So, here's the official thread to discuss PID. From what I've read, tell me if my layperson's interpretation is correct.

Proportional error is like saying, "hey, you're 1 inch behind where you're supposed to be, Mr. Cylinder. I'm going to move you ahead an inch to catch up."

Integral error is like saying, "hey, you've been 1 inch behind the last 10 times I've watched you, Mr. Cylinder. I'm going to move you ahead 2 inches to catch up."

Differential error is like saying, "You're supposed to be moving at 5 inches per second, Mr. Cylinder, but you're only moving at 4 inches per second. I'm going to speed you up."

How correct am I?

edit: changed second "proportional" to "differential."

This message has been edited. Last edited by: Josh Cosford,

quote:

Originally posted by Josh Cosford:
How correct am I?

Pretty good for starters.

You have brought up a couple of points. For linear actuators the error is almost always a position error. The error is determined by the difference between the target and actual position

Error=TargetPosition-ActualPosition;

The proportional term is something like this

ControlOutput = Kp*Error

Kp is the proportional gain and it has units of volts/InchOfError). Therefore a gain of 10 volts/InchOfError means an error of 1 inch would cause the controller to output 10 volts which is full signal to the valve.

Your second example is also a position example for how the integrator works. The integrator sums the error as a function of time. Every millisecond the integrator error is updated by doing this

IntegratorError=IntegratorError+(0.001 seconds)*Error

The integrator error has units of error*seconds
The integrator gain, Ki, has units of volts/(error*seconds). An integrator gain of 1 volts/(error*seconds) would just need the integrator error to to be 10 error*seconds to provide 10 volts. Normally the error are small and the integrator gain, Ki, is large relative to Kp. Remember we want the errors small.

A PI controller uses the the two terms you have mentioned. The total equation similar to this

Error=TargetPosition-ActualPosition
IntegratorError=IntegratorError+(0.001 seconds)*Error
ControlOuput=Ki*IntegratorError+Kp*error;

This equation is execute every millisecond or 0.001 seconds. Each millisecond the error should get smaller and smaller so the IntegratorError build up ( we use the term winds up ) more slowly as the error decreases.
The actual position comes from the feed back device which is usually a Temposonic or Balluff rod.

The third example is more accurate in that is talks about controlling speed. The control output calculated above goes to the valve and this controls how fast the actuator moves. As the error decreases the control output will decrease and the velocity will decrease.

If the control signal to the valve is +5 volts the actuator will move ahead at 1/2 of full extend speed. If the control output is -10 volts the actuator moves at full retract speed. The full extend speed and full retract speeds are usually different because the actuator has a rod and there are different surface areas on the piston. What makes a hydraulic motion controller different, from a normal motor controller, is that it must use different gains while extending than when it is retracting UNLESS the cylinder is double rodded or we are controlling a hydraulic motor.

This should be good for now.
What we haven't talked about is the D part of the PID. The D or derivative is very important in hydraulic systems.

Do you know where the actual position comes from?
Do you know where the target position comes from?

We haven't talked reducing error to as small as possible. See the links to the graphs I posted. The difference between the target and actual positions is often just a few thousands of an inch.

OK, so what are Temposonic and Balluff rods?

josh
trt a google search to find the companies that th names represent and you eill find about all you need to know about these feedback devices.

Here is an excellent hit:
http://www.acshydraulics.com/temposonic.html

Here is a more detailed explanation
http://www.mtssensors.com/fileadmin/media/pdfs/551019.pdf

To see how they are used look at the diagram at the bottom of this.

http://www.deltamotion.com/pdf/fluid%20power%20basics.pdf

Notice that the MDT rod ( magnetostrictive displacement transducer ) is inserted into the cap end of the cylinder and that the rod is is bored out to make room for the MDT rod. This keeps the MDT rod safe in nasty industrial environments.

The simple diagram shows pressure transducers too. These are not required if all you want to do is position control. The pressure transducer are required for those applications that need to control pressure or force.

What is important to know is that these feedback devices can update very quickly. As fast as every 250 micro seconds and the resolution can be as fine as 2 to 5 microns. With this can of resolution you can do many machine control applications that required precision grind, stamping, pressing, cutting etc.

So, putting it in layman's terms again; A magnetic field on the piston generates a current on the rod. The time it takes for the current to travel from the rod to the sensor, tells the computer how far along the rod the magnet is, therefore telling us the exact stroke.

quote:

Originally posted by Josh Cosford:
So, putting it in layman's terms again; A magnetic field on the piston generates a current on the rod. The time it takes for the current to travel from the rod to the sensor, tells the computer how far along the rod the magnet is, therefore telling us the exact stroke.

You have the general idea right. It is like a sonar or radar. However, actually the transmitter sends a pulse down the rod. The pulse gets to the magnet at the speed of light or close to it. The current pulse in the magnetic field causes the wave guide inside the rod to twist. What is measured is the time this twist takes to get back to the head.

It isn't really necessary to know all the details above. What is important is that hydraulic applications can get reliable, fast and accurate feedback. This capability allows one to do sophisticated applications you see in the previous links and compete with the servo motor guys.

What you should also realize is that one can use a PID to control force too. One just adds pressure transducer to either end of the cylinder or use a load cell for feedback. Many press applications use both position controls and force control depending what part of the cycle the press is in.

