power transmission NOT using gears/chains?

[Frank(Aflak)]

I was toying with possible gearbox/transmission ideas as a start for next year, and I was wondering if polyurethane or however that is spelled rollers would work to transmitt power like a gear.

specifically, I was thinking about MrB's white pages gear ratio thing that had an idea for a CVT at the bottom, and was wondering is high traction rollers would work.

overview: mrB had two cones side by side, one with the base at the top of the drawing, on with the base at the bottom, so there was a constant distance between them. And in between them was a gear/roller that could move in the space between them. At one end it was close to the base on one and the narrow part of the other, at the other end it was reversed . . so as you slide it back and forth the gear ratio changes. Gears could not work for this design, at least not without very very complicated tooth patters, and maybe not even then. But what if you used high traction rollers? It would even act as a clutch of sorts, if you could vary the tension.

so, how would I get good sticky stuff to put around cones? Also, the roller in the middle may be hard to move, so maybe an omni wheel type thing would be best. or just really powerful actuator to move it.

MrB's spreadsheet is here. (http://www.chiefdelphi.com/forums/papers.php?s=&action=downloadpaper&paperid=166)


Well, really what I want is polyurethane/other sticky stuff suppliers to scout.

end.

[Al Skierkiewicz]

Frank,
Detroit has been experimenting with continuously variable drive systems for many years. They have been particularly suited for hybrid or all electrical vehicles. I know there is a great amount of info out there beginning with Popular Mechanics going back many years. I am surprised to not have seen any popping up in our robot designs but I do keep watching for them. There are a number of teams that have taken the shifting idea to a logical integration in robot design. Of particular note is Thunder Chicken's multi-speed transmission of last year. I believe that there is a white paper on their design. It was not only impressively designed and implemented but was surprisingly small and very well engineered.

[camtunkpa]

Our team has actualy built a CVT and it worked really well, but because of this years game we could not use it and fit everything else in to get under the bar.

[ahecht]

190 had a toroidal CVT last year, and will have one again this year if we can get it in under weight. The hard part about the CVT is programming it, as it is a very complex system to model.

There are many other kinds of CVTs out there, from the cone/belt design (like in the Honda Civic HX), to variable geometry designs, universal joint driven (Varibox), etc.

[mjt902]

i don't think anyone is still following this, but i am really looking for somebody who has knowledge of a robot clutch system that can shift on the fly?? i was informed earlier today that there were several teams that could, and me and a friend are wanting to make our own... are there any quick and easy (or long and ardruous) processes we might need to work through that you could help us solve??

[DougHogg]

Originally posted by mjt902
i don't think anyone is still following this, but i am really looking for somebody who has knowledge of a robot clutch system that can shift on the fly?? i was informed earlier today that there were several teams that could, and me and a friend are wanting to make our own... are there any quick and easy (or long and ardruous) processes we might need to work through that you could help us solve??

We don't use a clutch, but we do shift on the fly by using two sets of wheels, one large and the other small. You can see pictures at our web site. See below.

[Al Skierkiewicz]

I was talking to some of the guys on Christian Holland Team 107 at Midwest and they said they were working on a CVT drive system. They hadn't worked all the kinks out so they didn't include on this robot but what I had described to me sounded pretty cool.

[Mike Rush]

I know that 64 has a CVT this year. Don't know exactly how it is working out. Oh well, guess I'll find out in Houston...:cool:

[Greg Ross]

Originally posted by DougHogg
We don't use a clutch, but we do shift on the fly by using two sets of wheels, one large and the other small. You can see pictures at our web site. When I first saw these wheels, I thought they were spinning their wheels when they weren't even pushing! It took a little while to realize that those wheels weren't even contacting the carpet, and that they could lower them when they wanted to go fast.

[Wetzel]

Originally posted by gwross
When I first saw these wheels, I thought they were spinning their wheels when they weren't even pushing! It took a little while to realize that those wheels weren't even contacting the carpet, and that they could lower them when they wanted to go fast.

