Some movie monster robot , like this 1:10 Full function remote control monster Mechanical robot vehicles from movie “ taijqi “ , Two robot arm builds by computer precision cutting stainless steel and aluminum, not only reconstructs the actual mechanical function accordance with excavators in reality, the details also decorated to reproduce the Steampunk style in movie design. th-cam.com/video/-rgQFTTG3AM/w-d-xo.html
@@jamesbruton how it made here : th-cam.com/video/gLbkFyfXXl4/w-d-xo.html , I made a lot project for theme park too , thanks you for Inspired me a lot on your work .
Don't do it. With pneumatics it wouldn't work out well. With hydraulics it may work but it's probably to expensive and to complicated. For small and middle sized robots (up to human size) BLDC motors are the way to go. And there are lots of things possible in this area which may even turn out useful. Why not for example try to build a useful robotics plattform which can move around in- and outdoor and then attach various actuators to it to do stuff?
Tom Stanton isn’t that basically like your air powered car or engines but on a larger, more expensive scale or should I be looking out for amazing new things on your channel?
well, one thing you might have your hands on is a 5/3 valve which is made especifically for double-effect cylinders and the middle position keeps all the vias closed, so maybe this could be a better solution. other thing is that you sould put the regulator in 9:43 at the output of both tanks if you are going to use both and put a manual valve between both of them to balance them out, the regulator is made to lubricate and filter the air from the tanks, It keeps the cylinder from deteriorate because of the particles and also makes it run a bit smoother because of the oil. there are ones called "Quick Exhaust Valves" wich make the cilinder return faster if you need it, also they sometimes come with sound mufflers. I hope you found this useful, and I cant stress this enough, Put the regulator, It will keep the system nice and clean, that is a MUST
you're mistaken in implying its the volume of each side of the cylinder that determines the force it can output. It is the surface area being acted upon, so the plunger side with the rod on it has less face for the pressure to act upon.
Yea, and adding more pressure increases the force. I think the limiting factors are how much pressure the cylinder can hold with the shaft extended, and the compression load strength of the extending shaft.
He doesn't really have control over else than the amount of pressure in the tank in this test. Important to know if he codes a system to take the surface area into count at different positions though
@@jamesbruton I was thinking you'd need more "on" -time if it's carrying a load, but since you measure the position itself it kind of does that automatically.
Industrial systems that use big actuators (of many kinds, usually pneumatic or hydraulic, sometimes motor driven) often have String Encoders attached. Essentially a high accuracy encoder with a spring-loaded drum attached, and a string around the drum. As the string pulls out, it rotates the drum/encoder, and as the machine retracts, the spring maintains tension on the string and the encoder does its job in the other direction!
Hi james, you should really take a look at a 5/3 way valve, they also offer a locked position. Those that you have used are 3/2 way and are really only intended for use with single sided cylinders
People want so see james bruton building a lot of stuff for fun, nobody sees his capabilities and skill level to use it for real useful stuff. He should build devices and live enhancers for disabled people. This industry overcharge their patients anyway by far. A great way for james to do something helpful for the humams and make himself really really valuable to the world.
I bought an entire pallet of this stuff from a government auction. I still have no idea what to do with it all but I was told by the guy I outbid that I have EVERYTHING I need to make it all work except an air source.... I've got a nice big compressor in my shop though so problem solved. I just need to think of some epic project. I've got literally hundreds of different sized and types of cylinders.... Let the games begin. You have once again inspired me to build something crazy.
You could add a FCV (Flow Control valve like needle valve) combined with PCV (position control valve like 5/3 valve) for much better control I also recommend to use “automation studio” to simulate your system forehead There are also capacitive proximity sensors you could use for tracking position
Hi James, Inspirational man. As soon as I saw the speed of movement on OpenDog I though of pneumatics. My first experiments used 1 cylinder and 4 on/off solenoids rather than the Y solenoids (and a car tyre for air). All worked like your test except I could close all the solenoids so the cylinder didn't creep. I built a knee joint so just use a 10k pot in the angle for feedback. Then I discovered the Y solenoids, half the number of valves and drivers but picked up the mysterious creeping. Thanks for the solution. I now have one-ways on order plus more valves, cylinders and a compressor! Brilliant channel. Regards Steve
It would be interesting to use 3 state valves, it could allow a more precise control and lower the number of switching required. (Eg: lock one side while adjusting the other)
There is a nice MIT article on the use of pneumatic actuators. They used it in a legged robot but complemented it with an electrical motor in the hip. You can search the net for "Model-Based Trajectory Control of Robots with Pneumatic Actuator Dynamics" by Ryuma Niiyama
Nice project. Btw there are fully integrated high-flow cylinder positioning systems on the market that do exactly that and can be controlled by a PCL. These are usually way cheaper and much faster than servo actuators. These systems often include pressure sensors so you can even apply a set amount of force.
You could also control speed in code by creating a virtual input target that chases the actual input. Every loop of the code you would use some logic off the input pot to increment the position of the virtual input. Then use the virtual input as the target for the cylinder. The down side to this is that the cylinder would actually be pulsing to get to a position instead of going directly to it.
For proportional control, you could have a cam on a servo squeeze the tube. Sort of like a peristaltic pump, but it wouldn't revolve all the way around, just squeeze the tube in one spot. By rotating along an axis parallel to flow, you also wouldn't have to worry about the pressure trying to push the cam out of position.
