#### YukiWong

Joined Dec 5, 2011
27
The original continuous system has a overshoot of 16.3% and the discrete system of the continuous system with 100ms sampling time has a overshoot of 20.6%.

Can anyone explain to me why the discrete system for the 100ms sampling time has much more overshoot compare to the original continuous system. and also why the bode gain of the discrete system with 100ms have 2 gain margin. Thank you very much.

#### t_n_k

Joined Mar 6, 2009
5,455
One would expect a sampled system to give a more oscillatory response than its equivalent 'pure' continuous system. The sampling process gives rise to dead time which is known to destabilize a system. If you make the sampling frequency high enough (notionally sufficiently higher than the system bandwidth) then the discrete and continuous responses should converge.

Your bode plot shows two distinct gain margins because the sampled data phase plot crosses the 180° axis twice - the second (higher) value presumably being at the Nyquist frequency of 31.42 rads/sec.

Last edited:
• YukiWong

#### RRITESH KAKKAR

Joined Jun 29, 2010
2,829
The original continuous system has a overshoot of 16.3% and the discrete system of the continuous system with 100ms sampling time has a overshoot of 20.6%.

Can anyone explain to me why the discrete system for the 100ms sampling time has much more overshoot compare to the original continuous system.
Which software are you using??

#### YukiWong

Joined Dec 5, 2011
27
One would expect a sampled system to give a more oscillatory response than its equivalent 'pure' continuous system. The sampling process gives rise to dead time which is known to destabilize a system. If you make the sampling frequency high enough (notionally sufficiently higher than the system bandwidth) then the discrete and continuous responses should converge.

Your bode plot shows two distinct gain margins because the sampled data phase plot crosses the 180° axis twice - the second (higher) value presumably being at the Nyquist frequency of 31.42 rads/sec.
Thank t_n_k,
Which gain should be chosen according to the bode plot?

#### YukiWong

Joined Dec 5, 2011
27
• RRITESH KAKKAR

#### t_n_k

Joined Mar 6, 2009
5,455
Thank t_n_k,
Which gain should be chosen according to the bode plot?
Not the one at the Nyquist frequency.