pushrods and flex cable?

[Hank Fletcher]

Sorry for the overly mechanical other than electronics request, but I'm still on the lookout for this stuff for my recorder robot. I thought maybe some forum members might have more experience in this area of R/C models than I do (I regret not having the opportunity to explore that hobby in the past more!).

I think the pushrods I'm okay finding, but if anyone has any favourite suppliers for that sort of thing, or tips on the best products, I'd sure appreciate it!

For another potential project, I was wondering if there's any hobby-type product in terms of flexible axles? So instead of a pushrod going back and forth, it transmits rotary power like the flex cables that come with rotary tools? Is there such a thing?

 

[3v0]

Electrc seat drive cables (from cars) and spedo cables might be what you are looking for in the flex drive shaft area.

On another tack. Have you looked into using memory/muscle type wire to activate you musical devices? The major problem is that they eat a lot of current but if you are not on batteries it does not matter. Speed can be hand by using several thinner wires etc.

 

[Hank Fletcher]

On another tack. Have you looked into using memory/muscle type wire to activate you musical devices? The major problem is that they eat a lot of current but if you are not on batteries it does not matter. Speed can be hand by using several thinner wires etc.

I did, but came to the conclusion that they'd be too slow, I think? My understanding is that they rely heavily on ambient temperature to return to their original shape, and as such can also have an unpredictable range of power/movement depending on the environment.

My recorder robot fingers need to react and operate fast, faster even than I think a normal servo will permit. But the fingers need only be either fully pressed or fully released, at least for the first prototype of the recorder robot. I'm thinking some inexpensive DC motors connected to the pushrods, and some simple/efficient/fast way of latching the motors in either the pressed or released position if the finger doesn't move for longer than a second or so. This would limit the amount of power wasted in a stalled motor. Any ideas on what I might consider for the latching mechanism?

Thanks for the idea about the seat motor cables - I didn't realize they worked that way! I've been coveting that motor and waiting for a long time for my Sable to finally die so I can justify tearing it apart for the bits and pieces!

 

[jpanhalt]

For another potential project, I was wondering if there's any hobby-type product in terms of flexible axles? So instead of a pushrod going back and forth, it transmits rotary power like the flex cables that come with rotary tools? Is there such a thing?

Last question first. Torque rods were used in the ancient past (1950's early 60's), but have been replaced in aircraft with push rods and pull-pull cables (like full size airplanes). For going around bends, we use braided cable in a sleeve or pull-pull cables. When you put a cable in a sleeve, both ends of the sleeve need to be fixed to the structure.

Edit: The above comment relates to transmission over fairly long distances. There is a "rotary actuator system" that uses a torque rod, but the servo is located close to the control being moved. Some modelers are also attaching the servo directly to the control.

As for the type of pushrod, I use almost exclusively carbon fiber rods about 1/16 to 3/32 inch diameter. They also need to be in a sleeve for long distances. For non-sleeved applications, I use carbon rod (0.200 or so). Wood dowels and squares also work well where weight is an absolute premium and strength is not too important. Exotic modelers (like for indoor craft) use boron fibers, which are even more stiff than carbon wt/wt. John

John

 

[Ambient]

If you need hardware like blind nuts, screws, clevises, bolts, get Great Planes brand. For pushrods I prefer Sullivan. They are very smooth and flexible, and have different sizes. I have been using the gold-n-rods from Sullivan for all my airplanes.

I have a steerable nose wheel assembly in anyone is interested. And a Focus 3 ss Tx w/servos, Rx. AM.

 

[Ambient]

...My recorder robot fingers need to react and operate fast, faster even than I think a normal servo will permit. But the fingers need only be either fully pressed or fully released, at least for the first prototype of the recorder robot.... Any ideas on what I might consider for the latching mechanism?

How about solenoids? Those are very fast. If you need strength as well then you can try out pushrods connected to a car door lock mechanism. But that would get kinda bulky I imagine. You could also use stepper motors and winch setup. But all in all I would just use some fast servos. You can use the long servo arms and that will give you a good amount of travel, at least 2'' I think.