Have you considered an air over hydraulic system for this? Fill the cylinders with oil and use air for the energy in the system. Also many cylinders have position indicator options you could hook into your Arduino. Keep making cool stuff!
What you need, is a 5/3 valve with two solenoids. With no solenoids powered, it will be closed, power one solenoid, and it will change over to allow air to flow into one port, and release air out of the other. That air you will pass through a bleed valve so you can control the speed of the cylinder. Always control the speed of a pneumatic cylinder by restricting the air flowing out of the system. If you denergize the solenoid again and powers the opposite, the cylinder will move in the opposite direction.
There are already electro-pneumatic valves for increasing and decreasing air flow for controlling the speed and curve. So no need to make them yourself, they are quite small and cheap
@@rock-afire-fan i dont know about those ones. I'm guesing you are talking abou industrial high pressure or high air flow control valves. Because I remeber using some small servo pneumatic control valves back in uni and they were relitively small and like 70$ for one. They could handle up to 9 or 10 bar I think. It's not a lot but for what hes got in the video, that would be plenty. I maybe wrong
I was actually thinking a ball valve attached to servo. Might not be good enough for a proper robot, but should be plenty good enough for testing purposes
Hi James. Most linear actuators (air) have a small magnet inside. There are small hall sensors that attach to the outer body on a circular part of the extruded body. Those sensor only give end to end indication. Thought I drop that info here for future reference. Keep up the good work! =)
If you wanted something crazier, you'd could look into electro-hydraulic actuators. Something like a tilt-block pump (variable axial piston) moving the fluid with the direction and rate it pumps controlled by a servo, and powered by a constant speed motor. Then that can be directly connected to the piston it actuates. Trades some efficiency for a lot of available force in a compact size. I think that's the kind of system being used by many fly-by-wire aircraft, and it can be made more redundant than the older style hydraulics with lines routed everywhere. Springiness or damping can be done with a bit of extra valves like bypass, and perhaps adding an accumulator tank to the system. Might be interesting for robot stuff.
That is a good start with a basically bang-bang system with no load and the friction is so high to dampen the whole unit. I wonder how it would behave if a heavy load was applied to the output. Presumably, it would oscillate, and some velocity feedback would be necessary. I think it would do without integral control. Congratulations. One would have appreciated velocity control, acceleration control, and integral control. in addition to just the positional error sensor.
Hydraulics don't compress much, are very controllable, and the force is through the roof. Kubota makes some very nice compact diesel engines that are made specifically to drive a hydraulic pump. Any robot that can pick up a car and squeeze it into a ball is a proper robot.
Lufferov hydraulics is used for higher duty applications. It’s stronger and can carry more weight, but it’s heavy and takes up tons of space. In addition, it can’t keep up with the precision pneumatics have at high speeds.
Instead of no-return valves, you could use 2 valves on each side: one to let air in, one to let air out. That way, you could control the pressure in both sides of the cylinder independently, and actively control the amount of compliance. It would also (theoretically) give you a way to do "regenerative braking", pushing air back into the reservoir when you backdrive a cylinder (e.g. landing after a big jump... ;) )
Take a look at push fit fittings instead of barbs and Jubilee clips. They are quick to assemble and you can get a wide variety of non-return valves, flow restrictors etc in push fit - pneumatic Lego. I use them a lot for various projects.
May I be pedantic for a minute? When you had the valve Teed together you have the same pressure in both ends of the cylinder but different areas which yield different forces. Force = Pressure * Area. So you are correct in the in balance of forces causing it to be difficult to hold a position but it's not due to a difference in pressure.
Before jumping deeper into project, I suggest you study What is available in the pneumatic components. There are valves that will operate both in and out stroke within one component (5/3 valve). Also there are other gadgets built into the cylinder to feedback the relative position of the action (pushing or pulling). The easiest problem of the noise as the air is pushed out can be achieved using silencer. Unless you want to use what you already have available and not invest in any more component, then the solution may be impractical like 2 air tanks. Because of the weight involved in compressor and /or engine to run the compressor, doing moveable project is rather less feasible than stationery project.
you should definitely do an aliens powerloader lol hacksmith is doing one so it would be cool to see both youtubers versions of it. If you do build one, I will absolutely be looking forward to watching!
V2P (voltage to pressure) proportional valves are available, that turn a 0-10v signal into a 0-10bar pressure output. But they are expensive. A sensible halfway house would be to just use 4 on/off valves , mounted with two at each end of the cyl, as close as possible to minimise the working volume, and use two of them as "fill" and two as "empty". This would allow the controller to independently modify the pressure in each end of the cylinder, and provide a much better fail safe situation (loss of power = all valves shut). A smarter control system, which now has 16 states (4 valves = 16 states) could easily use say 4 rules to determine the best operating order for the valves etc
Have you thought about trying to make an LVDT (linier variable differential transducer) for this setup? The basics are simple, three coils of wire in a half or full bridge setup. They're better for promotional control as there is no contact unlike pots, so they don't wear out.
Based on your comparison between pheumatics and electrical actuators, I think you should really take a look at hydraulics. You can get much better positional accuracy with hydraulics (compared to pheumatics), hydraulics don't compress like pneumatic systems. They also tend to be a bit safer than high pressure pneumatics as they don't fail so explosively.
