Drive train questions
 

I'm running over some possible drive train ideas, with physics not starting till next year i have a few questions.

1) What (if any) advantages is there to having 2 gear boxes with 2 cims in each as appose to 4 gear boxes with a cim in each?

2) Just wondering if this would work. A drive train with 2 mecanum wheels in the center and an omni wheel in each corner. Only the mecanum wheels are powered. If I'm thinking of this correctly you always are crabing with no way to turn your robot. Is this correct? I'm not looking for people telling me how stupid it would be to have a robot that can't turn, i know this, just would this happen?

3) In the two years I've been on the team we have not used pneumatics. Just making sure, the cylinders can be fully extended, fully compacted, but can they be put somewhere in between? Like half extended.

I'm sure I'll come up with more questions to ask about drive trains, that's all i can think of right now.

Re: Drive train questions
 

1. Aside from complexity, weight is a big savings. It is 2 less sets of reductions you need to take care of.

2. Mecanums need to be used in sets of 4. I'm not going to go into all of the details, but the physics of mecanums require 4 to achieve "crabbing". Thus, using 2 mecanums to drive your robot will not do what you are trying to achieve.

3. The short answer is yes they can. The long answer is that its fairly complicated to do so. The rules this past year on pneumatics were loosened slightly, and in some cases I saw teams that were using 3 position cylinders.


-Brando

Re: Drive train questions
 

You could go forward or backward, and you could do a combination strafe+turn left or right. You couldn't strafe without turning, or turn without strafing.


I think the word you are looking for is "retracted". The simple pneumatic cylinders that came with the 2010 KoP (by request) are force devices, not servos. Operated open-loop, they're not designed to stop halfway and hold that position against an external load. You could, of course, add a position sensor and try to turn the cylinder into a linear servo by doing closed-loop control in software. You'd have a tough time holding the position against a varying load due to the compliance of pneumatics. Getting a robust stable solution would be challenging at best.

Re: Drive train questions
 

Thanks for the help, I thought mecanum wheels worked by pulling at a 45 angle from the rotation and you only needed 4 to turn, guess i was wrong. Another drive train to the trash.

Re: Drive train questions
 

I know it can be done because in 2008 we did it with a 12" long 1.5" bore cylinder and basically we de-pressurized the cylinder after so many milliseconds so it stopped half way but it would not hold its position.

I know Team 254 used a different type of piston seen here http://www.team254.com/media/photos?func=detail&id=3146 and here http://www.team254.com/media/photos?func=detail&id=3133 and I want to say it was used to shift their drive system into natural

Re: Drive train questions
 

That is how they work. And it's why you need 4 to strafe. You need the back wheels to cancel out the fore-aft force vector components.


~

Re: Drive train questions
 

Yes, the 2010 rules on pneumatic cylinder choice were very relaxed so we were able to use more complicated cylinders. Note that 3-position cylinders will require more than a standard single or double solenoid valve (we used two single solenoids).

Also, back in 2008, we used a center-closed pneumatic solenoid from SMC. This solenoid allowed us to hold our intake arm in the middle of its stroke.

Re: Drive train questions
 

For your first question (2 versus 4 gearboxes), Brandon's reply concerning weight covers a large portion of the correct answer, but there are nuances.

• Another clear advantage of 2 gearboxes & 4 CIMS is that if the front or rear wheels of your robot become under-weighted (and therefore lose traction, such as when climbing a ramp) the drive power becomes available to the other, over-weighted wheels. So you do not lose effective drive power when the weight distribution to your wheels shift.
  A second advantage of 2 gearboxes is that you can better afford the weight (and cost) of a shifting gearbox.
  Some drivetrains, however, require independent drive to function. Mecanum is a good example of such a drive.

I have nothing new to add to the Mecanum subject.

Regarding pneumatics, it is possible to have a position control on pneumatic devices, although the devices for FIRST are designed for binary (extended / retracted) use. In the chemical process industry (I am a Chemical Engineer), many flow control valves are pneumatic and they control flow by controlling the valve stem position. They work very well & reliably. Such analog pneumatic devices work using a controlled analog pressure signal which varies between 3 and 15 psig. I/P (current to pressure) converters convert higher pressure instrument air to this 3-15 psig signal proportional to a 4-20 mA input analog electrical signal.

In FIRST, however, our pneumatics are designed for on/off service. It is possible to stop at an intermediate point with the standard parts, but it is difficult and complicated. We've done this in test-beds, but have never incorporated this concept into a working robot.

