Looking for sources for air muscle control valves....

got a metal lathe and some aluminum laying around?

--A couple of options come to mind: Festo Pneumatics sells these "muscles" but they're all metric, but then so are you, yes? :-) They have a full line of accessories, but they're pricey. --Depending on how small your project is, it might be possible to make do with some of the stuff sold by Robart, who make valves, etc for model aircraft pneumatic landing gears. The catch is, these would have to be moved manually by a servo or solenoid that you'd have to get seperately from somewhere else. --Trying to remember the name of the third outfit, but IIRC it's robotcombat.com; check under the section for "Sozbot" parts. They make a

3-way (for single acting cylinders) and a 5-way (for double acting cylinders) control valve that can couple directly to the output of an R/C reciever although, IIRC, there needs to be a signal boost from the 3v that comes from the reciever to 5v to operate the valve. I'm using one of the single-acting variety to drive the air cylinder on a 12 lb "hobby weight" bot and it does a dandy job...

Try Princess Auto in Coquitlam. They used to have lots of surplus solinoid valves at pretty decent prices. How many are you looking for, how much can you afford to spend, and what tools do you have access to? It might be possible to build your own valves using Nitinol wire as the actuator. You'd need a lathe, but Busy Bee has 7X14 lathes on sale for $600cdn.

Remove nospam_ to reply by email

Jeff H........

Lies, All lies. Don't believe a word Difool/sayNO says. He fears the truth!

This company makes a proportional nitinol based valve with analog or digital interface. Very nice price as well. They also have a very interesting McKibben's muscle/servo system. Again at a very nice price. (most pneumatic servo systems are VERY expensive).

I've not used them, but am anxious to hear how well other people's projects have gone with them.

Direct link to the servo actuators.

Direct link to the valves alone; Joe Dunfee

Thanks for the link.

I'm building my own air muscles, as they're just a bit cheaper! I originally thought I might be able to use motors or steppers but got into the range of torque/speed/cost where the air muscles looked better.

TJ

Nope.... but it might be possible to get access to one... do you have plans/schematics? And what do you use for an actuator?

Thx, TJ

There's actually a number of ways to build your own air valves if you have a decently equipped shop. A nitinol wire pulling a needle valve would work, but you don't get much movement so airflow might be a problem if you built VERY small valves. You could also go with a diaphram valve made out of .5" lexan and powered by an RC servo. You could even use a servo driving a cam to sqeeze a section of surgical tubing. (use a coil spring around the tube to keep it from expanding except where you sqeeze it) On the other hand, you could simply use air cylinders and solinoid valves. I've got a bunch of cylinders and valves from an old robot I built years ago that I could give you a decent price on.

It would help to know what you're trying to make, if it's mobile or stationary, how much force you need, ect. A pick and place arm for a machine is going to take a lot less matched parts than a 6 legged walker.

I'm in New Westminster, so if you want to get together and compare ideas sometime, e-mail me and let me know. I'd love to learn how to build air muscles.

Remove nospam_ to reply by email

Jeff H........

Lies, All lies. Don't believe a word Difool/sayNO says. He fears the truth!

build a solenoid from some nail and copper wire, or buy one. for me it's cheap dc motors and gears. i've banged this up, i usually don't draw them out, but this is the general idea. nylon nipples with teflon tape for tubing, a plastic plate to prevent disastrous leaks, a nylon washer and nut on the shaft, and some splines for bellcranks or gears. with higher pressures the plastic plate could be backed with metal and chromed brass nipples used.

I'm a big fan of the Mead Isonic valve for this kind of application. It's not quite computer-controllable, but 5 minutes with a ULN2803 and a printer port can resolve that.

Lee make some really nice small valves, as well - but they're much more pricey. And the pressure tolerances aren't as good.

But if you're needing to put air at 120psi in a muscle, you're doing something wrong :->

cheers, Rich.

Hi Jeff,

I'm trying to build a mostly to scale mobile full 8 legged spider, built the first draft for last Halloween.

Legs are aluminium, and not fully to scale as the torque worked out too high so I had to shorten the legs, and I'm redesigning the frame in plastic for weight. I'm still evaluating leg X/Y mechanics to balance out the speed/cost/power issues, which is why I like the air-muscles, although that brings in an issue with the air source, but I think I've got that covered via some simple chemistry...

I'm planning for an 8 legged gait as well as a 6 legged so I can use the front as "graspers", but will probably start with the 6 legged as that's easiest to program at first.

Thx, TJ

That sounds like an extremly difficult project to do with air power. I built a 6 legged air powered walker a few years back and found a lot of problems in doing it on a budget. One of the main issues was the air source. I had an array of small CO2 cylinders that I'd planned on using. They were about 3" dia X 12" long. The problem was that even with 6 of them I didn't have enough volume to work for long. I also considered a scuba tank to supply air, but those are heavy. I can't imagine how you're going to get usefull pressures from a chemical reaction. It's also hard to position air driven legs. Normally they move till they hit a mechanical stop.

What about using water in the muscles rather than air? It would add some weight, but allow you to stop them in mid stroke. (assuming you've got a small pump or valve for each joint)

To be honest, since you need a solinoid for each joint anyway, wouldn't it be easier to simply use small cheap motors to drive a threaded rod moving a nut. The nut then moves the leg through a linkage. If you like I can send you a couple of sketches on how that could work.

Are you planning on having a knee joint? IF so, it's going to be hard to make a hinge that's tight enough to wobble but loose enough to move with a reasonable ammount of power.

