I am working on a Drivetrain CAD Project and I am debating using A direct drive system with 6 Falcons with planetary gearboxes Connected directly to the wheels. However, almost no one ever seems to do this instead using belts or chains to get power to all 6 wheels. I just was curious as to why direct drive is not very popular among West Coast drivetrain designs.
- We only need 4 motors (and PDP slots) for a standard WCD
- We know that all 3 wheels on a side are turning at the exact same speed
- We have not had good experiences with planetary gearboxes under high load.
- A flipped motor WCD makes great use of typically unused space.
Aside from the extra weight and efficiencies*, the major problem is this:
If you have a drop center (or you hit a bump, etc.), what happens to the wheel and motor that is off the ground? It’ll spin, and the motor doesn’t contribute anything to motive force. If the wheels were chained together, though, all motor power could be redirected to the wheels that are on the ground.
Plus, when that wheel does get back on the ground, it’ll probably leave a nice skid mark as it’s been revved up. I can’t imagine that producing good behavior…
If you’re doing autonomous routines, which wheel do you use for odometry?
*Planetary gearboxes are generally less efficient than traditional spur boxes. This is at least partly owed to the fact that their gears usually don’t even spin on bushings much less bearings- it’s metal-on-metal contact.
Agreed. There isn’t one solid reason the proposal is a bad idea, but rather the sum of small factors (weight, packaging, control) that likely make it less of a good idea.
OP, what planetary are you considering? There are beasts like this banebots gearbox that are more than enough— way too much, in fact. I wouldn’t trust VersaPlanetary level durability on my drivetrain.
That can be done, although personally I would use Vex Clamping gearboxes rather than planetaries. Saves some money there:Single Reduction Clamping Gearbox - VEX Robotics. I also wouldn’t typically trust a versaplanetary in this application (maybe others would?). An Andymark sport gearbox could probably do it, but that cost adds up fast.
The big down side to it is the reduction of power. If you run a drop center wcd (as you certainly should) only 4 wheels tough the ground at a time. That means that only 4 falcons are helping you at any given time.
Also, this opens up the possibility for 1 wheel to “spin out” while another wheel maintains friction. This means that for portions of the match where this is happening (turning is a big example) you have less than 4 falcons giving you power.
Hope that makes sense. It would be functional, but running belts/chains for maximum power is generally the recommended way to go.
A couple of nitpicks. Direct drive implies a 1 to 1 ratio. Once you introduce a gear box it is no longer direct drive. While FRC style planetary gearboxes might be less efficient than FRC spur gear boxes, this in not inherently true.
As others have noted. The point of the six wheel drive is to lower the effective wheel base so turning is easier. So you always run drop center. One set of outboard wheels are always either unloaded or lightly loaded.
Besides the problems already outlined above, there’s the synchronization problem. Even with the very good controllers on the Falcons you’ll have some trouble getting them to operate at exactly the same speed all the time on each side of the robot. By far the simpler solution is to just use the belt/chain system to a single gearbox.
I will note that the load problems on direct drive using planetaries aren’t as bad as a lot of people will tell you, as long as you handle them properly. First, you do need a robust planetary, like an old-fashioned Banebots or an Andymark Sport, which are made to take loads better than VersaPlanetaries (these are great gearboxes, but built light.) Then you need to make sure to support the output shaft of the planetary with a bearing to ease some of the load. We’ve done this for years when using this kind of setup for mecanum drives at it works fine without undue stress on the gearbox. We were even able to drive off the level 2 platform at speed in 2019 with this kind of setup and never had a problem.
I wasn’t planning on Versaplanetary. I actually had those Banebots Gearbox in mind but I couldn’t remember who manufactured them so thanks for reminding me.
One of the advantages of a 6WD drivetrain is that with all three wheels on one side chained together, even if one or two loses contact, the total motor power can still be directed to the wheel that still has traction.
