Dak,
The VI and the documentation being different isn't good. As far
as saying which one is correct, depends on what you want to
achieve. The implementation in both the documentation and the VI
could be correct. The implementation described in the
documentation allows you to set alpha and then have the cut-off
frequency of the derivative action depend on the derivative parameter
you choose. This can be convenient. However the
implementation of the VI lets you set a fixed cut-off frequency that
does not depend on the derivative coefficient. This is good if
you have noise of a known frequency that you want to eliminate.
The misleading part is that the input to the VI is called Alpha.
A better name might be Derivate Filter Time Constant.
Luckily it is easy to convert between the two implementations.
Specifically to get the behaviour described in the help, multiply Td
(or Kd depending on the type of PID) by your desired Alpha and use that
as the input to the VI. I have attached a snippet of code to show
this (CD_PID.vi).
Arun,
If you make Ti arbitrarily large you will see the effect of the
integral action become smaller. But there will always be some
intergral action present. For this reason a Ti of 0 is special
and means no integral action. What happens in the VI is that Ki
is calculated from Ki = Kc/Ti. If Ti is 0 then Ki is Inf which
then triggers the PID Integral Action.vi to ignore the integral action
part. So set Ti to 0 and you will see no intergral action.
Carl L
National Instruments