Re: Drive train questions
 

Not necessarily a fault of multiple gearboxes, but if you have say 4 CIMs and 4 gearboxes, and each gearbox drives one of four, two of 8, etc. wheels, then when one wheel leaves the ground, you're transmitting less torque and aren't getting all of the power you could. With both wheels on one side chained together (ether with a single gearbox for both motors, or two gearboxes for two motors but all of it connected) you can still use both CIMs with a wheel lifted.

Re: Drive train questions
 

This is why I'm confused. Let's say that you do have the design I mentioned, if you drive both forward you drive forward because the right wheel is pulling forward and left, and the left wheel is pulling forward and right. Therefor left and right cancel out (assuming same speed). Same can be said for driving both backwards. Now if you drive the right wheel forward it pulls forward and left, and the left wheel backwards, backwards and left. Shouldn't you go left? And running the right one back and the left one forward shouldn't you go right?

As I see it the only reason i would need 4 mecanum wheels is to turn, not to strafe.

Someone explain to me how I'm wrong considering no one else sees this happening.

Re: Drive train questions
 

that would work if the two mechanum wheels were occupying the same point in space, however this method would cause the robot to twist when you tried to strafe due to the fact that they wheels being apart from each other would cause a rotational force to be applied about the center of gravity. in a perfect world this would work, but not in the real world.

Re: Drive train questions
 

These days, I live in the real world.

Re: Drive train questions
 

Here's where your upcoming physics course will be helpful. Since the forward force from the right wheel and the backwards force from the left wheel are not colinear, they create a torque moment which causes the vehicle to turn left as it is strafing left.


If you want to strafe without turning, the rear wheels are necessary. To strafe to the right, the front left and rear right are driven forward, and the front right and rear left are driven backward. The front left and rear right generate forward and right force components; the front right and rear left generate backward and right force components. The backward force component of the front right wheel is colinear with and cancels the forward force component of the right rear wheel; the forward force component of the left front wheel is colinear with and cancels the backward force component of the left rear wheel. What remains is the right force component of each of the four wheels. So the vehicle strafes to the right, with no torque to cause turning.

Re: Drive train questions
 

B,
There is nothing that prevents you from using two cylinders in series to obtain a three position mechanism. Apply air to one for the middle position and both for the fully extended.

Re: Drive train questions
 

Darnit, Al beat me to it. We used a system like this to good effect in 2010 for our blocker. With two pistons retracted we could drop the blocker completely to go through the tunnel, with one set deployed the blocker was at 45deg for hitting balls out of the return, and with both sets deployed the blocker was vertical, good for blocking shots.

We found the best way to do this was to use a threaded coupler and jam nuts to attach two piston shafts together and use the supplied brackets to mount the 'base' of each piston. It was a very robust and reliable system, and flow-control valves can turn the pistons from essentially undamped springs into spring-dampers, which may or may not be useful to you.

Re: Drive train questions
 

Thanks everyone for your help. Just so i make sure I have this right...

1) Just weight, cost and complexity

2) Doesn't work

3) Attach 2 pistons together and it works great

Thanks again

Re: Drive train questions
 

^^ You got it.

Re: Drive train questions
 

There's a good way to visualize how this would happen too. Imagine a merry go round you might have had in a playground growing up. Now, put a person on the east (right) and west (left) side of it. If both people push north, it doesn't spin (ie the robot would go straight forward). However, if the guy on the east pushes towards the North West, and the guy on the West pushes towards the South West, they can spin the merry go round (Because one is pushing towards the north and the other towards the south). It's not a perfect spin, however, because they aren't pushing tangentially to the merry go round - some of the force goes into translation.

This analogy isn't perfect, as it's hard to visualize that translation with it. But it does a great job with rotation. For a normal Mecanum drive train (with 4 Mecanum wheels in the corners), you can stick people on the North East, North West, South East and South West corners and have them push in different combination's. You'll find that when they're all pushing tangent to the merry go round, in the same direction (clockwise or counter clockwise), the merry go round will turn without translation. And when you have them working in opposing pairs there won't be any rotation, but the force has to go somewhere - you should be able to figure out what the translation would be.


In physics, this is called a free body diagram. You imagine your forces acting on a pivot arm coming from the center of mass of your object (arguably the center of your robot, for all practical purposes... although your mileage may vary based on specific robot designs). If the forces line up properly, they'll cause the object to rotate or translate (or both!). It's one of the more important concepts in physics, and one that students seem to have the hardest time grasping.