@@jamesbruton Fair enough, I'd not looked at the costs involved. With respect to pheumatics, I'd suggest one thing to consider very carefully if strapping it to someone would be failure modes - what happens if the feed to one end of a cylinder fails and the cylinder is quickly thrown to one of it's extents.
You don't have to work with higher pressures just because it's a hydraulic system. There are low pressure hydrolic switches on the market that don't cost a lot. Only REAL downside is that leaks get pretty messy with oil spraying everywhere.
9:15 technically it's not because of higher capacity. It's because the piston face on the side without the ram has a larger effective area than the side with the ram as the ram takes up some of that area where it mounts to the piston face.
In an Robotics Book I have they use pneumatic actuators that run the cylinder against retractable pins in linear ways to position things. Though that is not proportional.
You don't need a motorized proportional valve, or even a non-return valve. If each of the solenoid valves you have opens and shuts quickly under PWM control, they can do the job themselves I think.
Thoughts: You could return to a single split source and place the regulator in front of the high side to balance directional pressures..your multiple air sources wont stay balanced, throwing any attempts to calibrate. Second - how much directional strength do you lose when this "braking" mechanism is active?
I was actually looking into something last year that might be useful to you for this project. Servo controlled ball valves, and they are relatively cheap to make you just need a servo that has the torque to turn the valve. There are a few designs up on Thingiverse that you could look at to see if it would suit your purposes for restricting outflow.
It would be possible to create a pressure cylinder that has a built in tank(s) in one or more segments of the cylindrical structure, and the 2 pressure valves could also built into the casted structure. This would allow for keeping things super light weight and also the placement of pressure sensors for more controlled accuracy within the walls of both cylinder/combination tank. Jim, it can be done!
I don't much like compressed air for power storage, but it seems okay-ish for transmitting power and driving a mechanism. Not great, though. I think hydraulics are a bit better for robotics, if only because liquid fluids are less compressible than gas. I don't know how the prices of pneumatic versus hydraulic versus electric components of the roughly the same utility in robotics, compare to eachother. I do wonder how much cheaper the pneumatic stuff is, such that it would tempt you to experiment with using it for a robot, though. I wouldn't think it very practical for something that needs to carry its own power source. There's energy density, and then there's energy efficiency. Density affecting how much energy you can carry around without refueling. And efficiency effectively affecting how quickly you use what energy you have. Carrying extra weight around usually affects efficiency. So if pneumatic systems, powered by a tank of compressed air, have poor energy density, and add a lot of extra weight to a robot, that means they'll suffer from a double whammy hit to how long they last between refueling. And yet… There are some situations were they could still be useful! Situations, specifically, where time between refueling is not a problem, like inside a workshop, or warehouse, that has air compression stations for the purpose of maintaining a robot. In which case, you can look for other advantages that pneumatic systems might have over electronic systems. Perhaps pneumatic systems wear out slower? I don't know.
There are pneumatic speed controllers that are very cheap and since they are controlled by a screw it should be relatively easy to interface. Also there are pneumatic mufflers that cut down on the sound.
Im not a mechanical engineer so this is probably a stupid question, but why wouldnt he connect the two i/o ports, on the pneumatic actuator, to each other via some small two-way air pump and an inlet valve, and then pressurize the system through that valve. In that way the system would not lose any pressure by constantly blowing air out, and the actuator could still be moved by pumping the air from one i/o port to the other.
It would work but the air pump would need to be just as big as the compressor to drive enough force, and I'd have to wait for it instead of compressing the air in advance.
@@jamesbrutonAhh i see, then yes it wouldnt be a practical solution at all. Im a computer guy so i was just thinking of the very small water pumps that are used for water-cooling systems, but i guess the major difference is that they dont need to drive the water at a very high pressure. Thanks for the clear answer :D
Connect both outlets and just use one valve there. There's just one open at any given time and should get less problems. You can even use the same statement/code for both directions. Less is more...
Hey James! Last year I was at the 2018 CSME International Congress (it's a Canadian engineering research conference) and actually attended a talk that addressed the problem you're trying to solve. The speaker proposed a couple modified algorithms that outperform current algorithms with low tracking error and much less switching that using PWM. Might be a good read or at least includes references that might help further this project. Here's a link to his paper: www.researchgate.net/publication/326305197_Direct_Switching_Position_Control_Algorithms_For_Pneumatic_Actuators_Using_OnOff_Solenoid_Valves
I wonder if it would be possible to put a return line into your holding tank so as the cylinder releases the air it goes back to the tank rather than wasted?
If i am correct, the pressure is the same on both sides of the cilinder. But it is the surface that is different on both sides. So the pressure is not different, what I heard in the video
I can't help but think of sim chairs when I see big actuators like that. How'd I'd love to have one that could do things like flip me upside down in a split second!
Could you incoporate some kind of use for the bleeded air to reduce energy usage somewhere else? Your batterias are quite bloated. They have been through some good use 😄
Could you use hydraulic fluid and a hydraulic pomp? Don’t need to store compressed air, the hole system stays pressurised so the rams stay in place, and isn’t compressible like air. Also easier to control the speed of a valve.
what about oil pneumatics? I'm sure there would be less play and you could use lower pressure also you can probably find or make a good pump to run off li-pos to power the pnuematics
HI. Is there a way to speed up a linear accuarator other than the speed you bought? It take 20 seconds for my weather maps to move behind a wall I've made? Thanks.
