Hi,

When designing a controller, my understanding is that to look for stability, you have to consider the gain and phase margins. However, i have a design that has both positive gain and phase margins (>10 dB and > 60 degrees) yet the step response to the system is still unstable. What other tests can do i to test the stability?

i have also heard of the root locus plot, but i think usually looks at the gain parameter and its effects on the poles and zeroes. Can this be unstable even with the above conditions? (+ GM, +PM)

TIA

 

May I ask what is the transfer function of this unusual system? As you suggest the root locus shows how the closed loop poles vary with loop gain variation.

 

Can you post a plot showing the gain and phase margins of your system?

 

Did you actually test (measure) the gain/phase or is this something that came out of some simulation?

 

Hi,

When designing a controller, my understanding is that to look for stability, you have to consider the gain and phase margins.
TIA

The critical point for most systems is what the gain is when the systems total phase shift reaches -180 degrees. If you still have positive gain then, the system is an oscillator. Another way to look at it is to see what the phase margin is when the system's gain crosses 0dB (unity gain). Phase margin is the difference between the measured phase shift and -180. Ideally, you would like about 30 - 60 degrees of phase margin at the unity gain crossover point, but in some cases 20 degrees is OK.