Summer Drive Train Prototyping

[nrogals]

Hey Everyone,

We are looking into some new drive train layouts during the summer. Our team is definitely going to try to keep the complexity to a minimum, because we want to come out of the summer with a reliable, strong drive train.

Although holmonic, swerve, crab look cool, we don't really view them as appropriate in risk versus reward analysis.

We really would like to have the capability to be both powerful and agile. So an idea is that we would use a transmission for the first time.

4 directly powered high traction wheels Possibly smaller ones to transverse objects

Our team likes to stay away from chains, belts and so forth to a degree.

With a Transmission:
One setting would be geared low- high torque, high acceleration, fast turning, strong pushing, slow maximum velocity.
Second setting would be geared high- low torque, low acceleration, high maximum velocity.
The body of the robot would be made out a light weight material.

Does anyone have any comments on the planned features? Please be blunt. I'm ready for the heat.

Would orchestrating the shifting on four independent gearbox modules be difficult?

Has anyone looked at coupling additional motors together such as banebots to cims to get more power?

Would putting a lot of time into making the shifting and gearbox mechanisms be worth it? Does anyone see any other alternatives in getting agility and power?

Thanks for your help!!! And I wish everyone a great summer and awesome off-season.

[tickspe15]

Originally Posted by nrogals

4 directly powered high traction wheels Possibly smaller ones to transverse objects

Our team likes to stay away from chains, belts and so forth to a degree.

.

So you want to have a direct drive drivetrain with no gear box'

CIM motors dont have nearly enough torque to move a 150 pound robot, am i understanding you right

[runneals]

Originally Posted by tickspe15

So you want to have a direct drive drivetrain with no gear box'

CIM motors dont have nearly enough torque to move a 150 pound robot, am i understanding you right

I suggest you take a look at our Butterfly Drivetrain if you are looking for a direct drive train that takes out the gearbox. It works pretty well!

[EricH]

I believe a 4WD high-traction drivetrain has been compared to a Mexican jumping bean. You WILL bounce when turning, if your drivetrain is significantly longer than it is wide! (Unless you're from FRC 25.) Even some 6WD high-traction drivetrains with dropped center wheels have had that problem.


That said, a shifting gearbox will accomplish most of what you're looking for. However, 4 of them is going to be a bit on the heavy side, so you want to find some really light ones. Orchestrating the shifting would depend on what you used; servos would be as simple as a trio of Y cables to split one output into 2, then each of those into another 2, while pneumatics could be done off of one solenoid by splitting up tubing runs (though I might suggest 2 solenoids with the equivalent of a Y cable).

The other "problem" is that you'll probably want to use higher reduction than most teams use in their boxes, due to not having any chain/belt to knock it down a bit farther.

If you have the funding, I'd spend the money on a couple of AndyMark or VexPro shifting gearboxes and spend some time tinkering with them to find out exactly how they work, rather than making the gearbox from scratch on the first iteration. (Later iterations, go for it. But if you've never done shifting, learn from the masters.)

BaneBots to CIMs, don't do it. Not because it can't be done, but because when you stall the drivetrain (never "if", in FRC), you'll probably be replacing the BaneBots due to magic smoke release--they don't handle stalling well. MiniCIMs or BAG motors would be a better choice in terms of durability. You'll need to match up the motors to get the power/speed you want.


I would suggest looking into the West Coast Drive--typically 6WD, dropped center wheel (eliminates the bouncing problem) directly driven off the single gearbox for each side, all other wheels chained. Or the Kitbot on Steroids that 1114 has been spreading. Adapting one of those to fit your preferences might be a little easier than going completely from scratch.

[Pault]

I think I have a suggestion that could give you what you want without the turning problem.

Last year my team had a 4WD long bot (actually, it could also move sideways, but I am going to leave out the features that let it do that because they don't matter right now). We used AM 8in Pneumatic Wheels, so as you can imagine turning was horrible. However, we also had 4 stingers in the corners of our robot. And each one had a slippery foot* to touch the ground. So what we discovered was that, because they were motor powered rather than pneumatic, we could lower 2 of them just enough so that our back wheels were lifted off of the floor. Now, we could turn really well while still being able to convert into high traction mode.