Try a 5/3 valve with closed mid position. And flow restrict valves (with a manual adjustment, a screw) on the ports of the cylinder. And air dempers on the exit porta on the 5/3. You don't use two different valves to conttrol one cylinder
I really want to use a proportional valve on the output - it'll be pretty hacky to start with. The on/off solenoid valves work ok though - I just need those check valves.
PWM is a method used sometimes in pneumatics but it isn't common in industrial applications as the high on/off rate leads to shortened valve life. PWM isn't great in most mechanical systems for the same reason
If you were to only use a pneumatic system to transmit power to mechanical components in your robot, you could conceivably create a closed-loop system for the gas you use. If you do that, you could have the luxury of choosing what gas to fill the system with. You could use a heavier gas, like argon, or a lighter gas, like helium. Are heavier gasses hard to compress? I feel like that might be the case, but I can't think of any specific evidence of it in my memory. (Might be a quirk of my neural pathways.) I have doubts about the usefulness of a closed loop pneumatic system, compared to a purely electric system. Something has to motivate those air pumps, after all, and that's likely going to be an electric motor, unless you install an internal combustion engine, just to run an air pump. (I wonder how the energy density of ethanol compares to a lithium ion battery, when the ethanol is running a smallish average internal combustion engine. I heard somewhere that alcohol burn hotter than gasoline, though, and some engines don't handle this well.)
I once tried to build an automatic door shutter with one of these. I didn’t use a flow control valve and when I tried it the first time… well… the door was shut. But in the end I decided it was too dangerous to use it. Don’t wanted to kill someone by accident with that door.
Hi James, great video! Are there any commercial systems/kits available for this sort of thing? And how would you measure the force output of the actuator?
When you push against the cylinder are you programming the Arduino to let out the air on the side of the force to increase the force against the hand motion?
Thanks for the great video, just what I was looking for. What's the advantage of pneumatic over electric ?? I am looking for a pneumatic based system that accurately positions a liner carriage at set positions, we now do this using stepper motors. Can the same result be achieved using pneumatic actuators or sliders ? What about the cost ? Will such a pneumatic system be cheaper than an stepper motor based system??
i kno1 this video is 2 years old but its the only relevant and in depth video out there. i work mainly with electric actuators but i am planning on doing a quite large animatronic project and i want to use hydraulics. does all of this work with hydraulics or is it just used for pneumatics?
Would these be good for an injection molding machine? I just bought a hydraulic cylinder but im thinking this may have been easier, and possinbly cheaper approach.
Hi James would you think it possible to control both a hydraulic slew drive and a hydraulic cylinder simultaneously with an Arduino? controlled by joystick rocker buttons.
What sort of (giant) robot could I make with this?
Some movie monster robot , like this 1:10 Full function remote control monster Mechanical robot vehicles from movie “ taijqi “ , Two robot arm builds by computer precision cutting stainless steel and aluminum, not only reconstructs the actual mechanical function accordance with excavators in reality, the details also decorated to reproduce the Steampunk style in movie design. th-cam.com/video/-rgQFTTG3AM/w-d-xo.html
I thought about a T-rex! maybe something that can walk...
@@jamesbruton how it made here : th-cam.com/video/gLbkFyfXXl4/w-d-xo.html , I made a lot project for theme park too , thanks you for Inspired me a lot on your work .
Don't do it.
With pneumatics it wouldn't work out well. With hydraulics it may work but it's probably to expensive and to complicated.
For small and middle sized robots (up to human size) BLDC motors are the way to go. And there are lots of things possible in this area which may even turn out useful. Why not for example try to build a useful robotics plattform which can move around in- and outdoor and then attach various actuators to it to do stuff?
Something for the porn industry?
James! This has given me an incentive to try some more pneumatic projects!! I might be bugging you for a parts list soon. Thanks
Hi Tom, sure let me know. Most of it was from eBay ;-)
Tom Stanton isn’t that basically like your air powered car or engines but on a larger, more expensive scale or should I be looking out for amazing new things on your channel?
Ah, eBay... the friend of many a hobbyist!
I have four bimbas for sale
well, one thing you might have your hands on is a 5/3 valve which is made especifically for double-effect cylinders and the middle position keeps all the vias closed, so maybe this could be a better solution.
other thing is that you sould put the regulator in 9:43 at the output of both tanks if you are going to use both and put a manual valve between both of them to balance them out, the regulator is made to lubricate and filter the air from the tanks, It keeps the cylinder from deteriorate because of the particles and also makes it run a bit smoother because of the oil.
there are ones called "Quick Exhaust Valves" wich make the cilinder return faster if you need it, also they sometimes come with sound mufflers.
I hope you found this useful, and I cant stress this enough, Put the regulator, It will keep the system nice and clean, that is a MUST
you're mistaken in implying its the volume of each side of the cylinder that determines the force it can output. It is the surface area being acted upon, so the plunger side with the rod on it has less face for the pressure to act upon.
yes this is true. Force = Pressure x Area. Same pressure on each side of Piston, but less area on underside of Piston due to Piston Rod
Yea, and adding more pressure increases the force. I think the limiting factors are how much pressure the cylinder can hold with the shaft extended, and the compression load strength of the extending shaft.
