Filters in closed-loop control

[Riscilla]

Hi friends,

I have a question regarding the use of filters in closed-loop control systems. Very often in a system like that there is noise from the sensor, and we have to put a filter after the sensor to eliminate the noise. However, all filters produce a phase shift as well as an amplitude attenuation into the measured signal, therefore the signal we get is already distorted. My question is that how could we control the system in these circumstances?

To make things clearer, I will give an example: Suppose I intend to setup a digital control system with a update frequency (sample rate) of 1 (kHz). But the measured signal is very noisy, so I have to use a filter to reduce the noise. After several tries, I come up with a filter which has a -3db of 200 (Hz). With this sensor, all noises are removed, however, there are also a lot of phase shift and amplitude attenuation. Can we still control the system with that filter?

Thank you for reading my question. Any answer would be very appreciated.

Regards,

Riscilla

 

[schip666!]

It all depends on the rest of the system's time constants and response requirements. A simple motor speed control updated every .01 sec is probably pretty good, but other applications may need quicker response or more accurate tracking.

 

[crunchman]

i deleted original post. i did not carefully read your question.

the problem is in the sensor. i think you only need to add a compensator to correct phase shift. there should be a chapter on compensator design in control systems books.

another way is to mathematically apply filtering/ compensation/ control by programming a microcontroller or interfacing with a scientific package (scilab, matlab, labview).