EE 3CL4, §4

1 / 55
Tim Davidson

Stability
Condition in terms of
poles
Condition in terms of
denominator
coefficients EE3CL4:
Routh Hurwitz
condition Introduction to Linear Control Systems
Basics
Disk drive example Section 4: Stability and Routh-Hurwitz Condition
Dealing with zeros
Zeros in first column
Zero rows

Using Routh
Hurwitz for Tim Davidson
design
Turning control of a
tracked vehicle
McMaster University

Winter 2020
 EE 3CL4, §4
2 / 55
Tim Davidson Outline
Stability
Condition in terms of
poles
Condition in terms of 1 Stability
denominator
coefficients
Condition in terms of poles
Routh Hurwitz
condition Condition in terms of denominator coefficients
Basics
Disk drive example
Dealing with zeros
Zeros in first column 2 Routh Hurwitz condition
Zero rows
Basics
Using Routh
Hurwitz for Disk drive example
design
Turning control of a Dealing with zeros
tracked vehicle
Zeros in first column
Zero rows

3 Using Routh Hurwitz for design

Turning control of a tracked vehicle
 EE 3CL4, §4
4 / 55
Tim Davidson Stability
Stability
Condition in terms of
poles
Condition in terms of
denominator
coefficients

Routh Hurwitz
condition
Basics
Disk drive example
Dealing with zeros A system is said to be stable if all bounded inputs r (t) give
Zeros in first column
Zero rows rise to bounded outputs y (t)
Using Routh
Hurwitz for
design Counterexamples
Turning control of a
tracked vehicle
• Albert Collins, Jeff Beck (Yardbirds),
Pete Townshend (The Who), Jimi Hendrix,
Tom Morello (Rage Against the Machine),
Kurt Cobain (Nirvana)
• Tacoma Narrows
 EE 3CL4, §4
5 / 55
Tim Davidson Conditions for stability
Stability
Condition in terms of
poles
Condition in terms of
denominator
coefficients

Routh Hurwitz Z ∞
condition
Basics
y (t) = g(τ )r (t − τ ) dτ
Disk drive example −∞
Dealing with zeros
Zeros in first column
Zero rows Let r (t) be such that |r (t)| ≤ r̄
Using Routh
Hurwitz for
Z ∞ 
|y (t)| =  g(τ )r (t − τ ) dτ 
design
 
Turning control of a
tracked vehicle
−∞
Z ∞
 
≤ g(τ )r (t − τ ) dτ
−∞
Z ∞
 
≤ r̄ g(τ ) dτ
−∞

R∞  
Using this: system G(s) is stable iff −∞
g(τ ) dτ is finite
 EE 3CL4, §4
6 / 55
Tim Davidson Condition in terms of poles?
R∞  
Stability
We want −∞ g(τ ) dτ to be finite
Condition in terms of
poles
Condition in terms of
Can we determine this from G(s)?
denominator
coefficients
We can write a general rational transfer function in the form
Routh Hurwitz
condition Q
Basics
K i (s + zi )
Disk drive example G(s) = N Q Q 2 2 2
Dealing with zeros s k (s + σk ) m (s + 2αm s + (αm + ωm ))
Zeros in first column
Zero rows

Using Routh
Hurwitz for Poles: 0, −σk , −αm ± jωm
design
Turning control of a
tracked vehicle
Assuming N = 0 and no repeated roots, the impulse response is
zero for t < 0 and for t ≥ 0 it is
X X
g(t) = Ak e−σk t + Bm e−αm t sin(ωm t + θm )
k m
R∞
Stability requires −∞ |g(t)| dt to be bounded;
that requires σk > 0, αm > 0
In fact, system is stable iff poles have negative real parts
 EE 3CL4, §4
7 / 55
Tim Davidson Marginal stability
Stability
Condition in terms of
poles • Consider integrator: G(s) = 1/s; simple pole at origin
Condition in terms of

• y (t) = t r (λ) dλ
denominator
R
coefficients
−∞
Routh Hurwitz • if r (t) = cos(t), which is bounded,
condition
Basics then y (t) = sin(t). Bounded
Disk drive example
Dealing with zeros
• If r (t) = u(t), which is bounded,
Zeros in first column then y (t) = t. Not bounded
Zero rows

Using Routh
Hurwitz for • Consider G(s) = 1/(s2 + 1), simple poles at s = ±j1
design
Turning control of a • Unit step response: u(t) − cos(t). Bounded
tracked vehicle
• What if r (t) is a sinusoid of frequency 1/(2π) Hz?
Not bounded

If G(s) has a pole with positive real part,

or a repeated pole on jω-axis
output is always unbounded
 EE 3CL4, §4
8 / 55
Tim Davidson Routh-Hurwitz condition
Stability
Condition in terms of
poles
Condition in terms of
denominator
coefficients