When the NEO came out, I did toy around with having a motor above each wheel in a 6WD chassis, but they were also all chained together: 6 NEO "outside-in", WCD, AWD drivetrain sketch
They would be Bolted Into a Chassis like this. And this Chassis still needs some work done to it as my original design will not fit everything.
I think I chose my words poorly. I do not plan on having any drop center on the drivetrain.
My Idea Is that the wheels could be hot-swapped out for Mechanums or All-Omnis. If you felt the need to, and through software, you could change the controls this way.
You would be leaving the center drive units out for mecanum, right? There really isn’t a good way to do a six wheel mecanum drive. In theory you could put omnis in the middle position, but a lot of the more sophisticated movements possible with mecanum (like diagonal movement) would still be a problem.
A drivetrain that is capable of being a lot of things is rarely good at being any of them. I would argue that this concept of the “configurable” drivetrain is more harmful than helpful.
Instead of designing an “all in one” drivetrain, consider designing three different drivetrains that are optimized for each drivetrain style - that way you don’t need to lose out on the benefits of one type in order to include the other types.
In 2014 we designed a 4-wheel drivetrain that was swappable between mecanum and traction wheels. We weren’t sure whether we’d rather have the extra agility or pushing power, so we tried making a drivetrain that would be able to do both. In the end we got a drivetrain that couldn’t really do either. As a mecanum drivetrain it was too slow to be effectively agile (and in reality mecanum was a terrible choice that year). As a traction drivetrain it had a lot of trouble turning (being a near-square 4WD) and lost half its power when it was pushing anything (because the wheels weren’t chained together). We would have been much better off picking one drivetrain and sticking with it.
You should not make major design trade-offs on the chassis to only be multi-function.
That said, the popular AM14U is an example of a chassis frame that can be adopted to Mecanum, 6 wheel , and 8 wheel tank with no major compromises other than perhaps more machining time.
Could be worth looking at how they do it.
Also…you definitely want a drop center. 1/8" is generally good. Even if you plan to run omnis in the corners, you should run a drop center: especially if your team is not really experienced with chassis design. Without it, your robot will be terrible at turning.
I’m going to go back a few years here… you can do a drop-center 6WD and still keep mecanum capability. All-omni is a little trickier but you could just run a 4-omni. What a lot of people don’t know is that 330’s 2005 robot was designed to run either 6WD drop center OR mecanum. We had the gearbox slots and everything. But, we determined experimentally that going sideways was a waste of time that year, and the mecanum never made it onto the comp bit.
If you don’t have a center drop, there’s a pretty good chance you end up with turning issues. You’ll need to choose your wheels very carefully.
@Strategic it’s been done though not with 6 mecanums. Center omni with a ball differential as I recall–1322, about 2010. 8 mecanums has also been done but that’s considerably easier.
If you want to support mecanum and WCD, keep the center drop and use clamping gearboxes in the corners. You can then chain motors on one side together (in WCD mode) or leave the corners independent (for mecanum).
You should work out what the torque will be on the output shaft of the planetary gearbox would likely be for the case where only one of the three wheels on one side is touching the carpet. Then multiply that number by a factor of 5 X to 10 X to account for shock loading and compare with the ratings for the particular gearbox you are going to use. Otherwise, your gearboxes may explode after a few hits or jumps.
I’ve seen the 8-wheel mecanum proposed before, which makes sense given that you can balance the wheels in two properly opposed sets (and theoretically increase your effective pushing power over a standard mecanum since you have twice as many motors involved.) I suppose the six-wheel as you describe might work well enough as a mecanum and certainly would give increased push in the forward/back direction.
Yes, with Mecanums the middle wheels would be deactivated via software.
Yeah, so maybe direct drive isn’t a great idea.
To me more than anything, I wan’t to build a drivetrain that drives and rides smoothly. I will probably use chains (or belts) in the final design. While the ability to run mecanums or omnis sounds good on paper, In the real world it probably doesn’t make much sense. I am a big fan of H-Drives anyways when I want to do anything holonomic.