I thought they were different material for traction on the carpet and ramp, but then they were used on both. Very interesting idea.

116 designed a 2-speed transmission that can shift on the fly, and be built with limited shop capability. It is clutch less, but not extensively tested in competition.
This is from 2002, so note that the drill motor is different from this year. It was updated for the new drill motor, but no white page as yet.

http://www.team116.org/2002/EDTransmission.pdf

Wetzel
~~~~~~~~~~~~~~~~~
mmmm...manual shop tools

[Jared Russell]

Take a look at team 45's ball drive demonstration bot...by far the most interesting drive train I've seen.

[Matt Attallah]

A CVT is very, very nice, but one draw back. Sure, you can have a huge "gear selection" but their 1 drawback is their ability to hold torque. I'll assume for FIRST drill motors, FP motors, and others in our kit, it will be fine, but that is why you don't see more CVTs out there today. Their belts will slip under that high torque demand situtation!

I have tought of cone-shaped gears or so. I'll be working on that during the summer and see if i can get it to work w/o the weight and size that I am currently thinking of!! :D

[sanddrag]

Originally posted by Matt Attallah
I have tought of cone-shaped gears or so. I'll be working on that during the summer and see if i can get it to work w/o the weight and size that I am currently thinking of!! :D I don't think it will work at all. Either the cone gears will have a constantly varying pitch or more teeth will majically appear the further you go toward the larger side of the cone. Niether of those ways would work. So, unless you've got some patented gear technology I don't know about, you're probably bettr off sticking to high friction rollers or something of the sort.

[AnimatorSarah]

I'm not sure if this will help but our team this year shifted on the fly by using a little pnumatic cylinder to shift back and forth. We also added a spring to help the cylinder shift back to high from low. I don't know if this helps you at all since I'm not exactly what kind of drive train your going for.

[KenWittlief]

The Honda CIvic has a CVT in it.

One easy way to make a two speed tranny is to use the freewheeling mechanism on the back wheel of a ten speed bike.

(Not the shifter)

when you pedal a ten speed, as long as the wheel is going slower than the pedals, the pedals stay engaged. If you go down a steep hill, and dont pedal faster, then the pedals are just spinning, and the back wheel is going faster, right?

so you could have two separate motors driving a link the same way, with different gear ratios

when you are going slow, both motors will be engaging the wheel, and the high torque motor will be doing most of the work - once the machine gets beyond the top speed of the high torque motor its sprocket will freewheel (like the pedals on your bike) and the high speed motor will take over.

Only drawback, you cant go backwards, unless you add a reverse gear.

[KenWittlief]

BTW - the best way to drive a bot with no gears or pulleys is to use a motor that spins at the RPM you need for your wheels.

There are several electric scooters on the market now, with the motor designed into the hub of the wheel itself.

Of course, we have to live with the motors that FIRST gives us :c(

[sanddrag]

Originally posted by KenWittlief
when you pedal a ten speed, as long as the wheel is going slower than the pedals, the pedals stay engaged. If you go down a steep hill, and dont pedal faster, then the pedals are just spinning, and the back wheel is going faster, right?

Only drawback, you cant go backwards, unless you add a reverse gear. If I'm not mistaken the majic behind that is called a one-way bearing. Interesting how you described it being applied to a robot. Instead of the one-way bearing, what about something like manual locking hubs on an old 4x4 truck/suv?

[KenWittlief]

I dont think its called a one way bearing, its more of a cog. Ive taken them apart to grease them. It looks like an inside gear, with the teeth square on one side, and sawtooth shaped on the other.

The tab that engages it is spring loaded - when you pedal forwards it catches, when you pedal backwards the tabs slides on the sawtooth shaped edges.

Now that I think of it, the high speed motor doenst need this, only the high torque motor. So you could have a two motor tranny and use the high speed motor to go backwards when you need to.

[Al Skierkiewicz]

More accurately, this is a one way clutch. It is the same design as a ratchet wrench. The wrench is just able to change direction as well as apply power in one direction and slip in the other.

[Jnadke]

Originally posted by KenWittlief