 

[Hank Fletcher]

How about solenoids? Those are very fast. If you need strength as well then you can try out pushrods connected to a car door lock mechanism. But that would get kinda bulky I imagine. You could also use stepper motors and winch setup. But all in all I would just use some fast servos. You can use the long servo arms and that will give you a good amount of travel, at least 2'' I thin

I started but then gave up on solenoids. I was just disappointed by how inefficient they were - lots of heat, energy suckers, and unreliable in terms of magnetically sticking in the activated position when energized too long. I thought about the car lock and the steppers, too. Trouble is, being a music application the amount of sound the motor/mechanism makes is a bigger factor than most other applications have to be bothered with.

What do you mean by "fast" servos? I'm a little ashamed to admit that I've never worked with servos before, but I'd thought that all hobby servos only refreshed their position every 50ms (at 20Hz, or do I have those numbers backwards?). I've seen servos with plenty enough torque/speed to get the job done, I'm just wondering about the latency, and consequent coordination between multiple servos.

Just to give an idea of the ideal I'm aiming for:
An allegro tempo in music is 120 beats per minute, or 2Hz. The beat in music is often called a quarter note, which can be subdivided into further fractions. So at an allegro tempo, a sixteenth note would be a note that occurs at a resolution (can I call it that?) of 8Hz (or 125ms = 500ms quarter notes divided by four). The sixteenth note resolution at 120bpm is what I'm aiming for in the prototype. Typically, the note needs to be sounded (with the fingers in a holding position, whether each finger is used or not) for at least half the value of the note (this is close to normal practice for articulated notes at this speed by a real musician). What that means is that at its fastest speed, there is only 62.5ms for the servo to react and move each robot finger to its new position between different notes.

That might be my ideal for the prototype, but suppose I were to say I'd be satisfied with something that worked at half or even a quarter of the reaction speed? Would it still be possible to get the servos at least coordinated so that they refreshed simultaneously? What I don't want to happen is for the servo/fingers to get out-of-sync, as it were, so that the wrong fingers are going down at the wrong time. Or am I thinking about this all wrong, or too much?!

 

[Hank Fletcher]

I'm thinking some inexpensive DC motors connected to the pushrods, and some simple/efficient/fast way of latching the motors in either the pressed or released position if the finger doesn't move for longer than a second or so. This would limit the amount of power wasted in a stalled motor. Any ideas on what I might consider for the latching mechanism?

Sorry, I'd forgotten that I'd already come to a conclusion as to what would be sufficient in terms of a "latching" mechanism for the recorder robot. If I'm using a geared servo, or geared motor, then I think that will be sufficient for keeping the finger securely pressed or left released with no power.

Another question about the servo option:
I'm wondering how I'd detect whether the finger was pressed or released sufficiently? I guess I could do this by just controlling the servo and setting it to a satisfactory position, but is there a way to get more feedback? For instance, if I was just using a geared DC motor, I was thinking of reading the change in current as the motor stalled - can you do that with servos, too?

 

[jpanhalt]

What do you mean by "fast" servos? I'm a little ashamed to admit that I've never worked with servos before, but I'd thought that all hobby servos only refreshed their position every 50ms (at 20Hz, or do I have those numbers backwards?).

It's 20 mS and 50 Hz.

The term"fast servo" refers to the speed of rotation of the top wheel. Some are faster than others.

Would it still be possible to get the servos at least coordinated so that they refreshed simultaneously?

That would seem possible with a microcontroller.

What I don't want to happen is for the servo/fingers to get out-of-sync, as it were, so that the wrong fingers are going down at the wrong time. Or am I thinking about this all wrong, or too much?!

As for synchronizing the movement of several servos, that might be a bit more difficult. However, some of the larger models have separate servos on the elevator for redundancy. I have read that some fliers put the servos on separate receivers, yet the airplane flies well coordinated. That is, one elevator half is not moving differently enough from the other to make a visible difference. That application is not nearly as stringent as a musical application.