He doesn't really have control over else than the amount of pressure in the tank in this test. Important to know if he codes a system to take the surface area into count at different positions though
@@Asphyr The valves are only on or off, so no amount of coding will help me at this stage.
@@jamesbruton I was thinking you'd need more "on" -time if it's carrying a load, but since you measure the position itself it kind of does that automatically.
Boston Dynamics? Nahh... Bruton Dynamics.
Industrial systems that use big actuators (of many kinds, usually pneumatic or hydraulic, sometimes motor driven) often have String Encoders attached. Essentially a high accuracy encoder with a spring-loaded drum attached, and a string around the drum. As the string pulls out, it rotates the drum/encoder, and as the machine retracts, the spring maintains tension on the string and the encoder does its job in the other direction!
You should attach this to an a-frame ladder, put ratcheted wheels (that only spin one way) on the legs and watch it go.
Hi james, you should really take a look at a 5/3 way valve, they also offer a locked position. Those that you have used are 3/2 way and are really only intended for use with single sided cylinders
People want so see james bruton building a lot of stuff for fun, nobody sees his capabilities and skill level to use it for real useful stuff. He should build devices and live enhancers for disabled people. This industry overcharge their patients anyway by far. A great way for james to do something helpful for the humams and make himself really really valuable to the world.
I bought an entire pallet of this stuff from a government auction. I still have no idea what to do with it all but I was told by the guy I outbid that I have EVERYTHING I need to make it all work except an air source....
I've got a nice big compressor in my shop though so problem solved. I just need to think of some epic project.
I've got literally hundreds of different sized and types of cylinders....
Let the games begin. You have once again inspired me to build something crazy.
You could add a FCV (Flow Control valve like needle valve) combined with PCV (position control valve like 5/3 valve) for much better control
I also recommend to use “automation studio” to simulate your system forehead
There are also capacitive proximity sensors you could use for tracking position
Yes it's all a bit hacky at the moment
Hi James, Inspirational man. As soon as I saw the speed of movement on OpenDog I though of pneumatics. My first experiments used 1 cylinder and 4 on/off solenoids rather than the Y solenoids (and a car tyre for air). All worked like your test except I could close all the solenoids so the cylinder didn't creep. I built a knee joint so just use a 10k pot in the angle for feedback. Then I discovered the Y solenoids, half the number of valves and drivers but picked up the mysterious creeping. Thanks for the solution. I now have one-ways on order plus more valves, cylinders and a compressor! Brilliant channel. Regards Steve
thanks - I'll have another video coming up on this soon
Wow, I always had these questions and felt the need to test... this setup gave a great amount of insight into the working and pros/cons. Thankyou
It would be interesting to use 3 state valves, it could allow a more precise control and lower the number of switching required.
(Eg: lock one side while adjusting the other)
There is a nice MIT article on the use of pneumatic actuators. They used it in a legged robot but complemented it with an electrical motor in the hip. You can search the net for "Model-Based Trajectory Control of Robots with Pneumatic Actuator Dynamics" by Ryuma Niiyama
Nice project. Btw there are fully integrated high-flow cylinder positioning systems on the market that do exactly that and can be controlled by a PCL. These are usually way cheaper and much faster than servo actuators. These systems often include pressure sensors so you can even apply a set amount of force.
Uploaded just in time for my train to work!
You could also control speed in code by creating a virtual input target that chases the actual input. Every loop of the code you would use some logic off the input pot to increment the position of the virtual input. Then use the virtual input as the target for the cylinder. The down side to this is that the cylinder would actually be pulsing to get to a position instead of going directly to it.
This is one of my favourite videos of yours James. Really interesting, thanks for sharing.
For proportional control, you could have a cam on a servo squeeze the tube. Sort of like a peristaltic pump, but it wouldn't revolve all the way around, just squeeze the tube in one spot. By rotating along an axis parallel to flow, you also wouldn't have to worry about the pressure trying to push the cam out of position.
Have you considered an air over hydraulic system for this? Fill the cylinders with oil and use air for the energy in the system. Also many cylinders have position indicator options you could hook into your Arduino. Keep making cool stuff!
What you need, is a 5/3 valve with two solenoids. With no solenoids powered, it will be closed, power one solenoid, and it will change over to allow air to flow into one port, and release air out of the other. That air you will pass through a bleed valve so you can control the speed of the cylinder. Always control the speed of a pneumatic cylinder by restricting the air flowing out of the system. If you denergize the solenoid again and powers the opposite, the cylinder will move in the opposite direction.
There are already electro-pneumatic valves for increasing and decreasing air flow for controlling the speed and curve. So no need to make them yourself, they are quite small and cheap
actually the pneumatic servo valve that do exactly what you just said are quite large and expensive. A good servo valve costs around $1,000
@@rock-afire-fan i dont know about those ones. I'm guesing you are talking abou industrial high pressure or high air flow control valves. Because I remeber using some small servo pneumatic control valves back in uni and they were relitively small and like 70$ for one. They could handle up to 9 or 10 bar I think. It's not a lot but for what hes got in the video, that would be plenty. I maybe wrong
"Festo"
I was actually thinking a ball valve attached to servo. Might not be good enough for a proper robot, but should be plenty good enough for testing purposes
Hi James.
Most linear actuators (air) have a small magnet inside. There are small hall sensors that attach to the outer body on a circular part of the extruded body.
Those sensor only give end to end indication.
Thought I drop that info here for future reference.
