Position control of pistons with simple solenoids

[viggy96]

I wanted to get more accurate position control of some pistons, and using a DART linear actuator was not viable, as it needs to provide more force than the DART can offer. I came across this paper (http://user.das.ufsc.br/~victor/Artigos%20ProjetoC/V%C3%A1lvulas/Accurate%20Position%20Control%20of%20a%20Pneumatic %20Actuator%20Using%20on-off%20solenoid%20valves.pdf) on the Internet. I wanted to know if anyone else could get a better idea of the diagram in the paper and see if something like that would actually be possible.

[MrForbes]

wow, I'd like to see a team implement that! It might be rather challenging, though. We like to use pneumatics on stuff that needs to be in either of two positions, not in between. We like to use motors for things that need to be put in different positions...although we'd rather make our machine so it only needs things to be in one position or the other, if possible.

My guess is that you probably won't get good results using the method in the paper, as it probably requires you to modulate the pneumatic solenoids in a way that is difficult to do with our FRC control system. But I could be wrong.

[Kevin Sevcik]

I think it the main hurdle in frc is the fact that they used a linear slide with a pneumatic brake. I'm not certain the brake would be a legal pneumatic actuator. The other difficulty is the PWM period of 16ms. You'd have to get sub millisecond update rates out of the pneumatic control module, and I don't think that's possible. After that, it's sourcing a fast enough solenoid valve, and everything else is frc legal. If the pneumatic brake was legal, then I think you could have done this on a cRIO controller.

[BoilerMentor]

In 2013 1747 attempted to build a pneumatic climber. We used a fairly complex, but available set of manifold style solenoids from SMC. The climber ultimately did not work due to significant mechanical issues that were encountered, but the pneumatic end of the climber was successful. The climber was ultimately ditched after a single regional event spent playing defense the whole time and never hanging for a Frisbee shooter and a simple low bar pneumatic hang.

IIRC there was a double acting solenoid and two single acting solenoids per cylinder. The double acting valve controlled direction and the single acting valves were basically an on/off for air supply to the cylinder. What this resulted in was the ability to select a direction (extend or retract) and then either allow flow in to and out of the cylinder or cut that flow off at both ports which maintained the position of the pneumatic as long as the load on the cylinder didn't change. This wasn't instrumented with something like a string pot or linear pot, so I don't know what accuracy was, but I suspect with appropriate flow control it could be accomplished.

In hind sight, I'm wondering if this is/was legal considering the single-solenoid-output-per-cylinder rule, but maybe since they were in series it wasn't an issue. I know we had tons of questions, as you'd expect with a 5 gallon aluminum air tank on the robot.

I wish I had a video to share, but that robot, pre-Frisbee shooter, is our robot-which-shall-not-be-named.

[pribusin]

There are 4-way 3-position solenoid valves available from SMC (and other valve suppliers) that should let you maintain any position of a pneumatic cylinder at other than end stops. Keep in mind that any load change on the cylinder rod will have a result on the cylinder rod position as this is not an active system but a passive one.

I say 'should maintain position' because we're about to test that with our first ever 3-position valves arriving today. The key is to get a 3-position valve with closed centers which effectively closes off the exhaust ports of both sides of the cylinder when the valve is in the OFF position. That way you would be working against 60psi of pressure on both sides giving you a significant amount of position hold force in both directions depending on your bore diameter.

We hope to assemble a test bed tonight to experiment with these valves and I'll report our findings.:)

[JCharlton]

A closed centre valve is a good solution to your problem, however beware R89 which requires a pressure vent plug to release all stored air. With a closed centre pressurised air would remain trapped in your cylinder.

In the past I think there have been rules saying something along the lines that if a mechanism could generate pressures higher than 60 psi (by an external force) you would have to have a way to relieve that pressure. I don't see that in the 2016 rules, but beware.

[viggy96]

There are 4-way 3-position solenoid valves available from SMC (and other valve suppliers) that should let you maintain any position of a pneumatic cylinder at other than end stops. Keep in mind that any load change on the cylinder rod will have a result on the cylinder rod position as this is not an active system but a passive one.

I say 'should maintain position' because we're about to test that with our first ever 3-position valves arriving today. The key is to get a 3-position valve with closed centers which effectively closes off the exhaust ports of both sides of the cylinder when the valve is in the OFF position. That way you would be working against 60psi of pressure on both sides giving you a significant amount of position hold force in both directions depending on your bore diameter.

We hope to assemble a test bed tonight to experiment with these valves and I'll report our findings.:)
That sounds perfect. Any idea if its any different to program one of those SMC valves versus normal solenoids? I really appreciate the information you've provided.

[pribusin]

The difference is only in our program logic. You have two solenoids that you need to activate alternately to make the piston move in either direction, just like with a normal 4-way double solenoid valve. But now you also have the option of stopping the piston mid-stroke by turning both solenoids off.

We didn't finish our valve test assembly last night and I have nothing to report today but tonight should give us results to report.

[FrankJ]

Interesting article. The IBM PC sort of dates it though. :] It would be a big consumer of air since the valves are constantly cycling. Pneumatic brakes are neither legal or illegal. You just need to use an legal pneumatic cylinder to actuate it. It can be an integral part of the brake as long as R77 J is met. The equivalent rule in past years was a lot tighter for passed for a pneumatic actuator.

A closed center valve works well for positioning assuming the dynamic forces aren't too high. The trouble is R89 requires the system to fully vent through one vent valve. R82 has an allowed working pressure of 60 PSI. A closed ended cylinder might exceed that. Since check valves are legal this year, you can comply with the rule by clever use of check valves to vent the closed ends of the cylinder back to the working pressure line. R90 will not be violated since you will be plumbing the outlet of the control valve to the inlet check valve.

[DonRotolo]

Just remember: Air is compressible. Put a load on your cylinder shaft and it will move.

In general, it is a best practice to use pneumatic cylinders for two positions only.

[pmangels17]

I don't know if you want to control the cylinder along a continuous length, or only have discrete points you want to reach. However, if you simply have more that two discrete points you want the cylinder to actuate to, there are a couple options. First, there are multi-position cylinders available, which may be a viable option. I have also seen teams successfully attach one cylinder to the end of another, such that two separate cylinders combine to be able to reach four different points. I would imagine that, given enough creativity, one could even stack two three-position cylinders together, or a few two-position cylinders, though this may require more custom parts, time, or weight. That being said, it is certainly a viable option if you don't need a continuously variable position.