Unit 3: Time Response, Part 3: Systems With Additional Poles or Zeroes
Uploaded by
Nikhil PanikkarUnit 3: Time Response, Part 3: Systems With Additional Poles or Zeroes
Uploaded by
Nikhil PanikkarSystems with Additional Poles or Zeroes
Pole-Zero Cancellation
Unit 3: Time Response,
Part 3: Systems with Additional Poles or Zeroes
Engineering 5821:
Control Systems I
Faculty of Engineering & Applied Science
Memorial University of Newfoundland
February 5, 2010
ENGI 5821 Unit 3: Time Response
Systems with Additional Poles or Zeroes
Pole-Zero Cancellation
Systems with Additional Poles or Zeroes
The formulas for Tp , Ts , etc... are specific to second-order
systems. However, sometimes systems with additional poles or
zeroes are well approximated as second-order.
Consider the addition of a third pole to a two-pole system. First,
note that the only way to add exactly one more pole is for the pole
to be on the real-axis.
The system response is as follows,
bc A Bs + C D
C (s) = = + 2 +
s(s 2 + as + b)(s + c) s s + as + b s + c
In the TD, the third term adds De −ct to the second-order step
response. What effect does this have?
ENGI 5821 Unit 3: Time Response
Consider the three following cases for the position of the third
pole:
If the third pole is near the other two then its effect is strong.
However, as it is moved to the left it decays at a faster rate and
therefore has less effect on the overall signal.
Even if the exponential decays quickly, we may still be concerned
about its initial effect. Actually, the effect of the third pole
decreases in magnitude (as well as decay rate) as the pole moves
to the left. Consider again the system response:
bc A Bs + C D
C (s) = = + 2 +
s(s 2 + as + b)(s + c) s s + as + b s + c
We can evaluate these constants:
ca−c 2
A=1 B= c 2 +b−ca
2 2 a−bc
C = cac 2−c
+b−ca
D= −b
c 2 +b−ca
As c → ∞ D → 0. Thus, for a third-pole that is far to the left of
the dominant poles, the system response is approximately
second-order. How far is far enough to discount this third pole?
Rule of thumb: If the pole’s real part is five times the real part of
the dominant poles, then the system can be approximated as
second-order.
e.g. Which of T2 or T3 is a better approximation to T1 :
24.542
T1 (s) =
s 2 + 4s + 24.542
24.542
T2 (s) =
(s + 10)(s 2 + 4s + 24.542)
24.542
T3 (s) = 2
(s + 3)(s + 4s + 24.542)
The step responses for T2 and T3 have the following form:
ci (t) = 1 + k1 e pi t + k2 e −2t cos(4.532t + φ)
Solution: The pole for T2 is 5 times the real component of the
dominant poles and can therefore be approximately neglected.
Additional Zeros
What is the effect of adding a zero to a second-order system? The
addition of zeros affect the response, but the form of the response
is preserved (it will still be an exponentially damped sinusoid).
Consider adding a real-axis zero to the left-plane. We begin with a
second-order system with poles at −1 ± j2.828.
It appears that the closer the zero is to the dominant poles, the
more impact it has.
Consider the partial-fraction expansion:
s +a A B
T (s) = = +
(s + b)(s + c) s +b s +c
(−b + a)/(−b + c) (−c + a)/(−c + b)
= +
s +b s +c
What if a is large relative to b and c?
h 1/(−b + c) 1/(−c + b) i a
T (s) ≈ a + =
s +b s +c (s + b)(s + c)
In this case a acts as a simple gain factor, which does not alter the
form of the response.
This can be illustrated in another way. Our original system’s
response is as follows:
C (s) = R(s)G (s)
Assume the system is now modified such that
G 0 (s) = (s + a)G (s). The new system response will be,
C 0 (s) = R(s)G 0 (s) = R(s)(s + a)G (s)
= (s + a)C (s) = sC (s) + aC (s)
The response is composed of the derivative of the original response
plus a scaled version of the original response.
C 0 (s) = sC (s) + aC (s)
The derivative of the second order response is initially positive
which means that the zero affects the response by increasing
overshoot.
SIMULINK DEMO
If a is negative then the initial response may be negative, causing
the system to initially move in the opposite direction:
Such a system is known as a nonminimum-phase system.
Pole-Zero Cancellation
We can cancel poles and zeros even when they are not exactly
equal. Consider the following system:
26.25(s + 4)
C1 (s) =
s(s + 3.5)(s + 5)(s + 6)
1 3.5 3.5 1
= − + −
s s + 5 s + 6 s + 3.5
The residue of the pole at 3.5 is not negligible compared to the
other residues. We cannot accurately approximate this system as
second-order. However, we can for the following system:
26.25(s + 4)
C2 (s) =
s(s + 4.01)(s + 5)(s + 6)
0.87 5.3 4.4 0.033
= − + −
s s + 5 s + 6 s + 4.01
0.87 5.3 4.4
≈ − +
s s +5 s +6