The tab that engages it is spring loaded - when you pedal forwards it catches, when you pedal backwards the tabs slides on the sawtooth shaped edges.

Now that I think of it, the high speed motor doenst need this, only the high torque motor. So you could have a two motor tranny and use the high speed motor to go backwards when you need to.

As said above, this is a one-way clutch. However, there are a few problems with your design:

1. The high-speed motor would be constantly operating in near-stall condition when accelerating, wasting a lot of electricity.

2. No reverse direction, unless you use a shifting clutch design that reverses the rotation of the gears. You cant just have the one-way clutch on the low-speed motor, because when you rotate the clutch in reverse, the engaging mechanism will catch. Think about it.

[KenWittlief]

Its ok if the high speed motor is always engaged - if you know the speed of the robot, you can contol the power to each one seperately - and you wouldnt 'lean' on the high speed motor until your speed was within its desired range.

and when you reverse, you would power both motors - the high torque motor would freewheel (not catch) and the high speed motor would have all the load. You would not have a lot of torque in reverse, but you could at least back up when you need to

or maybe you could use the ratchet wrench mechanism - make it reversable - then you have high and low torque in forward and reverse directions - you just need to have a shift button, like a car does.

The more I think about this idea, the more I like it. Its simple and easy to implement.

[Jnadke]

Although I am certain you could make it work, the design is turning out rather complex. Overall, if you put the drill (in high) along with the chiaphua directly to a shaft, with the chiaphua on the clutch bearing, the free speed of the chiaphua is 3.5x the drill. This is probabaly the simplest method to make it work. Then just gear it down from there to the output shaft.


It might just be easier to use the Thunderchickens 2002 CCT. With the stronger motors, and higher current constraints, coupled with the use of a worm gear with a 11 degree lead angle, it might solve the problems they were having. The overall theory is quite similar to what you are trying to do, but how it works is very different.

[KenWittlief]

but putting two motors on one shaft, then gearing it all down after that, gives you a high torque, low speed drive train

that is the whole point - you can have the drill motor, geared down low (top speed of maybe 5 MPH) and the chalupa motor, geared up around 10 or 15 mph

and you will have a drive train that will be able to spin tank treads on carpet when you need to move mountains

and fly like the wind when you need to be fast

all without any shifting, and without overloading any of your motors.

[sanddrag]

Why is everyone thinking of having one motor engaged only some of the time? Besides current draw, I can't understand how this would be advantagous over having all the motors engaged and powered all the time.

[KenWittlief]

to get lots of torque, like you want to play king of the hill, you need your drive train geared way down

if you gear a motor way down then you cant drag it up to a higher speed with a second motor - it can only go so fast and that is your top speed

to go fast you need to be geared up - to have both on your drivetrain you need a transmission that can shift gears, or a variable speed transmission.

Using this one way 'ratchet' clutch idea lets you have two motors, geared at the oppose ends of torque and speed, without the need to shift gears on the fly, or to stop and shift gears.

thats why! :c)

[KenWittlief]

Its kinda the same idea as the solid rocket boosters on the space shuttle - they give lots of thrust for the first minute, then they disengage and the rocket on the shuttle itself provide less thrust, but higher speed, to reach orbital velocity.

[DougHogg]

I am wondering if any of this could be applied to our robot for post season competitions.

We are using the 2 CIM motors with custom non-shifting gear boxes and chains going to 2 set of wheels. We shift to high gear by lowering 4 large wheels and raising the 4 small ones.

We have one motor that we aren't using in our drive train or one of our 3 arms: a drill motor.

I had thought of using it like a rocket booster. I guess we could change the gears on our wheels so our CIM motors would give us higher speeds and then use the drill motor as a low speed pusher that would rachet at high speed. Alternately, we could use the drill motor as a booster for high speed and "not lean on it" at low speed. I personally don't have the experience/knowledge to design it. Also it seems tough to stick something like that into an existing robot. However any ideas would be appreciated.

