[wklove]

Hey guys, I am wanting to characterize the capabilities/performance of a cnc and need to do this by delivering small amplitude, variable frequency input commands and measure the response to place on a Bode Plot (amplitude vs. frequency). I am retrieving the x,y, and z values from the machine controller in real time as it moves. So my question is how do I accomplish this?

I am having trouble trying to understand how this machine motion (x,y,z) correlates to the bode plot. Does anyone have insight on this or has done this themselves before? I think I might be misunderstanding the point or instructions I have been given. How do I apply a sinusoidal command to the machine, measure that amplitude and frequency?

[TaK]

A bode plot would be an unconventional way to characterize a mill/router.

Why would you do that?

[wklove]

Trying to do this to see if the machine is repeatable. As this machine is used day in and day out, there isnt anything characterizing whether this machine is working the same each day. So we want proof of this after warm up and it needs to be done remotely.

[harryn]

Hi, as usual, wikipedia has some insight on the matter.

Bode plot - Wikipedia, the free encyclopedia

My perception is that you aren't applying a sinusoidal input signal, you are applying a step function and measuring the response, but this interpretation might be wrong.

In order to measure the response, the cnc machine would need to be able to identify it's positioning independent of the programmed position. (closed loop feedback), if I understand the concept.

If you are trying to do this using info from all day operation, remote control, I guess what you are really plotting is setpoint and offset on the same plot.

[Khalid]

We use Bode Plot in Transient state of our turbo machinery... This gives us information about phase changes, critical speed and anyother abnormalities in proximity locations of the Rotor... I don't know how you are going to use with the CNC Machine and where you going to install the proximity probes?

This will only good for spindle startup/shuddown...

[daconfusion]

hopefully my comments can help, or be completely wrong (tends to happen after work and lots of beer):

I agree with the closed loop feedback comment. I think you could characterize your system in terms of position, velocity and acceleration by comparing the input and the output. (Then jusify adding PID or some P, I, or D.)

Also, what about looking at speciifc component natural frequencies? If you could plot your bode and identify the peaks, and figure out what each applies to, you could identify a potential resonance problem.

For example the critical speed of your ballscrew could be one resonance peak which should have its natrual frequency far away from lets say the frame?

If you have the Systems Dynamics book by Ogata, I believe there is a section on servo control/automation feedback analysis. This probably is standard in your typical systems engineering textbooks (undergrad).

Edit: What about looking at the frequency of the cutting tool to see if that has any effect on your entire system? Wouldnt that be a periodic function if you were to assume rotary imbalance?