Oscillating a pnuematics valve?

[Andrew Blair]

I was wondering if quickly oscillating a double solenoid valve would damage it after repeated use. The reason would be to create a multi-position pnuematics system using only 1 valve switching back and forth at the correct time, instead of controlling the exaust. Are the double's even fast enough to do this?So the question is has anyone done this, or is it unfeasible? :confused:

[Rickertsen2]

I have seen applications in the textile industry where pneumatic valves are actuated at speeds of over a hundred times a second all day long. (I was amazed to learn that pretty ordinary valves are capable of this). As long as you use a pretty solid valve, you should be ok. I don't know how fast any of the valves in the kit are but you don't need them to be all that fast. Keep in mind that that your setup will almost instantly drain your air supply if you are describing what i think you are.

[Andrew Blair]

I have seen applications in the textile industry where pneumatic valves are actuated at speeds of over a hundred times a second all day long. (I was amazed to learn that pretty ordinary valves are capable of this). As long as you use a pretty solid valve, you should be ok. I don't know how fast any of the valves in the kit are but you don't need them to be all that fast. Keep in mind that that your setup will almost instantly drain your air supply if you are describing what i think you are.

Darn, you're right. I didn't consider this. Perhaps if a flow control was used you could slow the air loss on the open side long enough to correct, then close it again. I'm considering an arm load where most of the time the load force is enough to allow control by opening only the lifting side for a moment. However, the arm itself wouldn't be heavy enough to lower the arm if it's not carrying something, so you would have to force it down using the other isde of the cylinder. Any ideas?

[gburlison]

Darn, you're right. I didn't consider this. Perhaps if a flow control was used you could slow the air loss on the open side long enough to correct, then close it again. I'm considering an arm load where most of the time the load force is enough to allow control by opening only the lifting side for a moment. However, the arm itself wouldn't be heavy enough to lower the arm if it's not carrying something, so you would have to force it down using the other isde of the cylinder. Any ideas?
In the past we have created multi-position pneumatics by connecting a cylinder to a double solenoid valve and then plumbing the exhaust port of the valve to a single solenoid valve. The double solenoid valve is used to control the direction that you want the system to move and the single solenoid valve is then opened to allow movement, and closed to stop movement. It was totally manual and somewhat imprecise. With some feedback (switches or potentiometer) it should be better.

[Rickertsen2]

Darn, you're right. I didn't consider this. Perhaps if a flow control was used you could slow the air loss on the open side long enough to correct, then close it again. I'm considering an arm load where most of the time the load force is enough to allow control by opening only the lifting side for a moment. However, the arm itself wouldn't be heavy enough to lower the arm if it's not carrying something, so you would have to force it down using the other isde of the cylinder. Any ideas?

The best thing to do would be to use something like a gas spring to pull the thing down and use a special type of 3-way valve called a 3 position closed center double soleniod valve.

this valve would allow you to alternate one side of the cylinder between pressurized, sealed off and vented while only using a single valve. When neither soleniod is energized, the output is closed off. When one of the coils in energized, the output gets pressure. When the other is pressurized, the cylinder is vented. I did some tests with this sort of setup and it worked pretty well. The biasing spring also leads to increased stability when lifting light loads. You could also use a 4-way version of one of these if you want to control both sides of the cylinder.

[Andrew Blair]

I'm not real versed in pnuematic's rules, and I'm considering this for a FIRST robot. What are the rules concerning the purchase of solenoid valves and which ones you may use?

[Tristan Lall]

I'm not real versed in pnuematic's rules, and I'm considering this for a FIRST robot. What are the rules concerning the purchase of solenoid valves and which ones you may use?The last revision of the rules allowed any unmodified off-the-shelf solenoid valve rated for 125 psi and consistent with the other generic parts restrictions (e.g. cost, safety, etc.). Rules change every year; verify that this is consistent with the 2006 rules when they appear.

[Mr.G]

The best way to do this is to use hydraulics, you can stop the cylinder at any place and it will be rock solid. If you want to do this for first, you need to get high presure on both sides of the piston or your set up will be spongy and not locate/hold location very good. You can make your cylinder way over sized to help make it stiffer. First's rules on pneumatics are very strick and really don't allow you to do much.

[Rickertsen2]

The best way to do this is to use hydraulics, you can stop the cylinder at any place and it will be rock solid. If you want to do this for first, you need to get high presure on both sides of the piston or your set up will be spongy and not locate/hold location very good. You can make your cylinder way over sized to help make it stiffer. First's rules on pneumatics are very strick and really don't allow you to do much.
Haha true. Unfortunately we are not allowed to use hydraulics :(

[Kevin Sevcik]

Several comments here. First, the double solenoid valve spec sheet says the maximum operating frequency is 10 Hz on the solenoid valve. So take that into account in your multi-positioning thing.