Keep up the good work! =)
If you wanted something crazier, you'd could look into electro-hydraulic actuators. Something like a tilt-block pump (variable axial piston) moving the fluid with the direction and rate it pumps controlled by a servo, and powered by a constant speed motor. Then that can be directly connected to the piston it actuates. Trades some efficiency for a lot of available force in a compact size. I think that's the kind of system being used by many fly-by-wire aircraft, and it can be made more redundant than the older style hydraulics with lines routed everywhere. Springiness or damping can be done with a bit of extra valves like bypass, and perhaps adding an accumulator tank to the system. Might be interesting for robot stuff.
Such a good video, clear and precise. Help move my project design along significantly.
That is a good start with a basically bang-bang system with no load and the friction is so high to dampen the whole unit. I wonder how it would behave if a heavy load was applied to the output. Presumably, it would oscillate, and some velocity feedback would be necessary. I think it would do without integral control.
Congratulations. One would have appreciated velocity control, acceleration control, and integral control. in addition to just the positional error sensor.
This makes me appreciate those Festo demo animals even more.
Hydraulics don't compress much, are very controllable, and the force is through the roof. Kubota makes some very nice compact diesel engines that are made specifically to drive a hydraulic pump. Any robot that can pick up a car and squeeze it into a ball is a proper robot.
So what about hydraulics? Would be interesting to see a video discussing the pros and cons of hydraulics versus pneumatics.
Lufferov hydraulics is used for higher duty applications. It’s stronger and can carry more weight, but it’s heavy and takes up tons of space. In addition, it can’t keep up with the precision pneumatics have at high speeds.
@@scrmnthn5233 but you can't compress the hydraulic fluid like you can compress air, so that would solve some of the issues?
Instead of no-return valves, you could use 2 valves on each side: one to let air in, one to let air out. That way, you could control the pressure in both sides of the cylinder independently, and actively control the amount of compliance. It would also (theoretically) give you a way to do "regenerative braking", pushing air back into the reservoir when you backdrive a cylinder (e.g. landing after a big jump... ;) )
Take a look at push fit fittings instead of barbs and Jubilee clips. They are quick to assemble and you can get a wide variety of non-return valves, flow restrictors etc in push fit - pneumatic Lego. I use them a lot for various projects.
May I be pedantic for a minute?
When you had the valve Teed together you have the same pressure in both ends of the cylinder but different areas which yield different forces. Force = Pressure * Area. So you are correct in the in balance of forces causing it to be difficult to hold a position but it's not due to a difference in pressure.
Before jumping deeper into project, I suggest you study What is available in the pneumatic components. There are valves that will operate both in and out stroke within one component (5/3 valve). Also there are other gadgets built into the cylinder to feedback the relative position of the action (pushing or pulling). The easiest problem of the noise as the air is pushed out can be achieved using silencer.
Unless you want to use what you already have available and not invest in any more component, then the solution may be impractical like 2 air tanks.
Because of the weight involved in compressor and /or engine to run the compressor, doing moveable project is rather less feasible than stationery project.
Pretty sure the solution is check valves like I said...
you should definitely do an aliens powerloader lol hacksmith is doing one so it would be cool to see both youtubers versions of it. If you do build one, I will absolutely be looking forward to watching!
This is well worth all the Patreon money! This is fantastic
V2P (voltage to pressure) proportional valves are available, that turn a 0-10v signal into a 0-10bar pressure output. But they are expensive. A sensible halfway house would be to just use 4 on/off valves , mounted with two at each end of the cyl, as close as possible to minimise the working volume, and use two of them as "fill" and two as "empty". This would allow the controller to independently modify the pressure in each end of the cylinder, and provide a much better fail safe situation (loss of power = all valves shut). A smarter control system, which now has 16 states (4 valves = 16 states) could easily use say 4 rules to determine the best operating order for the valves etc
Have you thought about trying to make an LVDT (linier variable differential transducer) for this setup? The basics are simple, three coils of wire in a half or full bridge setup. They're better for promotional control as there is no contact unlike pots, so they don't wear out.
Based on your comparison between pheumatics and electrical actuators, I think you should really take a look at hydraulics. You can get much better positional accuracy with hydraulics (compared to pheumatics), hydraulics don't compress like pneumatic systems. They also tend to be a bit safer than high pressure pneumatics as they don't fail so explosively.
It will cost a lot more though due to the higher pressures involved. Proportional valves will be 1000+ each
@@jamesbruton Fair enough, I'd not looked at the costs involved.
With respect to pheumatics, I'd suggest one thing to consider very carefully if strapping it to someone would be failure modes - what happens if the feed to one end of a cylinder fails and the cylinder is quickly thrown to one of it's extents.
You don't have to work with higher pressures just because it's a hydraulic system. There are low pressure hydrolic switches on the market that don't cost a lot. Only REAL downside is that leaks get pretty messy with oil spraying everywhere.
Also: with hydraulics you only neet one valve. So the complexity of the system is halved.
what if using same equipment and configuration in the video but instead of air using water with moderate pressure
Alien's Power Loader! Yes! Play Rock-em Sock-em Robots with Bumble Bee.
9:15 technically it's not because of higher capacity. It's because the piston face on the side without the ram has a larger effective area than the side with the ram as the ram takes up some of that area where it mounts to the piston face.
