I was reading about designing of controller by state state model from book by Norman S. Nise

And author suggested that if state space representations given in problem are of form other than phase variable form then best method to design a controller would be

1.transform given system to phase variable form

2.then design the feedback gains

3.then again transform the designed system back to original state space form

So my question

1.when we transform from one of state space representations to another , isn't zeroes of system may changed?

Or in other words

isn't transfer function gets changed due to transformation

As I try to figure out myself here is my attempt enter image description here

enter image description here

And after all this calculation to me it seems that both transfer function may not be equal always


example given in book shows that it has same transfer function in both transforms so I thought it(example) may be a special case

So can anyone explain

Mathematical expression I got can be solved further to obtain a relationship between both transfer function ?


my conclusion is right that transfer function may or may not changed due to transformation?

  • 2
    \$\begingroup\$ If the transfer function changes, then it is not the same system. The transformation is supposed to preserve the input output relationship; aka transfer function \$\endgroup\$
    – AJN
    Aug 31, 2020 at 13:43
  • \$\begingroup\$ @AJN , can you please tell then where my mathematical expression is wrong or how to simplify it? \$\endgroup\$
    – user215805
    Aug 31, 2020 at 13:53
  • 2
    \$\begingroup\$ \$sI-P^{-1}AP = sP^{-1}IP-P^{-1}AP = P^{-1}(sI-A)P\$. From this try to see if the last expression in the second figure can be simplified into \$C(sI-A)^{-1}B\$. I think it will be. Check the derivation carefully. \$\endgroup\$
    – AJN
    Aug 31, 2020 at 14:11
  • \$\begingroup\$ @AJN , thanks , very precise as always! \$\endgroup\$
    – user215805
    Aug 31, 2020 at 14:22

1 Answer 1


The similarity transformation is only a change of basis (co-ordinates) and doesn't change the input to output relations, viz, the transfer function (TF).

The similarity transformation preserves the eigen values; i.e. poles of the system; i.e. denominator of the TF. However, to show that the TF doesn't change, we need to show that the zeroes do not change either.

Assuming that \$P\$ invertible,

$$ \begin{align} \frac{Y(s)}{U(s)} ={} & CP\ (sI - P^{-1}AP)^{-1}\ P^{-1} B\\ {}={} & CP\ (sP^{-1}IP - P^{-1}AP)^{-1}\ P^{-1} B\\ {}={} & CP\ (P^{-1}\ (sI - A)\ P)^{-1} P^{-1} B\\ {}={} & CPP^{-1}\ (sI - A)^{-1}\ P P^{-1} B & \tiny{(ABC)^{-1} = C^{-1}B^{-1}A^{-1}}\\ {}={} & C (sI - A)^{-1} B\ \ \ \blacksquare\\ \end{align} $$

  • \$\begingroup\$ ,out of context ,but if you could suggest a good book for complete understanding of controller design because I have lots of trouble understanding it ,I'll thankful to you \$\endgroup\$
    – user215805
    Sep 2, 2020 at 17:04
  • \$\begingroup\$ Nise is my go to book too. I have referred Ogata and B.C Kuo also from time to time; but I don't know if they are easier to follow. Ask in the EE.se Electrical Engineering Chat for suggestions. \$\endgroup\$
    – AJN
    Sep 3, 2020 at 10:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.