Motion Madness

[Rod]

Norm,

Thanks, I wasn't thinking that far ahead.

A chip is a chip is a chip.

Rod

 

[Peter Nachtwey]

We are flooding the port with calls and responses. Our programmer can't find it, Acroloop can't find it and I am unwilling to give AL our source code.

Why all the calls? You shouldn't need to get the status any faster than the HMI needs it unless the PC needs the info for control.

They probably don't have the time to wade through 100,000 lines of our code and I can't afford to send our programmer there for the time it would take. We/I am STUCK!

THAT IS A LOT OF CODE. WHAT ARE YOU DOING? We can do a flying shear in less than 20 steps ( commands ). An injection molding machine may take 50.

We only buy about 3 motion control units per month, so we are too small to be of concern.

That enough to get some attention. We have plenty of customers that buy only 10 a year. They add up after a while.

On your hydraulic servo valve controls do you incorporate a 'dithering frequency/voltage' to prevent stiction in the valve spool? If so, what freq do you find to be the best? I've always used about 400Hz.

Thanks,
Rod

We did in our first generation product. We didn't on our second generation product. Dither wasn't necessary with the newer proportional and servo valves. Some valve drivers have the ability to add dither to the controller signal. 400 Hz should be just find for bigger systems with time constants in the 20 millisecond range. If your system is faster then you may want to go without the dither.
The frequency is really dependent on the update time of the controller. If the controller updates every millisecond then the frequency will be 500Hz. The dither period must be some even integer number times the the controller update period.

I have seen system with time constants as low as 2.2 milliseconds. A 400HZ dither would make the system vibrate at that rate. You really need to know the response of the system.
I don't like to use dither. It only keeps the valve spool from sticking which it shouldn't anyway with a newer better quality valve.

You mentioned PowerPC. You use a MAC for the front end?

No, it cost too much to develope the front end code to do it for a handleful of macs. I have a Mini Mac. I have used that to do assembly language and C programming for our product. Most gets done on a PC using Metrowerks.

To me that would imply using USB or Firewire ports and a stand alone motion control box.

The PowerPC version of our products supports Ethernet and USB.

I love my MACs but wouldn't base an industrial machine system on them. I've only seen two.

Macs aren't designed for industrial control. However, Darwin ( a Apple version of BSD unix ) is far superior to Windows. Even though the Mac hardware is not as fast as the PC hardware the Mac still perform better. Darwin multitask a lot more smoothly than windows. The windows message queue really slows it down.

I just about have the .pdf done that shows the need for a jerk feed forward and a higher order feed forward.

 

[Peter Nachtwey]

Full Circle.

ftp://ftp.deltacompsys.com/public/PDF/Mathcad%20-%20FeedForward3.pdf

I made a .pdf that show how the higher order feed forward terms affect motion.

Pages 1 and 2 are defining the system.
Page 3 sets up the state space arrays in discrete time.
Page 4 shows the systems response to a step input. I did this just to show that this system is compliant. Hopefully none of you see something like this. I have too many times.
Pages 5 and 6 Show how the actual will pretty much follow the desired motion profile using feed forwards. It takes a while for the actual to follow the target because of the abrupt change from stationary.
Page 7 shows the same system responding to the same motion profile but the higher order feed forward gains are off. Notice how the actual overshoots the target. This is because the sinewave is running at the same frequency as the natural frequency of the load on the shaft. Look at the Bode plot at the end. You can see there is a resonance peak at 10 Hz. Normally you would want to run at frequencies higher than 1/3 of the resonance peak. Norm,s motion profile is equivelent to about a 1/(2*.31) Hz ramp.
Pages 8-10 I generate a trapezoidal ramp using Norm's parameters. Since a trapezoidal ramp does not have terms higher than acceleration there can be no jerk or higher feed forward terms in the output. Notice the oscillations.
Page 11-12 I generate a 5th order ramp using Norm's parameters. The peak acceleration is actually higher than the .3G than what Norm specified. That makes up for the accleration being lower at the end of the ramp. However, the ramp times and distances are identical. Notice that it doesn't oscillate as much as the trapezoidal ramp. The difference may not seem like much but it would be much more noticeable if the acceleration and deceleratins times were shortened. Then the following error would approach that of the sine wave example.
Page 13 is a Bode plot showing the gain and phase response as a function of frequency.

Notice I am ONLY using feed forwards. I am not using a PID. In a real example the feed forwards get you very close but closed loop control is still required.

 

[kamenges]

Thanks for putting the time in on this, Peter. It really has been informative. And, you're right. I hope I never see the system you have modelled there. That would be a challenge.

Keith

 

[ndzied1]

Thanks Peter, now I'll be up late the next 4 weeks trying to figure out that MathCAD sheet

 

[Peter Nachtwey]

OT, This is impressive example of motion madness.

It is in spanish but it is easy enough to figure out.
http://www.tecnun.es/asignaturas/control1/proyectos/pdobleinv/video.htm
I think these guys have passed passed PID 101. What do you think? My hat is off to these guys.

The single inverted pendulum has been beat to death an can be done number of ways. The double inverted pendulum is something I have heard about but haven't seen. It is all just math and good engineering in both the motion and mechanical area. Still I find this impressive.

The double invert Pendulum examples take special code. I haven't seen a practical uses for this or any related applications in the industrial field.