In an Robotics Book I have they use pneumatic actuators that run the cylinder against retractable pins in linear ways to position things. Though that is not proportional.
You don't need a motorized proportional valve, or even a non-return valve. If each of the solenoid valves you have opens and shuts quickly under PWM control, they can do the job themselves I think.
Thoughts: You could return to a single split source and place the regulator in front of the high side to balance directional pressures..your multiple air sources wont stay balanced, throwing any attempts to calibrate. Second - how much directional strength do you lose when this "braking" mechanism is active?
I was actually looking into something last year that might be useful to you for this project. Servo controlled ball valves, and they are relatively cheap to make you just need a servo that has the torque to turn the valve. There are a few designs up on Thingiverse that you could look at to see if it would suit your purposes for restricting outflow.
You should build a BIG exosuit with them :)
yes revisit the exosit with what you've learned since you stopped the last one
Let him finish a project or 2 first ;)
@@lukedavid8099 finish a project? what does that even mean?
@@bruceneely4859 hah lol
I love it when the undergrad assumption of non compressible cannot be ignored.
It would be possible to create a pressure cylinder that has a built in tank(s) in one or more segments of the cylindrical structure, and the 2 pressure valves could also built into the casted structure. This would allow for keeping things super light weight and also the placement of pressure sensors for more controlled accuracy within the walls of both cylinder/combination tank. Jim, it can be done!
maybe, but you won't store that much air in there, ideally I need a much faster compressor to keep up with the bore size of the cylinders
9:15 it is same pressure, but rod takes some of area. So efective area of rod side is smaller, force is smaller, and it goes to one side.
I don't much like compressed air for power storage, but it seems okay-ish for transmitting power and driving a mechanism. Not great, though. I think hydraulics are a bit better for robotics, if only because liquid fluids are less compressible than gas. I don't know how the prices of pneumatic versus hydraulic versus electric components of the roughly the same utility in robotics, compare to eachother. I do wonder how much cheaper the pneumatic stuff is, such that it would tempt you to experiment with using it for a robot, though. I wouldn't think it very practical for something that needs to carry its own power source.
There's energy density, and then there's energy efficiency. Density affecting how much energy you can carry around without refueling. And efficiency effectively affecting how quickly you use what energy you have. Carrying extra weight around usually affects efficiency. So if pneumatic systems, powered by a tank of compressed air, have poor energy density, and add a lot of extra weight to a robot, that means they'll suffer from a double whammy hit to how long they last between refueling.
And yet… There are some situations were they could still be useful! Situations, specifically, where time between refueling is not a problem, like inside a workshop, or warehouse, that has air compression stations for the purpose of maintaining a robot. In which case, you can look for other advantages that pneumatic systems might have over electronic systems. Perhaps pneumatic systems wear out slower? I don't know.
Finally I'm waiting for this kind of project,if you could make similar like this but smaller e.g. 1/3 of this cylinder size ok
You should use a servo valve. Also, for control loop stability remember the cylinder is an integrator that determines the dominant pole.
yep that was the plan...
If you jam some cloth or cotton wad in there before restricting the flow you’ll make it a lot quieter
Have you ever heard of the "air over oil" concept. Just another thing that might be worth looking into
This project should be sponsored by smc pneumatics
Boston dynamics uses custom made hydraulic actuators. That should be lesson to all of us as to what is the optimum drive mechanism for robotics.
There are pneumatic speed controllers that are very cheap and since they are controlled by a screw it should be relatively easy to interface. Also there are pneumatic mufflers that cut down on the sound.
By a reed rod, widely used in industry for hydraulic cylinders or a rotary encoder
Im not a mechanical engineer so this is probably a stupid question, but why wouldnt he connect the two i/o ports, on the pneumatic actuator, to each other via some small two-way air pump and an inlet valve, and then pressurize the system through that valve. In that way the system would not lose any pressure by constantly blowing air out, and the actuator could still be moved by pumping the air from one i/o port to the other.
It would work but the air pump would need to be just as big as the compressor to drive enough force, and I'd have to wait for it instead of compressing the air in advance.
@@jamesbrutonAhh i see, then yes it wouldnt be a practical solution at all. Im a computer guy so i was just thinking of the very small water pumps that are used for water-cooling systems, but i guess the major difference is that they dont need to drive the water at a very high pressure. Thanks for the clear answer :D
I'm wondering if you can't recycle the compressed air somehow? Into a rubber bladder? Less load on the compressor and you may only need a single tank.
Connect both outlets and just use one valve there. There's just one open at any given time and should get less problems. You can even use the same statement/code for both directions. Less is more...
This is pretty cool! You should look at the SOMA project by TU Berlin, they got some neat pneumatic robot hands that they're working on.
Hey James! Last year I was at the 2018 CSME International Congress (it's a Canadian engineering research conference) and actually attended a talk that addressed the problem you're trying to solve. The speaker proposed a couple modified algorithms that outperform current algorithms with low tracking error and much less switching that using PWM. Might be a good read or at least includes references that might help further this project. Here's a link to his paper:
www.researchgate.net/publication/326305197_Direct_Switching_Position_Control_Algorithms_For_Pneumatic_Actuators_Using_OnOff_Solenoid_Valves
Why not put a pressure regulator on the higher power side + drop it down so it doesn’t move either direction?
I wonder if it would be possible to put a return line into your holding tank so as the cylinder releases the air it goes back to the tank rather than wasted?