If you're going with plastic for the frame, make sure you're using lexan rather than acrylic. Lexan is soft and pliable. Acrylic is brittle and will crack easy. You can get lexan sheeting at Home Depot or most industrial plastic suppliers. Personally I'd use a pair of 1/8" aluminum sheets with the drive hardware mounted between them. (similar to the frames Lynxmotion uses) You can also do your own anodizing with some basic equipment.

You don't mention how large this thing is going to be, but I'd think you'd need a body at least a foot in diameter to hold the hardware if you went electric and more if you stick with air. (since you still need an electrical system to run everything) That means the legspan is going to be too big to walk around inside.

Spiders are one of nature's coolest designs, but they've got so many moving parts that it's hard to copy them on a budget. :(

Remove nospam_ to reply by email

Jeff H........

Lies, All lies. Don't believe a word Difool/sayNO says. He fears the truth!

Jeff Holinski writes: [snip]

Erm, stop putting air in? Works for us.

(Alright, you want a real solution: mount a Hall effect sensor and magnet across the joint, gives you analogue position output, use it to drive the valves.)

Removes (most of the) compliance of the muscles, which is one of their big advantages.

For 8 legs, you can link the muscles so several leg movements happen simultaneously, thus reducing the valve cost.

If you're really on a budget, then you can get "washing machine" solenoid valves designed for water very cheaply - they can be run from DC with a care (110V AC solenoid => 45-50V DC operation - but that's getting to dangerous voltages)

[snip]

cheers, Rich.

Hi Rich,

Thanks for the reference, I'll take a look at Mead and Lee's valves, hopefully I can find a Canuck dealer.

As for the 120psi, I'm just using that as a max for burst pressure... I know the muscles like much lower pressure.

Thanks, TJ

While that's true, air is so easily compressible that you get soft spongy positioning. I've never been able to get nice solid positioning till I got to the end of the stroke at full pressure.

I've always thought of hall effect and magnets of more of a digital solution. A potentiometer would be easy to do, but you get into the spongy postioning mentioned earlier.

That was the idea. Isn't compliance a disadvantage for something like this? Good for a suspension though.

But then you lose the ability to adapt to terrain. If you just wanted to move 8 legs in a set pattern you could use a pair of motors running camshafts. You then run cables from these cams to move the leg joints.

I've seen 12v air solinoids for $5 cdn before. I doubt you'd save much with low pressure washing machine solinoids.

Remove nospam_ to reply by email

Jeff H........

Lies, All lies. Don't believe a word Difool/sayNO says. He fears the truth!

Have you looked at using the 12v sprinkler control valves? Should be available at the home improvement centers.

Hey Jeff,

I'm not expecting this to be super-easy.

I'm actually looking at a combination of stepper motor and air muscle, basically using the air muscle to get a lot of the torque/lift, and the stepper motor for the fine positioning. Granted the plans aren't finalized at the moment.

I'll currently be using just stepper motors for X positioning, and the combo for Y positioning. That way I can use a *much* lighter stepper motor, which gives me position encoding, and a reduced power requirement and and thus battery weight... which lightens the whole bot down again etc. The challenge in any robot project is to think synergistically about the whole combination of factors so as to get stuff to work together. This is part of why I'm doing a fair amount of research into actual spider motion, gaits etc. as well as the mechanics, in order to see where it might be possible to transfer energy from one leg to another. After all the spider faced energy constraints as well during evolution.

Air supply is an issue, and I'll probably run tethered during testing, but want an independant bot... There are a large number of chemical reactions that can provide enough pressurized gas, the challenge is doing so in a controlled and repeatable manner. Vinegar and baking soda are one cheap and relatively example that I've worked out the gas yields at the required pressures.

Rich points out a lot of issues about using water instead of air, most of which I agree with, but the additional one is weight. I'd rather go with proporitional air valves than try to bring in the weight of the water. An additional point is that if an muscle bursts, all I loose is that muscle... if I used a water muscle, I'd likely end up shorting out half the bot from the spray..

The legs only have one pivot point, basically a modified caster, ie. with wheel removed, so that I get a lever point that is rotateable. They were cheap and easy from Home Depot... I'd like to have more working joints in the legs, but one is fine for now... As I've said, I had to take out a few leg segments in order to get the MA down to something reasonable, while still giving the basic look.

The body is about 2ft long, and about 1 wide, legs are about 3ft each, which is why I need the torque, and why the air muscles seem like the best match. The abdomen will hold two plastic popbottles as an air reservoir plus other bits.

TTUL, TJ

Opposing muscles can solve this, by stiffening the joint.

Depends on the Hall effect - if you use a *switch* it will be. Look for "ratiometric Hall sensor" in e.g. Allegro's web site.

Well, if you don't need compliance, then fine :->

Only if you removed compliance to begin with :->

That's the kind of price we were buying the washing machine ones from.

cheers, Rich.

I don't see the point to using chemicals. Not only do you now need to worry about an additional fuel supply, but you must worry about disposal of waste biproducts, as well as electrically controlling the mixing of the chemicals. It's unnecessary complication. Just buy a $20 electric tire inflator from Target and remove the mechanics for an inexpensive air compressor.

Chris

Even then it's overkill. I've done some experimentation with air muscles, and 30 to 40 psi is usually more than enough pressure. Once you get around 80 psi even the toughest materials start failing, not that home-made air muscles are known for their quality ;) Also, the more pressure you use, the more dangerous those failures will become for both you and the robot. Best to keep things no more pressurized than need be.

Chris