a speed of 11.47 ft/s with a 4’’ wheel. That speed is slower than your first post but faster than your second post. I suppose one drawback could be the lack of a ready-made encoder mount.