Simple position control made smarter?

I have a slide gate with analog position feedback (0-100% open) and an air cylinder to actuate the thing. I have two solenoid valves to operate the cylinder, one to open and one to close.

Currently for control, I turn the open solenoid on if the position is more than 2% under the target position and I turn the close solenoid on if the position is more than 2% over the target position. If I regulate my air supply and slow the thing way down (6 seconds full stroke) the gate will move to the target position and stop within the 4% window.

I’m not worried about making the move any faster, but the operators are unhappy with the fact that they cannot control the position to the exact percent, or if they want to bump the thing by a percent, they have to go down by four, then up by five. So my question is, what could I do to make these fine adjustments without worrying about the gate overshooting and bouncing back and forth?

I understand that there are other linear actuators that would be just the thing, but the challenge of this question is to get the job done with the existing air cylinder. I am all for keeping it simple, but I appreciate more the clever solution. Any thoughts?

Positioning with air is almost as hard as shooting pool with a rope. I think you're doing pretty good to get 4%. The sticktion of the cylinder is probably killing anything less than the 4%. By the time you overcome the static friction, the air pressure has built up so much you move that 4%.

Maybe a special low friction piston seal?

Other ideas:

• Electric actuator
• Proportional Hydraulics
• Short Electric actuator on the end of the cylinder. Bring the electric actuator to center while moving to position and use it to do the last percentage of the move. (sort of a Rube Goldberg here but might be cheaper than changing everything else out.

Peter should have some ideas on this but you could use a Prop air valve. That and a motion controller and you could do real good. If you can keep the air constant then you should be able to put in an offset to get close. If speed is not a problem then just slow it down with air flows.

Some years ago I used an SMC cylinder that had a built in encoder and a brake. I could stop it to a couple thou'.

We can do a very good job but...

CharlesM said:

Peter should have some ideas on this but you could use a Prop air valve. That and a motion controller and you could do real good. If you can keep the air constant then you should be able to put in an offset to get close. If speed is not a problem then just slow it down with air flows.

Click to expand...

It is easier to get a small linear actuator and motor. Just because we can do it doesn't mean it is a practical solution.

Here is an example of what we can do
ftp://ftp.deltacompsys.com/public/pneu/pneu5ext0.jpg
You can see we can zoom in on the position easily. This was a 2.5 inch cylinder pushing 120 lbs at about 10 inches per seocnd. If I stood on the 120 lbs and added my 280 to it the system would overshoot by about 0.050 out of about 10 inches even with the same tuining.

We test using pneumatics. If we can control pneumatic systems then hydraulic systems and most servo motor systems are easy.

Thanks for the thoughts guys.

I might have to look into the proportioning pneumatic valve. I thought about doing something similar with pulsing the solenoid valve, and varying the pulse width to get an analogish output from them. But I would probably wear the poor little things out in a hurry.

I have talked to these guys before but never used them. They may be able to help

http://www.enfieldtech.com/PS_systems_valves.htm

What's not mentioned here is the force & stroke required for the application. Apparently speed is not an issue due to the fact that you can tolerate a 6 second cycle time???

If that's the case,then it's my bet that you will spend way more $$ attempting this type of accuracy with pneumatics then you would be simply replacing the unit with a ball screw or some other purely mechanical actuator with an encoder feedback.

Simple. If your error is less than 2%, then actual = target and now your displaying bang on. Problem solved. It seems that all you want to do is keep them "happy" (quiet). You fall out of range or someone changes setpoint you display the actual value until it falls back into range.


Nick