(There are pictures of our robot and our drive train on our web site as listed below.)

[Jnadke]

Originally posted by sanddrag
Why is everyone thinking of having one motor engaged only some of the time? Besides current draw, I can't understand how this would be advantagous over having all the motors engaged and powered all the time.

The Chiaphua alone spins at a free speed of 5,500 RPM with a stall torque of 2.45 Nm. The Drill motor spins at a free speed of ~20,000 RPM with a stall torque of 0.80 Nm. As you can see, the easiest way to set this up would be to put the Chiaphua on a clutch bearing, and gear both motors to a shaft with no reduction. Then, you'd gear from there normally as if you were gearing the chiaphua alone for high-torque. It'd be easier, however, to do it the other way around and use the drill with bosch gearbox, for compactness. A 50:1 gear ratio is a lot harder to create than a 15:1 gear ratio.

As for your question, the reason you wouldn't want to operate the high-speed motor all the time is because, in the area of the high-torque motor, it is going to draw anywhere from 80A to 120A. Obviously, the motor will trip a few breakers. Also, in this range the motor is less than 30% efficient. A lot of the current is going to waste heat. As you know, as a motor heats up, its resistance becomes higher. Therefore, as the high speed motor heats up, it becomes less powerful. Also, more heat means you're closer to letting the magic smoke out.

[Madison]

Originally posted by KenWittlief
Its ok if the high speed motor is always engaged - if you know the speed of the robot, you can contol the power to each one seperately - and you wouldnt 'lean' on the high speed motor until your speed was within its desired range.


Couldn't you establish a correlation, in programming, between the output values of your high speed motor and your high torque motor?

So, rather than operating the high speed motor at near-stall loads under 12V, you could 'artificially' slow it down by applying less voltage via the program.

The free-speed we associate with each motor is for 12V only. If we reduce the voltage to the motor, we reduce it's free-speed. So, if I understand how this might work and what you are saying correctly, it's possible, I think, to make it so the high-speed motor sees little load at low speeds by matching it's potential free-speed to the actual speed, under-load, of the high torque motor.

Does that make sense?

Edit: I really like this idea.

[KenWittlief]

M.Krass - yes, you have the right idea. You could also drop the voltage (PWM output) to the high torque motor when the robot is going fast, or even stop it completely - since its not engaged, its only spinning and wasting power.

The thing I like about this idea the most, you could do this without the oneway clutch (ratchet) with gears and/or chain drive , and to make it a two speed self shifting tranny, all you need to add is one moving part - the little spring loaded cog level thingy.

One extra moving part - you cant get any simplier than that! :c)

[Madison]

Originally posted by KenWittlief
The thing I like about this idea the most, you could do this without the oneway clutch (ratchet) with gears and/or chain drive , and to make it a two speed self shifting tranny, all you need to add is one moving part - the little spring loaded cog level thingy.

How could this work without the one way clutch? If there's no clutch to slip when the output spins faster than what the high-torque motor is capable of at free-speed, that motor will drag and be spinning faster than it's free-speed.

...or, because you're making it spin faster than it's "no-load" spin mean that there won't be any appreciable effect on on the output? That is, at its free-speed, the high-torque motors does no work, so, conceivably, it requires very little work to keep it spinning at that speed? Would spinning it faster than it's free-speed damage it?

[Jnadke]

Originally posted by M. Krass
How could this work without the one way clutch? If there's no clutch to slip when the output spins faster than what the high-torque motor is capable of at free-speed, that motor will drag and be spinning faster than it's free-speed.


I think he's talking about using a reversible clutch bearings, like the one's used in socket wrentches. There's no way this idea could be implemented without the use of a clutch bearing.

I've thought about using such a design before, but ruled it out because reversible clutch bearings are rather complex.


As for spinning a motor faster than it's free speed, Lenz's law prohibits this. If a motor is electrically connected, it will act as a generator, creating a resistive force to the object that is turning it.