So, anyways, my suggestion to you would be to do the something similar. Have a little slippery foot on the back of your robot that can be lowers by pneumatics. Your turning problem is gone without sacrificing anything.

This could also be an excuse to not use transmissions. You claim that you want power and agility, but you don't say anything about speed. So if you don't mind having an average speed drivetrain, than transmissions is definitely an idea I would trash. It would be a big task to produce them, and 4 is a lot of weight to add on. If you do want speed, maybe try a 4 CIM, 4 miniCIM drivetrain. Or buy 2 COTS shifters and just use chain (this is what I would do, but I dislike direct drive, so I might be a little biased).

*Made from UHMWPE

[MichaelBick]

I know you are asking for a 4 wheel drivetrain, but I would highly recommend a kitbot on steroids. They use belt, however andymark has designed it so that it is very easy to do. On the old kitbots you could switch to supershifters, however I'm not sure if they kept that feature. I'm sure you could use supershifters and switch to #35 chain and that would be rock solid. If you want a solid drivetrain for next year, a kitbot on steroids or a variant would be your best bet.

[Chris is me]

There's a reason most drivetrains have six (or more) wheels in them. Given the pre-2013 robot dimensions, in the long orientation, four wheels at the corners just will not turn very well. This is occasionally a desired feature (see: a butterfly drive), but you need SOME way to turn.

One way to do this is to use different traction materials on each wheel. The problem is that by compromising traction on half of your drive, you are losing out on pushing power to an extent as well as pushing resistance.

Another way turning is accomplished with 4 wheels is to shorten the wheelbase such that the four wheels are closer together. To maintain stability, you have to add casters or unpowered omni wheels that will hold your robot frame off of the ground, as your wheels are now much closer together. However, unpowered wheels rob normal force from your powered wheels, which results in less pushing power.

Going to six wheels, with the center wheel "dropped", effectively halves your wheelbase length, to the point where the wheel base is notably wider than it is long. Depending on the drop, wheel material, etc. you either ride on four wheels at a time, or all six wheels touch the ground due to tread compression. In either case the middle wheel takes most of the load and turning about your robot's center is easy.

The tricky thing is that a six wheel drive can't all be direct driven out of transmissions like you want to. If you must avoid chain or belt at all costs, and you have access to precision manufacturing of some sort, you can do a gear drive, similar to what 829 did this year. With aluminium gears readily available, it's not as much of a weight penalty as it used to be, and it does prevent any chain or belt issues from popping up. That said, using 35 chain is honestly not that hard.

[Woolly]

If you're going 4 wheel skid, the sides of the robot you're driving from need to be shorter than the non-driven sides, otherwise your maneuverability will be compromised.

Also, I'd say that you should be fully prepared to use chain or belts. Each side of the skid steer should probably have 2-3 CIMs though the same shifting gearbox that directly drives 1 wheel, and then chains or belts to the others on that side.
I'd try start with Vexpro gearboxes to begin with, and in future years upgrade to West Coast Products gearboxes if you need that little extra.

We're also tossing around the idea to do some drive train R&D this summer, and we're looking to make a 2-speed 8 Wheel Drive with placation wheels and a belt drive. Potentially we would put omnis on the outer 4 wheels if we want to trade some resistance to defense for maneuverability, but who knows what we'll find.

[cmrnpizzo14]

Welcome to CD Nate! To expand upon the OP a little bit, we have used AM nanotubes for the previous three years. 2 of those 3 have been mecanum, one was 6W tank with only the back 4 wheels direct driven (with gear reduction) by CIMs. We have limited machining capabilities so while a WCD would be nice, as far as I know it is a little bit beyond our capabilities. Kitbot on steroids is exactly the style that we are looking for. Articulating drive trains or swerve are essentially out of the question. Super shifter or ball shifter reviews and comparison would be appreciated as well.

[Nemo]

I think you should strongly consider the AndyMark 446 Drive Module. You get a COTS two speed drive module with gear drive for the four center wheels. You shouldn't be too scared of the chains that drive the outer wheels, because you're still mobile even if all the chains break, similar to a West Coast Drive.

