Free turning shaft - Basic scheme

[atferrari]

My mechanical knowledge / ability is very limited so bear with me.

I want to built a small platform (for a robot) supported by three wheels. I need shafts A, B & C, able to rotate in any sense with no limitations. Ever heard of "synchro drive"?

https://i1000.photobucket.com/albums/af124/atferrari/shafts.jpg


Can anyone suggest a simple scheme showing what kind of bearings / bushes do I need thinking of a light platform, no more than 1,5 Kg batteries and electronics included?

As a reference, the platform is of square shape, 30 x 30 cm, aluminium maybe 4mm thick.

In the future each wheel will have its own motor / battery and slip rings to receive the on/off commands.

Would anyone be so kind to show a basic scheme to help me to realize how it should / could be?

 

[dknguyen]

I know of it as Omni-Drive. This is what you mean right? A three-wheeled holonomic base? Of course, this can be done with more than 3 wheels.

**broken link removed**

Use the "Search" on these two websites for "Omni-wheel" to get a bunch of results for platforms, wheels, and related parts. You can also find good gearmotors on these websites too.
http://www.robotcombat.com/store.html
http://www.superdroidrobots.com/shop/

But you made a couple statements that seem really misguided. Why on earth would each wheel need it's own battery? Just route power from one big battery to all the motors. With dedicated batteries to each motor, the motor that is being worked hardest would die first leaving your robot dead in the water.

And why would you need slip rings for anything? Slip rings are used to transmit power or signals across a continuously rotating joint. They're insanely expensive, unreliable, delicate, and really hard to find. They're only used when you have absolutely no other alternative. Like sensors sitting on a wheel, propeller, or rotor, or rotating pan-tilt head assembly. In most of these cases, there are usually much simpler approaches to mounting an electronic component on the spinning part (the one exception is the pan-tilt head assembly).

The only things you should need is:
- a base
-omni-directional wheels
-motors to mount them on
-reversible motor controllers (preferably with speed feedback) for each motor.
-a computer or controller smart enough to control the motors to move in a certain direction

 

[dknguyen]

Or are you talking about this:
Robo

A holonomic base that uses regular wheels that each have independent steering?

If that's the case, what I said about using slip rings and one battery for each motor still holds true. You don't need slip rings for the wheels because the wheels don't have to have an infinite steering rotation. Slip rings are also delicate so having them right on the load bearing shaft that takes bumps and shocks isn't a good place for them to be. And using one battery for each motor still ensures that the motor that is being worked the hardest dies first leaving your robot immobile.

Personally, I would favour the omni-drive approach (non-steering omni-wheels that control direction by spinning them at different speeds) rather than the synchro-drive approach (regular wheels with independent steering) if you are only driving on flat ground because it has a lot of advantages. It's simpler, lighter, stronger, and more reliable than the synchr-drive. It can also change directions instantly since it doens't need to turn wheels to steer. It's biggest downside is it can't climb over rough terrain very well in which case the synch-drive is better.

The problem with the synchrodrive is it has a lot of moving parts. What you need is:

-a compact wheel and drive gearmotor assembly
-a way to mount the wheel/motor to the steering mechanism.
The easiest way would be just to mount the whole thing on a small plate and mount the plate to the steering servo. It's also the ugliest way but most other ways take custom machining to directly mount the motor on to the steering mechanism
-a rugged strong steering servo. These are hard to find and still likley not as strong as you want them to be. The strongest method would be an indirect drive. A steering shaft rigidly bolted to the wheel/drive motor assembly that runs to the main base through a bearing so that it can rotate in the main base. Then you drive the steering shaft with the steering servo through some gears, belt drive, or chain drive. You lose some steering accuracy though with a belt or chain drive, or poor gear drive.

Most of those parts are pretty hard to find. Particularily the indirect drive or steering mechanism and the wheel-drive assembly if you want something that doesn't look ugly and bulky.

 

[atferrari]

You seem to have spent lot of time in replying what I thanks.

I know what synchro drive is (that's why I mentioned).

Please note that I am asking for suggestions on how to implement the shafts that would turn more than 360º.

Could you help with a scheme even if rough? That is all what I need by now.

Gracias.

 

[dknguyen]

Hold on. I'm trying to look for parts that can make things a bit easier after I finish my lunch.