On the last question about monitoring push, consider measuring current flow in an analog servo. That relationship might be proportional to force.

John

Edit: Just reread the question about push. If you are looking for positioning accuracy, go with a digital servo. Moreover, if they are held just a little out of position, the current demands go up very steeply.

 

[3v0]

I did, but came to the conclusion that they'd be too slow, I think? My understanding is that they rely heavily on ambient temperature to return to their original shape, and as such can also have an unpredictable range of power/movement depending on the environment.

A few people have been working on speed. I seen reports of 30-40Hz using water cooling.

Maybe using two wires, one to close and another to open? Both in the same path but on opposite sides of a pivot or pivots.

To a large extent I know that it is a technology looking for an application. The stuff fascinates me.

 

[Hank Fletcher]

Thanks for the servo tips, jpanhalt. I think I'm going to go with the gearmotors for the recorder robot after all, though, just taking into consideration cost and the fact that I really just need something for two positions (pressed or released) rather than the finer control over the positions in between. But while were on it, how do continuous rotation servos work? Does the difference in the PWM signal just change the speed, or is it control over 360 degrees of rotation? Is there any feedback for counting rotations? I know there are servos with encoders attached, so I guess I'm just asking about servos that don't have encoders as well here.

3v0, I've said it before: there's something cool about the muscle wire idea. I just don't think it's something I have the patience or funds to experiment with for this project.

My question about the flexible cable for rotation transmission was me just thinking about building a walking robot. I've been mulling over the options in my mind for powering the various leg joints. My thought is that the best way to go about building a humanoid type walking robot is to have the weighty power (e.g. motors and batteries) in the torso, while keeping the limbs quite light, especially at the extremes (e.g. the ankle). That way the centre of gravity stays closest to the middle (I think!), and will be more efficient than, for instance, moving heavy feet.

But how to transmit the power to the various joints? Muscle wire? Hyrdraulics? Pneumatics? Motors at each joint? Axles connecting each joint to a central torso? That's what I was thinking about for the flex cable: a simple method of transfering the power from a central torso motor to the extremity joints, which themselves perhaps just consist of a worm and spur gear combination. I wonder if even pulleys and strings might be the better way to, at least for keeping things light to the extreme joints, but I'm having a hard time getting my head around that solution, for instance, whether cases of various independent movement of the joints might interfere with that type of solution.

 

[3v0]

3v0, I've said it before: there's something cool about the muscle wire idea. I just don't think it's something I have the patience or funds to experiment with for this project.

I understand. It can be spendy. The stuff is much cheaper in bulk and even more so if you train your own wire but then there is that patience thing. I would like to play with it but the intro kits are not what I would call a value.

...
But how to transmit the power to the various joints? ... I wonder if even pulleys and strings might be the better way to, at least for keeping things light to the extreme joints, but I'm having a hard time getting my head around that solution, for instance, whether cases of various independent movement of the joints might interfere with that type of solution.

There are so many ways to go. Choice to some point would depend on scale.

The pulley string thing is fun to think about. In part it depends on how many joints you want in the arm/leg. The first two are easy.

0123456789

 

[Pommie]

Hank,

To get a pic to control 14 (18 pin pic) servos simultaneously and receive data from a PC is really trivial especially if the servos only have 2 positions. You should be able to get servos for less than $10 each.

Mike.

 

[Hank Fletcher]

To get a pic to control 14 (18 pin pic) servos simultaneously and receive data from a PC is really trivial especially if the servos only have 2 positions. You should be able to get servos for less than $10 each.

Mike.

Trivial for you, maybe! But still, I'm starting to get my head around it. Even if I went with servos over the gearmotors, I'd still want some method of feedback to know that the fingers are pressing firmly against the holes of the recorder. I figure measuring the current of the motor as it starts to stall would be a good way to do that?

