06-29-2011 03:43 AM
Hi,
Currently I have a setup that basically operates a motor to move a certain object between certain movement limits and at a certain speed. As a result of the movement, another part of the machine will move also. This movement is registered by a sensor. I can successfully acquire data and move the motor, but sometimes we have to operate at extremely low speeds. In this case it's not necessary to move to the full extent of the limits. The important thing is that the movement of the second part is steady-state (that it reaches a DC value, or approximately a DC value). This steady state is a priori unknown.
The goal of my current task is to implement the motor movement as such, that when the second part has reached it's steady-state, the motor stops and does it's next movement.
To do this, I tried two approaches:
Is there anyone that has crossed paths with a similar problem yet? So basically, I want to be able to detect on a real time scale, whether a certain signal is saturated/stead-state.
Thx for your time!
07-06-2011 04:37 PM
That can be a tough problem.
The slope approach seems more reasonable. The standard deviation tells you more about the noise than about the speed changes, especially for slow changes.
It really depends on the signal to noise ratio, how fast the system stabilizes, and other parameters.
When you change speeds, is it a step change or a gradual one? How does the size of the signal from the sensor for the smallest significant change compare to the noise levels? Do you have any other signals available, such as the speed input command or a motor speed sensor?
Lynn