ac servo motor

[mfi.edu.my]

Dear Sir,

I am using 3 ac servo motor, in my XYZ axis system. I intend to develop the transfer function for each axis (or is it enough to discover one only)which later will be used to design the PID and Fuzzy controller. And then test on the real application using fielpoint FP 2000, FP AIO 610 and FP QUAD 510.

Currently I am using PCI 7344 motion card. Since it is embedded with only PID algorithm, I purchased the fielpoint to enable me the Fuzzy algorithm.

My question: what are the data that I should look on in order to discover the transfer function of the controller and the system; by using the system decription toolkits.

TQ.

SADIAH

[Irene B]

Hi Sadiah,

NI does not provide a transfer function for the controller for modeling, but there are some parameters that you might find helpful.  In Measurement and Automation Explorer, you can set the proportional, integral, and derivative gain (under Control Loop Settings) and view the Bode plot (under Calibration >> Servo Tune).

Hope this helps,
Irene Chow
National Instruments
Applications Engineer

[sadiah]

I already played around the MAX and tune the system manually.

Since it is very tedious to figure out the transfer function(dynamic modelling) of the system since it involves many nonlinearities, I bealive using NI software toolkits - i.e SYSTEM DESCRIPTION toolkits, might help me to figure out the modelling (so i can create the right controller offline - to avoid amaging the system.

How do i go about this?

TQ 

[Irene B]

Hi TQ,

We do have two LabVIEW toolkits that you can look at: the NI LabVIEW Control Design Toolkit and the NI LabVIEW System Identification Toolkit.  If you have any specific questions about using them with your application, feel free to post them and another applications engineer that supports these products can help you.

Good luck,
Irene Chow
National Instruments
Applications Engineer

[xf]

Hi,

I am using DC motor, the input voltage is a 30% duty cycle PWM. Ouput is velocity of a print head, by a linear encoder reads counts. To get the plant model, i've got them as stimulus and response signal(see attachment). But the processing of data is hard. Here, it gives big MSE. How should I pre-process the data to make them in the same form? Sorry if my phrasing is bad.

Or is my response data kind of weird?

Thanks...