Very helpful, thank you.
If i am correct, the pressure is the same on both sides of the cilinder. But it is the surface that is different on both sides. So the pressure is not different, what I heard in the video
I can't help but think of sim chairs when I see big actuators like that. How'd I'd love to have one that could do things like flip me upside down in a split second!
This is so cool!!!
Could you set up two of these systems with the ends facing each other ( --->
Could you incoporate some kind of use for the bleeded air to reduce energy usage somewhere else? Your batterias are quite bloated. They have been through some good use 😄
What about using the solenoid valves in reverse on the out flow to restrict the out going pressure...
I did that on the Transformers build to stop it crashing down of power failed. There are also multi-position valves that can do it.
Give Metalwork pneumatics in Milton Keynes a call, they have everything you would ever need.
Could you use hydraulic fluid and a hydraulic pomp? Don’t need to store compressed air, the hole system stays pressurised so the rams stay in place, and isn’t compressible like air. Also easier to control the speed of a valve.
Yes but that would be a lot more expensive, the proportional valves are 1000+ at least due to the higher pressures involved.
Mr. B - you grock! Thanks for the vid and info.
what about oil pneumatics? I'm sure there would be less play and you could use lower pressure
also you can probably find or make a good pump to run off li-pos to power the pnuematics
You could use double coil closed center1/2 5-3 directional control valve
you should combine forces with the Hacksmith. They are working on something like that but im sure you know that
HI. Is there a way to speed up a linear accuarator other than the speed you bought? It take 20 seconds for my weather maps to move behind a wall I've made? Thanks.
Try dry ice or LN2 with a warmer tank to convert/ gassify... less space, higher energy density?
Try a 5/3 valve with closed mid position. And flow restrict valves (with a manual adjustment, a screw) on the ports of the cylinder. And air dempers on the exit porta on the 5/3. You don't use two different valves to conttrol one cylinder
I really want to use a proportional valve on the output - it'll be pretty hacky to start with. The on/off solenoid valves work ok though - I just need those check valves.
Can you do PWM valves on the outlets?
PWM is a method used sometimes in pneumatics but it isn't common in industrial applications as the high on/off rate leads to shortened valve life. PWM isn't great in most mechanical systems for the same reason
If you were to only use a pneumatic system to transmit power to mechanical components in your robot, you could conceivably create a closed-loop system for the gas you use. If you do that, you could have the luxury of choosing what gas to fill the system with. You could use a heavier gas, like argon, or a lighter gas, like helium. Are heavier gasses hard to compress? I feel like that might be the case, but I can't think of any specific evidence of it in my memory. (Might be a quirk of my neural pathways.)
I have doubts about the usefulness of a closed loop pneumatic system, compared to a purely electric system. Something has to motivate those air pumps, after all, and that's likely going to be an electric motor, unless you install an internal combustion engine, just to run an air pump. (I wonder how the energy density of ethanol compares to a lithium ion battery, when the ethanol is running a smallish average internal combustion engine. I heard somewhere that alcohol burn hotter than gasoline, though, and some engines don't handle this well.)
I once tried to build an automatic door shutter with one of these. I didn’t use a flow control valve and when I tried it the first time… well… the door was shut.
But in the end I decided it was too dangerous to use it. Don’t wanted to kill someone by accident with that door.
Hi James, great video!
Are there any commercial systems/kits available for this sort of thing? And how would you measure the force output of the actuator?
When you push against the cylinder are you programming the Arduino to let out the air on the side of the force to increase the force against the hand motion?
Have you ever watched AvE? He recently did a whole little mini series about pneumatic systems and playing around with them.
once more, this is very very interesting !
I would love to see a build with air muscles! Perhaps Festo would even sponsor such a project?
I dont think they do... At least in México we dont do that but we have very nice techology to play
@@fernandocastro2075 do you work for Festo in Mexico? :o
Is it possible to hold in a position of cylinder by using 5/3 Direction control valve
lets say you wanted the cylinder piston to go back and forth by itself
without you using the hand switch , how would you do that?
Thanks for the great video, just what I was looking for.
What's the advantage of pneumatic over electric ?? I am looking for a pneumatic based system that accurately positions a liner carriage at set positions, we now do this using stepper motors.
Can the same result be achieved using pneumatic actuators or sliders ? What about the cost ? Will such a pneumatic system be cheaper than an stepper motor based system??
Try using 3 position, 5 way valve. It should hold the pressure quite well.
That's probably what I need in reality
very cool! gained one more subscriber
i kno1 this video is 2 years old but its the only relevant and in depth video out there. i work mainly with electric actuators but i am planning on doing a quite large animatronic project and i want to use hydraulics. does all of this work with hydraulics or is it just used for pneumatics?
Can the air cylinder provide constant force?
As is, it has a nice air suspension / shock absorption feature.
Would these be good for an injection molding machine? I just bought a hydraulic cylinder but im thinking this may have been easier, and possinbly cheaper approach.
Do air over electric... small electric actuator to just do the final adjustment. Let air do the heavy lifting and electric do the fine tune?
Holy crap, I've spent years reading your name as Burton... It wasn't until you said your name that I realized I was wrong.
Hi James would you think it possible to control both a hydraulic slew drive and a hydraulic cylinder simultaneously with an Arduino? controlled by joystick rocker buttons.