Re: Sampling time

Jan Jantzen (jj@iau.dtu.dk)
Fri, 21 Feb 1997 07:16:43 +0100


It should be chosen somewhat smaller than the dominating time constant in
the process, for instance between 1/10 and 1/5 of the time constant.
If the sample period is too short, the computation of CE will become
too sensitive to noise. This normally shows up as a restless control
signal. If CE happens to be implemented as a difference, several
gains depend on the sample period; a change in sample period must then be
followed by a compensation in a gain factor to keep the proportional,
integral and derivative gains intact.

The sample period can also be related to the derivative gain Td. Astrom
and Wittenmark (1984) suggest that the sample period should be between
1/10 and 1/2 of Td. In connection with the Ziegler-Nichols rules,
this implies that Ts should approximately equal 1-5 percent of the
ultimate period Tu.

Regards, Jan Jantzen

Technical University of Denmark | Tel +45 4525 3561 (direct)
Att: Jan Jantzen | Fax +45 4588 1295
Dept. of Automation, Bldg. 326 | Jantzen@iau.dtu.dk
DK-2800 Lyngby, DENMARK | http://www.iau.dtu.dk/~jj

On Thu, 20 Feb 1997, Ali Khalaf wrote:

> What is the required sampling rate required for a fuzzy logic
> controller. I am mainly concerned with the application of fuzzy
> controller in power system stabilizers. Also what effect does varying
> sampling rate have ?