If you want two speeds and a long robot that can turn, completely eliminating chains/belts will end up adding a lot more complexity than it removes.

If you want zero chains, simplicity, the ability to turn, and traction, I only know of one way to do it - these nanotubes (also mentioned above). Those are single speed, and you don't get the power from all four drive motors unless the correct four wheels are all touching the carpet.

Or just do kitbot on steroids. Or find a custom way to direct drive the center wheels of a 6WD to eliminate two of the chains. Like Eric said.

[KrazyCarl92]

A great summary of many of the drive train fundamentals you are asking about can be found here:
http://www.chiefdelphi.com/media/papers/2429

I am a huge proponent of making reliability the number one priority in drive system design. It simply must work. This is why I like direct drive live axles. If every chain or belt on the robot broke, the thing would still move. That being said, I would highly encourage you to consider the possibility of direct driving 2 wheels on a 6 or 8 wheel drop center drive train, or even consider the possibility of direct driving the 4 dropped wheels of an 8 wheel drop center drive. The remaining wheels can be chain or belt driven to gain the advantage of all wheels being driven, but you still get the reliability from the dropped wheels being direct driven from the gearboxes.

One thing that others have not yet touched on is the true advantage of shifting in an FRC drive train. While you had noted that you would like a high gear and a low gear, it is important to understand why this is desirable in order to execute well.

With just about any single speed 4 CIM FRC tank drive, there is a tradeoff between the abuse the motors see and a high end speed. You could gear for 4 CIMs to never see more than 40 Amps per motor, but you would only go 6 feet per second. You could also gear 4 CIMs to travel 15 feet per second, but the motors might see upwards of 100 Amps a piece at times.

A shifting drive train allows you to get the best of both worlds: rather than settling for an in between gearing where the desire for a high top speed cannot be satisfied without sacrificing motor "health", both of these goals can be met.

All this being said, with many teams using 6 CIMs in their drive train, the availability of miniCIMs, and the possibility of game design completely changing drive train design criteria (2009...), there are ways that drive train performance can be improved without shifting.

When designing your drive train, keep in mind the reasons for using a shifting gearbox in the first place so you can reap the rewards of your work. Consider gearing the low gear such that the CIMs pull no more than 40 Amps at the traction limit. Figure out what your top speed should be for high gear based on your strategy for the game and an understanding of the inevitable trade off between acceleration and speed.

I HIGHLY recommend the Kitbot on Steroids.

[nrogals]

Thank you everyone for you recommendations. Can someone check my understanding of how gearing, shifting and game play work?

First, from game play, there are two main criteria that define a drive train's performance, outside of reliability, maintenance and so forth; Namely they are: speed(linear speed) and pushing power(acceleration).

The best way to address both these criteria is through gearing and shifting.

With no opposition, we want to go fast. Aiming for around 15ft-18ft/s. At this speed it would be easy to get pushed. Although we would be agile.

For the above situation, you want to gear it such that you have the highest speed possible, enough torque to overcome dynamic friction and accelerate (dependent on mass of robot) fairly fast.

Amperage draw for each motor would like to be under 40 for the sake of motor "health." In the powerpoint, http://www.chiefdelphi.com/media/papers/2429 it says to design max power for 40 amps. How would you go about doing this in choosing the gearing? To do this, I would pick the point at which I get max power out of the motor at 40 amps(data), determine what torque I need for the situation above(experimenting/estimation), adjust the speed accordingly to get the same power output as the motor(calculation), and then after that choose the appropriate gearing(calculation).

For opposition and turning, namely times when the robot is at a low speed, the robot will be in low gear. The gear needs to be low enough such that it can generate enough force to push other robots. Low enough to turn against whatever friction the design requires. High enough to turn quickly.

Again would I use the same process of calculation as above. Matching my max power for max power needed when pushing a robot. Determine torque needed. Determine speed and gear, "design" accordingly.

Following the advice above, I think the best drive train setup for us would be a KITBOT on Steroids with custom gears to fit our need for the game.

Does anyone have a recommendation to what type of transmission mechanism to use: super shifter, or vex pro shifters.

Thank you for your help everyone. All the best.