Up to now all I've been able to figure out is programming a PIC with code that contains the information of, for instance, a single song. I really haven't done much into exploring options for interfacing to the PC for re-programming new songs into the PIC, or having the PIC connected to some sort of memory for downloading new songs from the PC. I've got some ideas for translating MIDI files into PIC-adapted data for the recorder, but I still don't know anything about connecting memory chips to the PIC and the PC for storage. I should say that I'm expecting the recorder to operate autonomously to PC control after it's programmed, in case that's not obvious by now!

 

[Pommie]

The beauty of servos is that they go to where you tell them to go. So, you have a weak spring holding the finger up and a strong spring from finger to servo, you can drive the servo past the contact point until the strong spring exerts enough pressure to form a seal. If the servo is too slow then you put a longer arm on the servo. If it is then too weak, you get a higher torque servo. The point I'm trying to make is that you don't need any feedback method.

If you get the mechanics sorted out then I am happy to help with the codeing part. In a previous life I wrote a few music drivers for various computer games.

Mike.

 

[Omar.M]

But while were on it, how do continuous rotation servos work? Does the difference in the PWM signal just change the speed, or is it control over 360 degrees of rotation?

The continuous rotation servos are just normal servos with the potentiometer glued in place and the little tab on internal gears cut off (which is there to stop the servo from moving more than X amount-- used for position).
With continuous servos, PWM signal changes the speed of the rotation, instead of being able to control the position of the motor.

Perhaps I am on a different page here, but I don't quite think that you would require feedback from the servos (if you went that route)-- just find the position that would be covering the hole, and get that pulse length.

Technically, you'd only need to move a tiny bit away from the hole for it to be in the "uncovered" state (thus distance between "covered" and "uncovered" state is small, meaning you could play the recorder pretty fast).

-Omar
Edit: Aw shucks, Pommie beat me to it!

 

[jpanhalt]

After a little more thought, I think the suggestion of using a spring loaded cover will be the simplest, rather than trying to develop feed-back loop to sense pressure.

We have all seen comic pictures of the robot trying to pick up a soft object, with the end result being it it squeezed to pieces. Servos go to a position, and if that position is just a little past where it should be, say because of temp drift of the amplifier, the servo will exert its full force to try to get there.

Attached are some pictures of a application where the servo moved a lever. The "up" position was not critical. The down position was spring loaded, like I would imagine your valve covers to be. It used minimal machining. I simply brazed a wire to a bushing that ran on a shoulder screw. Everything could be bought off the shelf, except, to save cost, I did turn the bushing myself.

A second suggestion has to do with mounting the servos. I assume there are about 6 holes in a recorder (never played any wind instrument), which means 6 servos. The cheapest hobby servos need to be mounted on rails, and if you refer to the geometry needed for the levers, that might be a little complicated for 6 servos. Servos are also made with the mounting lugs at 90° to what you see in common ones. They are often called "wing servos," and they are mounted on the flat side. Another option is simply to roughen the surface on the flat side and use an adhesive. I have had poor luck with double-stick tape, but have had good luck with hot-melt glue and two-part flexible adhesives. You can also use heat-shrink around a mounting board. Anyway, I think that mounting on the flat side would allow all servos to be mounted on a single board or metal plate.

Sorry for the length of these comment. John

 

[blueroomelectronics]

What would be nice and fast; perhaps would be what hard drives use. A voice coil motor.
**broken link removed**

 

[Hank Fletcher]

Sorry for the length of these comment. John

No problem, and thanks for the ideas. Nice photos!

What would be nice and fast; perhaps would be what hard drives use. A voice coil motor.

Thanks for the tip. I don't know much about them, but now have something to do with my old hard drive that fried awhile back!

 

[Ambient]

I just ordered myself a ssc-32 from lynxmotion.com. It will control 32 servos. You can send it serial commands for each servo from a PC or uC, and even tell it to move a bank of servos together. Each will move at different speeds(if thier positions are different) to get to their end points at the same moment. It's a pretty cool device. I will be using it to simplify my hexapod robot design.