Second, I just had a brain wave. From Eagle Master's post up there. It's hard to get high pressures on both sides of a FIRST piston and still move it and such. But you could use a piston to hold something in place if it was otherwise actuated. You connect the extend and retract ports of the piston together, except you put a solenoid valve between them. You close the valve when you want the piston to maintain position and the piston acts as an air spring. You open the valve to allow the piston to move freely. This has some interesting characteristics, however. The more you pre-pressurize this closed system, the stiffer the spring. The closer to the end of the stroke, the stiffer the spring becomes. At the middle of the stroke, the spring is reasonably linear over short displacements, but becomes less linear towards the ends of the stroke. The only real problem, I think, is that there would be a net extension force on the piston due to the shaft of the piston.

Similarly, you could make a shock absorber by putting a needle valve between the two ports.

Note, I'm doubting the FIRST legality of this. The pressures could get dangerously high. Especially if the system started at 60 psi.

[Rickertsen2]

Several comments here. First, the double solenoid valve spec sheet says the maximum operating frequency is 10 Hz on the

Figures. They are air piloted and very low wattage. You can faster acting ones that are FIRST legal.

Second, I just had a brain wave. From Eagle Master's post up there. It's hard to get high pressures on both sides of a FIRST piston and still move it and such. But you could use a piston to hold something in place if it was otherwise actuated.
This brings to mind the appropriate off the shelf part, which of couse the overly-prohibitive FIRST pneumatics rules prevent us from using.
http://www.bimba.com/break/news54.htm

Basically, it is an ordinary cylinder with two enhancements.
*A position feedback resistor
*a second single-acting cylinder which actuates an integrated device which locks the cylinder rod in place, preventing it from moving.

[Mr.G]

Or you could use 2 cylinders that are inline. Extending the first one gets you x distance and extending the 2nd gets you y distance. So actually if the two cylinders were different stroke you could get 4 positions.

Distance 1 = x and y retracted
Distance 2 = x and y extended
Distance 3 = x extended and y retracted
Distance 4 = x retracted and y extended

[Rickertsen2]

Or you could use 2 cylinders that are inline. Extending the first one gets you x distance and extending the 2nd gets you y distance. So actually if the two cylinders were different stroke you could get 4 positions.

Distance 1 = x and y retracted
Distance 2 = x and y extended
Distance 3 = x extended and y retracted
Distance 4 = x retracted and y extended

Or, If the pneumatics rules weren't soo lame, you could buy off the shelf multiple position cylinders. These are more or less what you describe only integrated into a compact off the shelf part. The custom cylinder order form is the only first rule i disagree with. It has no rationale.

[John Gutmann]

Well what would you be using a multi position cylinder for, because most likely you can use something else to do it that may be better or save weight so you dont have to have a pump and/or tanks and everything else.

[KenWittlief]

The idea of having an arm that is moved by pneumatics, and locked in place anywhere within its range of motion - thats very clever! I dont think I have ever seen it done on a FIRST robot

you dont need to get fancy with the valves, simply use the flow restrictors to limit the speed at which the arm will move

and to get precise control, use feedback. Instead of having the operator release the clutch and move the arm up or down, and then engage the clutch at the right position, have something like a potentiometer on the arm reporting position. the operator would move the joystick up or down, and the system will sense the arm is not where the joystick says it should be, unlocks the clutch, pressurizes the cylinder in the right direction, and engages the clutch when the feedback sensor = the joystick setting.

I really like this idea. You could have a very light, very simple, multiple axis or multiple elbow articulated arm, with very precise control of its motion - using nothing but pnuematics!

[Squirrelrock]

I like this - I have attempted to get a cylinder to run from two single solinoid valves as a multi-positioner, and it is hard to do, until you figure out how to do it. I no longer have the setup, though, but I believe that the exhaust of the first valve had to be routed through the input of the second, and the second had to have one output blocked and be set to it as default. Then, the system backpressures itself (probably above 60 psi, so this probably is not competition legal) until the arm stops whereever it is. The way you described it would seem to be a much easier (wiring and tubing wise) way to get this done, and it has a much higher likelihood of being competition legal.

The idea of having an arm that is moved by pneumatics, and locked in place anywhere within its range of motion - thats very clever! I dont think I have ever seen it done on a FIRST robot

you dont need to get fancy with the valves, simply use the flow restrictors to limit the speed at which the arm will move

and to get precise control, use feedback. Instead of having the operator release the clutch and move the arm up or down, and then engage the clutch at the right position, have something like a potentiometer on the arm reporting position. the operator would move the joystick up or down, and the system will sense the arm is not where the joystick says it should be, unlocks the clutch, pressurizes the cylinder in the right direction, and engages the clutch when the feedback sensor = the joystick setting.

