Peter Nachtwey convers this topic in many of his articles. (see www.powermotiontech.com) He recommends that the actuator natural frequency should be 3-4x the frequency of acceleration for sine waves. Larger piston area is the easiest way to increase the natural frequency. If the valves aren't already mounted directly to the actuator reducing the volume of oil between the valve and cylinder will also raise it. Sudden changes in the control valve such as using square waves or linear ramps, can excite multiple frequencies as well. ISZ

ISZ, Thank you for the help.

More information would be useful. The natural frequency can be lower than 3 to 4 times IF the right controller is used. However, for most controllers with only PID, the natural frequency should be 3 to 4 times the frequency of motion with 3 times being barely acceptable.
The natural frequency is roughly proportional to the diameter of the cylinder so larger diameter cylinders are better. The problem is that increasing the diameter by a factor of 2 increases the required flow by a factor of 4.

Peter Nachtwey
Delta Computer Systems
http://www.deltamotion.com
http://forum.deltamotion.com/
IFPE Hall of Fame Member

Hi, PNachtwey, Thank you for the help.
I used MTS Flex Test 200 to do the fatigue test of the vehicle. The load feedback commands are plotted in the attachment. Those cylinders of 8001 to 8004 subjected to unstable oscillations. I don't know why a stable system became unstable suddenly.
Oil pressure 3000 psi
Cylinder 8001L : 2.5" Bore, 1" Rod, 16" Stroke
Cylinder 8002L : 2.5" Bore, 1" Rod, 24" Stroke
Cylinder 8003L : 1.5" Bore, 0.625" Rod, 12" Stroke
Cylinder 8004L : 1.5" Bore, 0.625" Rod, 12" Stroke

How to tackle this phenomenon?

• https://files.engineering.com/getfile.aspx?folder=a9f0fb98-0fed-4ae8-941d-f

Yes, I can see the oscillations but there is no line for the control output.
I am assuming the plot shows the actual net force on the vehicle. What is the target force?

Quote:

The load feedback commands are plotted in the attachment

Are those the commands or the feedback. Commands are signals to the valves. Feed back comes from the pressure sensors or load cell.
I am assuming this has worked before.
Are you sure nothing has changed?
Air in the system will screw things up pretty quickly. Has there been any maintenance done on the system lately?
I am assuming that the motion is minimal and force comes from a load cell and the actuators are attached to the vehicle.
Are there any lose linkages or worn bearings?
What command are you giving the actuators? Simply hold a force?
Normally force is integrating the pressure ( times area) increase due to the flow. For the force to oscillate that means the flow is going in and out so the valve is shifting to let oil in and out. Have you tried reducing the gains on the controller. BEFORE CHANGING ANYTHING, SAVE THE CURRENT SETUP.

Peter Nachtwey
Delta Computer Systems
http://www.deltamotion.com
http://forum.deltamotion.com/
IFPE Hall of Fame Member

I revise the attachments.
Because the test is fatigue test, so the loads of all cylinders are not constant. Some cylinders are push and the others are pull.
Every plot contains one cylinder's load command and feedback from the load cell.
The results show the feedback following the command, although there is 20 lbfs difference.
Why the feedback becomes unstable suddenly, the oscillation frequency is approximate 15Hz.
This phenomenon has happened on different regions of the test object during the test process.
Depending on different regions, those oscillation frequencies are different which are between 15 to 40 Hz.
I have checked the linkages between the cylinder and the test object, and do my best to remove the air contained in the cylinder.

• https://files.engineering.com/getfile.aspx?folder=e417f43e-5e04-483c-ab9a-3

another cylinder

• https://files.engineering.com/getfile.aspx?folder=66f94419-0223-42ba-91ad-5

I am surprised that MTS doesn't have a better plotting ability. Is the "load command" the force set point or the control output to the valve? If it is the control output to the valve, I can see it that it isn't the controller/output causing the problem. The controller doesn't seem to be responding to the feedback at all. That is strange and doesn't make any sense. When the oscillations occur, can you hear them, feel them? Can something be vibrating the load?
The question is whether the cylinders are causing the vibration or something else is.

Isn't there someone from MTS that can help. They know their systems.
Do you know how to use an oscilloscope?

Peter Nachtwey
Delta Computer Systems
http://www.deltamotion.com
http://forum.deltamotion.com/
IFPE Hall of Fame Member

Because we are doing the fatigue test of the vehicle. The load spectrum is defined by force.
MTS Flex Test 200 controls every servo valve to control the corresponding cylinder.
The unstable oscillation phenomenon has happened on different regions of the test object many times.
Depending on different regions, those oscillation frequencies are different which are between 15 to 40 Hz.
During 15 Hz vibration, we can hear the sound of structural vibration and I think it is a structural resonance.
(because the vibration test shows the structural natural frequency around that region is approximate 15Hz)
But the other frequencies, no sign shows before unstable oscillations come.

Why do you think the cylinders are causing the vibration, how to check?

Why use an oscilloscope can identify this problem and how to use?