Routh Hurwitz
condition We have seen how to determine stability from the poles.
Basics
Disk drive example
Dealing with zeros Much easier than havingR to find impulse response
Zeros in first column ∞
Zero rows and then determining if −∞ |g(τ )| dτ < ∞
Using Routh
Hurwitz for Can we determine stability without having to determine the
design
Turning control of a
tracked vehicle
poles?
Yes! Routh-Hurwitz condition
 EE 3CL4, §4
9 / 55
Tim Davidson Routh-Hurwitz condition
Stability p(s)
Condition in terms of
Let G(s) = q(s) , where
poles
Condition in terms of
denominator
coefficients q(s) = an sn + an−1 sn−1 + . . . a1 s + a0
Routh Hurwitz
condition
= an (s − r1 )(s − r2 ) . . . (s − rn )
Basics
Disk drive example
Dealing with zeros
where ri are the roots of q(s) = 0.
Zeros in first column
Zero rows By multiplying out, q(s) = 0 can be written as
Using Routh
Hurwitz for
design
q(s) = an sn − an (r1 + r2 + · · · + rn )sn−1
+ an (r1 r2 + r2 r3 + . . . )sn−2
Turning control of a
tracked vehicle

− an (r1 r2 r3 + r1 r2 r4 + . . . )sn−3
+ · · · + (−1)n an (r1 r2 r3 . . . rn ) = 0

If all ri are real and in left half plane, what is sign of coeffs of sk ?
the same!
 EE 3CL4, §4
10 / 55
Tim Davidson Routh-Hurwitz condition
Stability
Condition in terms of
poles
Condition in terms of
denominator
coefficients

Routh Hurwitz
condition
Basics That observation leads to a necessary condition.
Disk drive example
Dealing with zeros
Zeros in first column
Hence, not that useful for design
Zero rows

Using Routh A more sophisticated analysis leads to the Routh-Hurwitz

Hurwitz for
design condition, which is necessary and sufficient
Turning control of a
tracked vehicle
Hence, can be quite useful for design
 EE 3CL4, §4
11 / 55
Tim Davidson R-H cond: A first look
p(s)
Consider G(s) = q(s) . Poles are solutions to q(s) = 0; i.e.,
Stability
Condition in terms of
poles
Condition in terms of
an sn + an−1 sn−1 + an−2 sn−2 + · · · + a1 s + a0 = 0
denominator
coefficients

Routh Hurwitz
condition Construct a table of the form
Basics
Disk drive example
Dealing with zeros
Row n an an−2 an−4 ...
Zeros in first column Row n − 1 an−1 an−3 an−5 ...
Zero rows
Row n − 2 bn−1 bn−3 bn−5 ...
Using Routh
Hurwitz for Row n − 3 cn−1 cn−3 cn−5 ...
design .. .. .. ..
Turning control of a
tracked vehicle . . . . ...
Row 0 hn−1
where
 
an−1 an−2 − an an−3 −1  an an−2 
bn−1 = = 
an−1 an−1  an−1 an−3 
   
−1  an an−4  −1  an−1 an−3 
bn−3 = cn−1 =
an−1  an−1 an−5  bn−1  bn−1 bn−3 
 EE 3CL4, §4
12 / 55
Tim Davidson R-H cond: A first look
Stability
Condition in terms of
poles
Now consider the table that we have just constructed
Condition in terms of
denominator
coefficients Row n an an−2 an−4 ...
Routh Hurwitz Row n − 1 an−1 an−3 an−5 ...
condition
Basics
Row n − 2 bn−1 bn−3 bn−5 ...
Disk drive example Row n − 3 cn−1 cn−3 cn−5 ...
Dealing with zeros
Zeros in first column .. .. .. ..
Zero rows . . . . ...
Using Routh Row 0 hn−1
Hurwitz for
design
Turning control of a
tracked vehicle
Loosely speaking:
• Number of roots in the right half plane is equal to the number
of sign changes in the first column of the table
• Stability iff no sign changes in the first column

Now let’s move towards a more sophisticated statement

 EE 3CL4, §4
14 / 55
Tim Davidson Stability (Revision)
Stability
Condition in terms of
poles
Condition in terms of
denominator
coefficients

Routh Hurwitz
condition
Basics p(s)
Disk drive example Let G(s) = q(s) , where
Dealing with zeros
Zeros in first column
Zero rows
q(s) = an sn + an−1 sn−1 + . . . a1 s + a0
Using Routh
Hurwitz for
design
Turning control of a
tracked vehicle System is stable iff all poles of G(s) have negative real parts