But you don't need shafts that turn more than 360 degrees. 360 degrees is enough to go in any direction you want. Actually, 180 degrees is enough if your motors and motor drivers are bidirectional. If they are uni-directional than you need 360 degrees. Why do you want them to turn more than 360 degrees?

 

[dknguyen]

A big problem seems to be a good gearmotor and wheel combination. The problem is that you steer the wheel you want the wheel to be directly centered on the axis it is rotating around when you steer it. If it's off center (like it is with most wheels that have a big long motor sticking out the side), then when you steer the wheel it would have to travel in an arc. That requries a lot of torque and makes a lot of skidding which wears down components.

Well...maybe you could get a metal plate cut some REALLY big holes into it where the wheels are supposed to go and install these things:
**broken link removed**
Even the smallest one is a bit large, maybe smaller ones exist. You could have steering servos turn the assembly from the top and the turntable would take the side loads off of the steering servo.

Other methods tend to make wheels at the end of legs since you need a shaft to take the side-load so the servo doesn't have to.
How to Build a Robot Tutorial - Society of Robots

Even in that design you see the wheel isn't centered on it's steering axis so there's going to be skidding when you steer the wheel. Look closely at the inside of the front and rear wheels that can steer on the Mars Rover:
https://photojournal.jpl.nasa.gov/jpeg/PIA04413.jpg
That little horizontal circular hing just before the motor that actually spins the wheel is the steering motor. They went to a lot of effort to try and get it as close as possible to the middle of the wheel's width so that the wheel could be steered without a lot of skidding. Even then it's not perfectly centered.

Trying to get the wheel centered in the steering axis seems to make the robot taller than most becuase the wheel has to be under the robot, rather than on it's side. It doesn't seem to lend itself well to a flat plate design.

 

[dknguyen]

These seem to be rather nice motors and wheels that might work (strangely enough one of the hardest things is figuring out how to mount a wheel to a motor):
**broken link removed**
**broken link removed**
**broken link removed**
**broken link removed**
**broken link removed**
**broken link removed**

These should allow you to mount a motor and wheels to a small metal plate that you can bolt onto the bottom of this:
**broken link removed** and then bolt that onto the base.

And this is where a lot of the pain comes in for the synchrodrive. It's really hard to get the wheel centered so that it does not skid or travel in an arc when you steer it. And it also means you need the wheel to be under the robot but the wheel can't hit the underside of the robot when it is steered so the robot has to be taller. It might mean you need to find a way to raise the height of the motor so it can mount underneath the base so the wheel can steer without hitting the bottom. maybe by mounting the motor inside a plastic or metal enclosure and then mounting that to the turntable.

In the flat base you can cut large wholes (just like in the turntable) where the turntable is mounted. Then you can mount a sprocket on top of the motor-wheel-plate/box assembly so that it passes through the hole in the turntable and the flat base.
**broken link removed**
Then on the flat base itself you can mount your steering servo and use another sprocket and chain to allow the servo to steer the wheel-motor assembly.

 

[atferrari]

Helpful post. Thanks

Your last post helped a lot to put me in the picture, regarding the assembling of the shaft.

Servocity seems to have a complete line of soluctions. Or at least they seem wise anough to show them quite logically, have to say.

I spent a lot of time trying to understand how the shafts of the Taurus were actually assembled.

What I do not know is what are the parts enclosed in the plate at mid-height between the top servo and the wheel. I expected that to be a plain shaft but it seems there are other parts that I do not recognize. Could you say something about?

 

[dknguyen]

The taurus legs are an outer tube formed by bending a piece of sheet metal into a U. The open end of the U closes off to form a tube by being screwed into sheet plastic that forms the rocker legs on the main chassis. It doesn't show explicity this in the photos but if you look at the images of the complete robotyou can see that it has another sheet of plastic 90 degrees to the main plastic sheet of the rocker to make a "T". The top of this T goes against the U to form a closed tube.

From there it just looks like he runs a shaft where one end is glued onto the side of the wheel servo modified for contiuous rotation, and the other end is glued onto the face of the sero mounting disc. There are those hubs and shafts on servocity that would let you do something similar. I don't know what the inside of the tube looks like, but is best to have a bearing or something in contact with the tube and the shaft. That way side loads on the wheel are transmitted to the wheel servo which is transmitted to the outside of the tube (part of the main frame of the robot) rather than the axle and bearings of the steering servo.