That's another thing I don't have a firm grasp of yet. Servos, torque motors and other electrohydraulic valves. The Mobile Hydraulic manual doesn't tell you much, and Bud's explanation in Fluid Power Basics is a little hard to understand, too.

quote:

Originally posted by Josh Cosford:
That's another thing I don't have a firm grasp of yet. Servos, torque motors and other electrohydraulic valves.

You do know you can look this information up on the web. There is excellent information on servo valves in this site.
http://www.hydraulicspneumatics.com/200/TechZone/Hydrau...Zone-HydraulicValves
http://www.hydraulicspneumatics.com/200/TechZone/Hydrau...Zone-HydraulicValves

I have an article about servo valves too
http://www.hydraulicspneumatics.com/200/GlobalSearch/Ar...False/13913/Nachtwey

Moog has excellent tech information.
http://www.moog.com/media/1/proportionalandservovalvetechnolog-fpjarticle.pdf

quote:

The Mobile Hydraulic manual doesn't tell you much, and Bud's explanation in Fluid Power Basics is a little hard to understand, too.

Moblie most hydraulics doesn't need the precision that servo valves provide or can tolerate the energy loses of a servo valve.

Basically there are two forms of servo valves. What is called a servo valve uses oil to move the main spool. A flapper or torque motor diverts oil to either side of the spool to make it move. The torue motor is moved by electric current from the controller.

There are also servo quality proportional valves. These uses a solenoid like the coil on a speaker to move the spool. These valves usually require a +/- 10 volt signal to control the spool position.

In either case the main thing to look at is the flow, obviously, spool linearity and frequency response. Comparing servo valves is difficult because all the manufacturers 'hide' the true perfromance.

Most use servo applications actually use 'servo quality spools' in proportional valves.

The important thing to remember is that the flow should be proportional to the control signal. The valve response should be fast enough for the application. Most valve frequency response ratings are a joke.

Peter;

When is the next TRAINING SESSION on Infinitely Variable Flow Valves?

It would be nice if all of us could attend but there seems to be no schedule. At least I can't find one.

Also, it would be nice if we could ask questions if that's not too much trouble.

It would be nice to have a little more detail though:

What is a "Flapper," a "Torque Motor?"

That solenoid that thinks it's a Speaker Cone sounds interesting also. Could you go into a little more detail?

How can "Oil move a Spool" with pressure on both ends.

'servo quality spools' in proportional valves" sounds like a real winner to me.

Also that obvious "Spool Linearity and Spool Response" is a new one to me and probably others.

????????????????????????????????

Just Kidding

quote:

Just Kidding

That is why I said

quote:

The important thing to remember is that the flow should be proportional to the control signal. The valve response should be fast enough for the application. Most valve frequency response ratings are a joke.

Unless you are designing valves you only care about the flow response to your control signal, not how it is achieved. You don't need to know how an car engine works to drive to operate it, keep it fueled and oiled.

quote:

You don't need to know how an car engine works to drive to operate it, keep it fueled and oiled.

No, but I know that when I hear a "tickity tick" under the hood of my car, then the hydraulic valve lifters are telling me I'm low on oil without even checking. I also know that 5W50 Mobil 1 Synthetic not only works great all year around here in Canada because of its high VI, but keeps my engine protected when I hit the racetrack as well. I also know that premium fuel, especially ethanol free fuels, have higher energy density, increasing fuel economy(though not always decreasing cost).

I know more about cars, than I do about hydraulics. Do I need to know how much I do? No, but I can guarantee I get more enjoyment out of my vehicles on a daily basis because I know what I know. Just because some things "work" with minimal background knowledge, it doesn't give us the excuse to be ignorant.

quote:

Originally posted by Josh Cosford:
I know more about cars, than I do about hydraulics. Do I need to know how much I do? No, but I can guarantee I get more enjoyment out of my vehicles on a daily basis because I know what I know. Just because some things "work" with minimal background knowledge, it doesn't give us the excuse to be ignorant.

Good, but if you are designing hydraulic circuits you don't need to know exactly how the valve works. There are other things to master first.

Peter stated:

quote:

Good, but if you are designing hydraulic circuits you don't need to know exactly how the valve works.

However, when you are Trouble Shooting a circuit that is not operating as planned it sure helps to have a thorough knowledge of all the components operating charecteristics and how they affect the function in that circuit.

Internal Pilot Supply, External Pilot Supply, Internal and External Pilot Supply, Internal Drain and External Drain are just a couple of minor changes to a lot of valves that can cause wierd things to happen when the wrong function is applied. Often it is not oly wierd but DANGEROUS to PERSONS, the Circuit or the Machine Members

In fact, anyone who is Designing or Trouble Shooting a circuit wthout a thorough understanding of how the components in that curcuit function is working Blind Folded.

Any circuit Designed by someone without a thorough knowledge of how components operate and should be applied reminds me of Peter's answer to a post in the "How were You Schooled/Trained" thread.

"There are too many hydraulic screw ups that just force people to use different technology (servos and VFDs) and ruin business for the rest of us."

I have seen a lot of mistakes on circuits because a wrong component was applied that worked "Sort Of" but not reliably and Smoothly as one desiged for the function. On the other hand I have come across components operating perfectly in functions I had never heard they were designed to do or were shown in any catalog or training material in that application.

Anyway, Fluid Power Training is still top priority on my list and will be until I hear better arguments to the contrary.