Although, the speed controllers do technically have the ability to disconnect a motor electrically. You could implement the idea without a clutch bearing as long as the 2nd motor is off when you try to use the high-speed motor. The negative side of this is that you'd be adding to the overally inefficiency of the drive system because you're rotating more mass inside the motor. Also, the brushes and bearings add inefficiencies. Overall, your efficiency would probabaly drop by 20%. I'd only do this if the high speed motor is 2x the low speed motor, because of the inefficiencies.

[KenWittlief]

what I was trying to say is you can build a drive train with two motors driving one shaft (or wheel) that is not self shifting

and to turn it into a self shifting two speed system, all you have to add is one moving part - the spring loaded tab that locks into the gear in one direction only.

This is not complex - its extreemly simple. Get you hands on the back wheel from a ten speed bike and take the rear gear cluster off (all the bearings will fall out - use a junk wheel)

You dont want to do this without the one way clutch - spinning a motor past its top speed, loaded, connected or otherwise, is still a huge drag on the driving motor.

and besides, motors have a physical top speed - if you spin them any faster the centrifical force rips it apart.

[GregT]

Why so complecated? Its possible to make a efficient CVT using only gears, its not used in cars because it requires 2 torque sources.

greg

[Adam Y.]

How could this work without the one way clutch? If there's no clutch to slip when the output spins faster than what the high-torque motor is capable of at free-speed, that motor will drag and be spinning faster than it's free-speed.
Actually I do not think you would destroy the motor by spinning it faster than its free-speed. The speed-controller is another story.

[Jnadke]

Originally posted by GregT
Why so complecated? Its possible to make a efficient CVT using only gears, its not used in cars because it requires 2 torque sources.

There is no way (that I know of) to create a CVT using only gears.

A real CVT allows for infinitely variable gear ratios (or at least between two points). Why? If you look at a motor curve, typically maximum efficiency (ratio of input power to work performed) is at 75% of free speed (25% power). The theory behind the CVT is to keep the motor operating at this free speed and vary the gear ratio during the acceleration phase.

The only CVT's to date are the Belt CVT and the Toroidal CVT. There are also hydro-mechanical CVTs.

[ChrisH]

Originally posted by Jnadke
There is no way (that I know of) to create a CVT using only gears.

I'm not sure if it would be considered a true CVT or not. But you can use an epicyclic or planetary gear system to create a similar effect. As GregT mentioned, it does require two torque sources and the ratio is changed by changing their relative rotational speeds.

I have a little hand operated demo version our machinist built that shows the concept pretty well. We were going to use it in 2002 but it didn't fit the robot concept. This year we didn't feel we had the weight available so we haven't put it into practice. Maybe next year, now that we feel a little more confident building gearboxes.

The Thunder Chickens' Crazy Chicken Transmission of 2002 worked on this principle. I think that the Gila Monsters used it this year. There may be others out there that I don't know about, but FIRST teams have done this in the past and it has been effective.

I'm not sure how efficient it is. Paul Copioli posted a white paper a while back that indicated the power of the second motor is basically lost. I went over the equations and didn't see anything obviously wrong, but a different configuration may require the equations to be set up differently and so give a different result. I haven't gone back and checked this myself, yet. I may get around to it someday, but I have a different area of interest at the moment.

[Jnadke]

Originally posted by ChrisH
I'm not sure if it would be considered a true CVT or not. But you can use an epicyclic or planetary gear system to create a similar effect. As GregT mentioned, it does require two torque sources and the ratio is changed by changing their relative rotational speeds.

I don't think I'd consider that a CVT. Yes, I've studied it before, and even created my own design. It's merely a dual-input viariable speed gearbox.

As for efficiency, in high speed it is extremely inefficient, because you must provide power to the central motor in order to give a surface for the high speed motor to react from. But it does serve its purpose since we have a limited set of motors to work from. I suppose you could make it fairly efficient if there was only a 2:1 difference between the two motors.