 

[dknguyen]

The parts on this page might also be of interest to you:
Lynxmotion - Servo Erector Set

In particular are these parts:

Lynxmotion - Low Profile Axis (no servos)
Use for the steering servo to take the sideloads off of the servo without the need or imprecision of a bulky belt or chain drive. If your robot is small enough you actually might be able to just have a single shaft without the need for a protective supporting case.

Use these to connect the shaft that runs from the steering servo down to the wheels and drive motor to give enough clearance between the wheel and base so the wheel can steer 360 degrees without hitting the base.:
Lynxmotion - Tubing and Hubs

To mount these shafts to the face of a servo or the geared indirect servo drive I linked to above you would use this:
https://www.lynxmotion.com/p-403-aluminum-tubing-connector-hub-pair.aspx

If you don't want to use continuous rotation servo motors for the drive motors (likely to be stronger and more rugged than any servo you can find that is modified for continuous rotation) then these should come in handy for mounting the motors onto the end of the shaft, whether or not there is just a steering shaft alone or a protective load bearing outer shaft:
https://www.lynxmotion.com/p-625-ses-aluminum-motor-mount-pair.aspx

Similar to mounting the shaft onto the face of a servo or the indirect servo drive you would use one of these at the end of the shaft to bolt the motor mount to:
https://www.lynxmotion.com/p-403-aluminum-tubing-connector-hub-pair.aspx

Pretty metal wheels that are easily attached to a servo modified for continuous operation:
https://www.lynxmotion.com/p-735-aluminum-ses-robot-wheel-275d-x-027w-pair.aspx

But if you want to use the wheels with a regular motor you'll need a hub:
https://www.lynxmotion.com/c-42-mounts-hubs.aspx

To remove side loads from the drive motor servo and allow it to be more rugged or your robot to be heavier you might consider using this though I'm not sure how:

https://www.lynxmotion.com/p-484-ball-bearing-hub-kit.aspx

It would basically have to involve running a rigid shaft down to the wheels alongside or around the steering shaft. then having the wheel sit on the shaft through that ball bearing hub. Then the weight of the robot is on the ball bearing hub rather than the servo axle. Details on how to get the servo to spin the wheel while it is on the ball bearing hub is up to you (perhaps drive the wheel from the opposite side). Perhaps using this as a load bearing rigid shaft:
https://www.lynxmotion.com/c-96-alum-channels.aspx

 

[atferrari]

My current idea

Hola dknguyen,

Thanks for your time and explanations. While I digest them, would you take a look to my sketch?

The usual disclaimer of having neither experience nor knowledge in mechanics is valid

https://i1000.photobucket.com/albums/af124/atferrari/BOT130Ideadepataconruedaymotor.jpg

Why steppers?, because I have lot of them available.

It is clear that I am not decided yet on giving up to the ++360º rotation. Yes, slip rings are by far the most difficult.

 

[dknguyen]

Steppers mainly because they are the poor man's solution to a servo motor. Steppers can be run into open loop and have higher torque at low speed without a gearbox than other DC motors. This means lower parts count and cheaper cost. The catch is that since you are running them in open loop with no feedback you must ensure that the stepper is always strong enough to move when commanded. Because if it isn't and slips, you won't know about it. UNless you use an encoder of course, but then that starts to defeat the purpose of low parts and low cost.

You will get better results with a DC brushed or brushless gearmotor with an encoder in a true closed-loop feedback system (though it requires a more complicated motor driver. an encoder, and a gearbox).

My opinion ( my preference) is that if you are going to go with a stepper for the steering motor, perhaps you might as well go with a brushed gearmotor rather than an RC servo for your drive motors. It's just more natural for the RC servo to be the steering motor and if you're not going that route, mgiht as well go all the way. They are more rugged too which is important if you are supporting a heavy robot on the motor bearings. It is rather mechanically complex to offload the weight of the robot onto the frame of the robot through something other than the motor bearings.

Of course, lots of parts for this kind of robot means higher cost but you seem to know that so I won't goad you on that. But what's the point of blocks A and B? Just to increase the thickness of the plate for the bearings? Seems like a rather expensive solution to a simple problem. Otherwise that looks like a very durable design. But you might want to consider the diameter of the wheel. Having the wheel centered on the steering axis is good because you have less skid when steering because the wheel doesn't need to move through an arc when you steer, but so is having a larger wheel because you can roll over obstacles more easily, but you do need a higher torque gearmotor because the lever-arm of the wheel, aka the radius is larger.