Re: Drive train questions
 

A (perhaps obvious) caveat: if you mean the above literally, it only works if the pistons are powered in the retract direction (and the load force acts to extend the pistons).

If you are powering the pistons in the extend direction and you just attach two pistons together, they will buckle unless there is some sort of supporting structure at the point of attachment.




~

Re: Drive train questions
 

One minor quibble with this explanation. The phrase "the force has to go somewhere" could be misleading to a new student. For example, if the front wheels are being torqued forward and the rear wheels being torqued backward, the "force goes nowhere". There is no motion. No translation, no turning.


~

Re: Drive train questions
 

Good catch... When we worked on implementing Mecanum, we only looked at the productive scenarios (since the one you point out would only serve to waste battery power), so any unproductive ones like that just didn't come to mind when i made my post. Thanks for helping to clarify that :)

Re: Drive train questions
 

If the pistons are joined shaft-to-shaft with a rigid coupler (like I described) one should probably check the design design using critical buckling load calculations, especially if the pistons are of a long stroke and small bore. Fortunately the actuation pressure can be reduced to help prevent the assembly from reaching its critical buckling load.

Re: Drive train questions
 

While that's all true, you run into another advantage of the 2 gearbox method when one of your wheels lifts off the ground. (I think this has been mentioned, but since you didn't summarize it...)

With 4 independent gearboxes+motors, you're effectively just losing/not using that drive power when the wheel lifts. With the wheels coupled (1 gearbox for 2 wheels), if you lift one wheel, the power can go to the other. This is doubly beneficial because that wheel is probably overweighted, since the other is lifted. So this is good for a traction standpoint.

(An obvious disadvantage of 2 gearboxes is if your drive requires independent steering.)

Re: Drive train questions
 

If you make these cylinders of two different lengths, you would have four positions - both rectracted, short one extended, long one extended, both extended.

Stopping a cylinder in mid-stroke is fairly straightforward by using a 3-position solenoid valve. Center position is all ports blocked, so the cylinder is pseudo-locked in position, although with some spongyness. With the right position feedback, programming, and flow controls, you could move to any variable position.

Re: Drive train questions
 

A few more questions for cd!

What is a good gear ratio from a duel cim transmission for

1) A high speed robot (reasonable I know you could go 1:1000 and just stall)

2) A high torque robot (again reasonable I know its possible to go 9001:1)

Thanks for the help again!

Re: Drive train questions
 

That all depends on how big of a wheel you plan on using. As an extreme example, a 2" wheel will be 10x slower and have 10x the 'pushing force' as a 20" wheel with the same motor gearing.

Do you know how to do the calculations? They are fairly straight-forward to explain if you don't (and it's okay if you don't:) ).

Re: Drive train questions
 

A rule of thumb I have heard thrown around is 8-12 feet per second for most games to make it easy to drive and minimize collisions. How you get there from the motor speed to transmission to drive coupling to wheel size is up to you.

Re: Drive train questions
 

8-12ft/s is about right. Think 5-6ft/s for pusher robots.

Re: Drive train questions
 

Actully, I disagree. I think that your robot should be geared to go as fast as the drivers can control it. If you look at teams such as 67, 1114, and 254 (just off the top of my head) they all have extreamly fast drivetrains that they can all control very well. This is generally because their drivers are very experienced from a combination of past seasons and practice robot driving.
Because games in FIRST are unfailingly "who can do more/faster" senerios if you want to win you have to go fast. So if you are really good but not that fast you can never hope to be as good as the robots that are both.

As for low gear make it low enough to push stuff but not low enough that other robots can run away, that aside its all preferance.

my 2 cents

Re: Drive train questions
 

Those teams also understand the tradeoffs of having 14-16fps drive trains.

To go 14fps+, a robot that expects to have all of its drive train motors last the entire season would need all 4 CIM motors on the drive train. The robot better also have a low gear because simply turning the robot would result in extreme torque loss at any gear ratio that gives such a high speed. I suspect that's why the poof's advertised speeds are 7fps / 16fps on a couple of their recent robots rather than 4-5 / 16. I also don't ever notice the poofs trying to push another robot in a defensive manner; usually I see them pushing through a situation and then zooming away, but I've only seen a small percentage of their recent matches.

To go even faster, even more drive train motors are needed. For a 60-lb (or so) robot that tumbled around the track in 2008, 148 used 6 motors on their 3-wheel crab that (iirc, off the top of my head) went somewhere around 18fps.