I really like this idea. You could have a very light, very simple, multiple axis or multiple elbow articulated arm, with very precise control of its motion - using nothing but pnuematics!

[Nitroxextreme]

last year we used multi positioning on 2 pneumatic cylinders and needed a total of 2 double solenoids and 2 single solenoids

http://www.chiefdelphi.com/forums/showthread.php?t=33258
has a good description as to what needs to be done for it to work

[KenWittlief]

the problem with using two valves to create a multiposition system is that the arm is balanced by the pressure in the cylinders. If you pulled or pushed on the arm at any given position, the arm would move (for example, if you dropped a tetra, the arm would rise up on its own).

using a locking mechanism, or clutch, would make the arm stiff in all positions.

[Cuog]

yeah thats what the potentiometer is for you make pregramming increase or decrease the pressure until the arm is within like 1 or 2 degrees of where the joystick is its very logical and a really great idea

What would happen is the arm would move due to the pressure change of shifting wieght but would automatically readjust itself to wherever it is supposed to be.

[DonRotolo]

Team 1676 tried that last year, but we had mixed results. The feedback loop (including the valves, movement of the air, etc.) wasn't fast enough to hold a position pneumatically when faced with variable loading. The problem is that air compresses, and so changes in load made the arm move: We hung a tetra on the arm and it drooped down. Also, the operator cannot expect instant response - we had something like a 1/2 second delay, and that was hard on the operator. We used a double-solenoid valve (3-way), no fooling with modulating exhausts and whatnot.

In short: Pneumatics are excellent for endpoint positioning, but awful for precision or intermediate positioning. Not that it can't be done - we did it - just that motors are a FAR better choice.

My advice would be to abandon ideas of variable mid-point positioning using pneumatics.

On the other hand, pneumatics were excellent for our gripping claw - open/closed.

Don



The idea of having an arm that is moved by pneumatics, and locked in place anywhere within its range of motion - thats very clever! I dont think I have ever seen it done on a FIRST robot

you dont need to get fancy with the valves, simply use the flow restrictors to limit the speed at which the arm will move

and to get precise control, use feedback. Instead of having the operator release the clutch and move the arm up or down, and then engage the clutch at the right position, have something like a potentiometer on the arm reporting position. the operator would move the joystick up or down, and the system will sense the arm is not where the joystick says it should be, unlocks the clutch, pressurizes the cylinder in the right direction, and engages the clutch when the feedback sensor = the joystick setting.

I really like this idea. You could have a very light, very simple, multiple axis or multiple elbow articulated arm, with very precise control of its motion - using nothing but pnuematics!

[KenWittlief]

Team 1676 tried that last year, but we had mixed results. The feedback loop (including the valves, movement of the air, etc.) wasn't fast enough to hold a position pneumatically when faced with variable loading. The problem is that air compresses, and so changes in load made the arm move: We hung a tetra on the arm and it drooped down.

I think you missed part of the description.

you include a clutch or locking mechanism on the joint, so once the arm is in the proper location, the clutch locks, holding the arm rigid. Unless you add enough weight to the end of the arm to overpower the clutch, its not going to move

and with this approach you dont try to control the pressure in the cylinder, or to apply pressure to both ends - you use the cylinder in the normal configuration, with the little flow restrictors to slow down their movement.

The clutch is in the feedback loop, not the valves. Once the arm has moved to the right position (up or down) the clutch locks and holds it against the pnuematic pressure, and against any added weight to the end of the arm.

[Rickertsen2]

I think you missed part of the description.

you include a clutch or locking mechanism on the joint, so once the arm is in the proper location, the clutch locks, holding the arm rigid. Unless you add enough weight to the end of the arm to overpower the clutch, its not going to move

and with this approach you dont try to control the pressure in the cylinder, or to apply pressure to both ends - you use the cylinder in the normal configuration, with the little flow restrictors to slow down their movement.

The clutch is in the feedback loop, not the valves. Once the arm has moved to the right position (up or down) the clutch locks and holds it against the pnuematic pressure, and against any added weight to the end of the arm.

I have tried this and it works! There are nifty off the shelf cylinders with pneumatic cluthes built in for exactly this application but they are of course not FIRST legal.
http://www.bimba.com/images/pfcrodlock.jpg
http://www.nimmofluidpower.com/img_fluid_power_product/bimba_line_rod_2.jpg

It might still be nifty to use a closed center valve in addition to a locking mechanism. By pulsing it on and off at different duty cycles, you could move at different speeds. Another strategy is to use more than one valve to control a cylinder. Each valve has it's flow restricted to a different extent.