If you try to get both of those things at the same time, them the centered motor mount has to be long and thin to accomodate the full height of the wheel. This makes the wheel/drive motor mount long and thin which makes it prone to flex because it is bent sheet metal. Switching the sides that the motor and wheel are mounted on(so that the motor is beneath the steering shaft) lets you have as big a wheel as you want by increasing the length of the steering shaft which is much more rigid than the motor mount is. But now the wheel is off center by a bit, but if you make the motor mount very narrow so the wheel is very close to the steering shaft, the benefit of a larger wheel might outweight the increased skiding arc when you steer. For example, outside on dirt, a larger wheel is definately better because there are more bumps and the loose dirt lets you skid more easily. On flat smooth asphalt, concrete or linoleum floors that don't let you skid so easily, a smaller wheel is definately better becaues bumps are no longer a problem but skidding is. For really bumpy high friction surfaces...well...that's tougher to decide.

It's your call:
-less lossy steering and smaller wheels (easier for a lower torque gear motor to drive but not as good with obstacles)
OR
-more lossy steering with a stronger mechanicaly assembly and larger wheels (need a higher torque motor but better for obstacles).

You might want to add "positive camber" to your wheels by bending the motor mount differently. Then if you mount the motor on the outside you can have a very large wheel while keeping the contact patch beneath the steering shaft. This is used on robots to help counter act the flexing/sagging that can occur due to the weight of the robot.

See "Frame sagging caused by weight" on page 219 and the photo on page 221:
https://books.google.ca/books?id=vY...&resnum=1&ved=0CBgQ6AEwAA#v=onepage&q&f=false

However in your case (where the motor mount is above the motor rather than beneath it like on the pg. 221 photo, this severely limits the length of the motor though the more you bend away from 90 degrees unless you increase the length of the motor mount which starts to add flex again. The angle you choose also restricts you to a narrow range of wheel diameters in order to keep the contact ptach below the steering shaft. This is bad if you want to change out wheels for different applications. See PDF.

Personally, I would probably go with a a large wheel mounted on off-center and go with only a tiny bit of camber, if any, to help get the contact patch more beneath the steering shaft. Any more and it would limit the length of the motor and the diameter of wheels more than I would like. But I also favour outdoor robots than can roll over anything more than indoor robots. Like I said in my first and second post , if I was making an indoor robot that did not need to go over bumps or obstacles, I'd use 3 or 4 fixed omni-wheels rather than a synchro-drive- much simpler and stronger.

 

[atferrari]

Willing to experiment

It took time to disgest your posts, but I feel I have a more clear idea about all this (and now I know what camber and caster is...)

I am willing to experiment a little.

By browsing further, I run across this which seems quite appropriate:

**broken link removed**

What do you think? In line with KISS, isn't it?

My next concern is a serious one: buying the few bits here and there with all the extra costs involved to have them sent to Argentina. And then..., Customs..! Wish everyting I want was sold by just one supplier.

Thanks for your incredible help and so much time spent.

 

[dknguyen]

Yeah I have that problem too with shipping and customs. Most of the parts I sent you can be bought through other websites (well the Lynxmotion stuff anyways). The parts I sent you aren't only available from one source. The links I sent you were the stores run by the manufacturer. You can often find other stores that also carry those parts as well as parts from other manufacturers.

For example, I would NEVER order from Lynxmotion directly because I did so once before and got burned by the fact they only ship with UPS. I would order from www.robotshop.ca. Not only are they based in my country but they carry stuff from Lynxmotion, Servo City, and Solarbotics. They also carry stuff from a crapload of other manufacturers- some low end and some very very high end. Some of the ones more priced for hobbiests are Devantech, Acroname, Parallax, Dimension Engineering, POB, CMU. So I am able to get them all into one shipment and avoid manufacturers that only ship with UPS.

You might want to check them out even though they are even further away than the US because they carry pretty much all the parts we have been talking about. Of course, you might also be able to find other similar stores close by.

For non-Americans, Canadians, and Europeans, their website is:
https://www.robotshop.com/world

The GM3 motor is qutie easy to work with and light. Just keep in mind they are small (which isn't necessarily a bad thing). It just means you have to make the rest of your robot small and light too which makes it cheap. That weight is also well within in the range of a typical cheap servo motor (not the super expensive high end ones which will burn a whole in your wallet).