Additionally, there are some slide decks on AndyMark's website that give insight as to why their SuperShifters have the ratios that they do -- anything lower than 4-5fps for low gear has a tendency to make the wheels slip on carpet.

I do agree with the "who can do more faster" concept, but that really only applies to Einstein-grade teams who are anticipating competition against other teams just like them. I say that because they also need extra practice in controlling the drive train at high speeds (as well as fast manipulators as high speeds), thus they allot time for extra practice & sensor integration before competition. Ergo, versus an average team an Einstein-grade team would (probably) win regardless of what drive train speed it chose so long as the speed was "fast enough".

Re: Drive train questions
 

I agree with you in principle, that a team should build a robot to be as fast as can be controlled. However I would assume that a team asking how fast they should gear their robot probably does not have the driver experience to handle a robot that's faster than 10-12 ft/s, which is remarkably fast.

Here's the mandatory "it depends on the game" disclosure: if it's a wide-open field, like 2008, where a robot can get up to speed and maintain that speed for a good portion of the match then a very fast robot would be more useful. If it's a segmented field, much like 2010, where there is not as much room to accelerate and maintain a high speed, a robot might be more effective (i.e. have a higher average speed) if it could reach a lower top speed faster.

Re: Drive train questions
 

Assuming no one does better then him next year with the practice bot, we will have a second year driver. Also just because i have no idea how to do the math on this, if anyone one wants to what would be the reduction for 8" wheels if I want it to go 4 fps and lets say 15fps. If anyone wants to do the math thanks, if not I understand, its summer.

Re: Drive train questions
 

In essence (be sure to keep units in check):

Robot Speed (feet/sec) = Wheel Circumference (feet/rotation) * Wheel Rotation Speed in (rotation/second)
= [Pi * Diameter] * [Motor Rotation Speed / Gear Ratio]

Assuming Diameter = 8" (0.67 feet) and Motor RPM for a CIM is ~= 5400RPM (90 rot / sec)
15fps = Pi * .67 ft * 90 rot/sec / GearRatio

or

Gear Ratio = Pi * .67 * 90 / 15
= 12.65:1 for a 15ft/sec bot with 8" wheels.

Feel free to put in the numbers for 4ft/sec.

Re: Drive train questions
 

"It depends". Say for case 2, your number one priority is pushing. All you need to do is push all day and you don't care about speed or anything else, just pushing.

The limiting factor will be the traction of your wheels. Once your wheels spin in place when your robot pushes against an immovable object, you're done. Gearing lower won't make you push harder, so you might as well gear your robot to right at that point in order to preserve whatever speed you have. However, if that takes more than about 40 amps of current per motor to happen, you will trip your robot's thermal circuit breakers. You can mathematically calculate that gear ratio based on your robot's wheel size, the motors you're allowed, the maximum current you want to draw, and whether or not you've reached your maximum pushing force ("traction limited"). The number's somewhere in the ballpark of 4.5fps for roughtop tread (CoF 1.3), but I'd encourage you to figure that out for yourself.

However, gaining that ability leaves you with a very low top speed. Not many FRC events could be easily won with a robot that only goes 4.5 feet per second! If you want to go faster, either you lose the ability to push against anything forever, or the traction of your wheels is going down reducing your pushing force. That, or you use a shifting transmission and go for both.

Now, the first question: The vast majority of drivetrains require skidding wheels (the exceptions are generally too complex for someone like me to have anything to do with), so your drive will need to be limited by traction at some amperage. This doesn't necessarily need to be 40 amps, since the breakers do not immediately trip and you rarely need to turn for more than a few seconds, but you're obviously constrained by the stall current of the motors, etc. I doubt you'll be gearing your robots close enough to the stall current to matter though because of the concern of acceleration.

You also want to weigh how fast you need to accelerate and how quickly you need to get to something x feet away. At lower speeds (under 8-9 fps or so) you'll probably get to something 5 feet away relatively quickly. Acceleration becomes more of an issue when you get to speeds in excess of 11-12 fps, roughly. How much of an issue depends on the game, how far and fast you need to go in that game, etc. so you'll have to decide for yourself on that one. Luckily someone from team 1640 posted a nice whitepaper on the topic.

In short, there's a lot to consider and no stock answer; that's why more teams don't build identical drivetrains every year. One more thing I want to add, is that how fast you can achieve a game objective is often less affected by the gearing of your drivetrain than you might think.