Recall, poles are solutions to q(s) = 0

Can we find a necessary and sufficient condition that depends

only on the set of coefficients {ak }nk =0 so that we don’t have to
solve q(s) = 0?
 EE 3CL4, §4
15 / 55
Tim Davidson Routh-Hurwitz condition
Stability 1 Consider q(s) with an > 0
Condition in terms of
poles
Condition in terms of
an sn + an−1 sn−1 + an−2 sn−2 + . . . a1 s + a0 = 0
denominator
coefficients

Routh Hurwitz
condition
Basics
2 Construct a table of the form
Disk drive example
Dealing with zeros Row n an an−2 an−4 ...
Zeros in first column
Zero rows
Row n − 1 an−1 an−3 an−5 ...
Using Routh
Row n − 2 bn−1 bn−3 bn−5 ...
Hurwitz for Row n − 3 cn−1 cn−3 cn−5 ...
design
Turning control of a .. .. .. ..
tracked vehicle . . . . ...
Row 0 hn−1
Procedure provided on the following slides

3 Count the sign changes in the first column

4 That is the number of roots in the right half plane

Stability (poles in LHP) iff all terms in first col. have same sign
 EE 3CL4, §4
16 / 55
Tim Davidson Constructing RH table
Stability
Condition in terms of
poles
Condition in terms of
denominator
coefficients

Routh Hurwitz
condition
Basics
Disk drive example
an sn + an−1 sn−1 + an−2 sn−2 + . . . a1 s + a0 = 0
Dealing with zeros
Zeros in first column
Zero rows

Using Routh
Hurwitz for
Step 2.1: Arrange coefficients of q(s) in first two rows
design
Turning control of a
tracked vehicle
Row n an an−2 an−4 ...
Row n − 1 an−1 an−3 an−5 ...
 EE 3CL4, §4
17 / 55
Tim Davidson Interlude
Stability
Condition in terms of
poles
Condition in terms of
denominator
coefficients

Routh Hurwitz
condition
Basics
Disk drive example
Dealing with zeros
Determinant of a 2 × 2 matrix:
Zeros in first column
Zero rows
 
 a b 
Using Routh
 c d  = ad − cb
 
Hurwitz for
design
Turning control of a
tracked vehicle
 EE 3CL4, §4
18 / 55
Tim Davidson Constructing RH table
Stability
Condition in terms of
poles
Condition in terms of
denominator
coefficients
Step 2.2: Construct 3rd row using determinants of 2 × 2
Routh Hurwitz
condition matrices constructed from rows above
Basics
Disk drive example
Dealing with zeros
Zeros in first column
Row n an an−2 an−4 ...
Zero rows Row n − 1 an−1 an−3 an−5 ...
Using Routh Row n − 2 bn−1
Hurwitz for
design
Turning control of a
tracked vehicle  
−1  an an−2 
bn−1 =
an−1  an−1 an−3 
 EE 3CL4, §4
19 / 55
Tim Davidson Constructing RH table
Stability
Condition in terms of
poles
Condition in terms of
denominator
coefficients
Step 2.2, cont: Construct 3rd row using determinants of
Routh Hurwitz
condition 2 × 2 matrices constructed from rows above
Basics
Disk drive example
Dealing with zeros
Zeros in first column
Row n an an−2 an−4 ...
Zero rows Row n − 1 an−1 an−3 an−5 ...
Using Routh Row n − 2 bn−1 bn−3 ...
Hurwitz for
design
Turning control of a
tracked vehicle  
−1  an an−4 
bn−3 =
an−1  an−1 an−5 
 EE 3CL4, §4
20 / 55
Tim Davidson Constructing RH table
Stability
Condition in terms of
Step 2.3: Construct 4th row using determinants of 2 × 2
poles
Condition in terms of matrices constructed from rows above
denominator
coefficients

Routh Hurwitz Row n an an−2 an−4 ...

condition
Basics Row n − 1 an−1 an−3 an−5 ...
Disk drive example
Dealing with zeros
Row n − 2 bn−1 bn−3 ...
Zeros in first column
Zero rows
Row n − 3 cn−1 ...
Using Routh
Hurwitz for  
design −1  an−1 an−3 
Turning control of a
cn−1 =
tracked vehicle
bn−1  bn−1 bn−3 

Step 2.4: Continue in this pattern.

Caveat: Requires all elements of first column to be non-zero
Will come back to that. Let’s see some examples, first
 EE 3CL4, §4
21 / 55
Tim Davidson RH table, second-order system
Stability
Condition in terms of
poles
Condition in terms of
denominator
coefficients
q(s) = a2 s2 + a1 s + a0
Routh Hurwitz
condition
Basics
Disk drive example
Dealing with zeros Row 2 a2 a0
Zeros in first column
Zero rows Row 1 a1 0
Using Routh Row 0 b1
Hurwitz for
design  
Turning control of a −1  a2 a0 
tracked vehicle b1 =  = a0
a1  a1 0 

Therefore, second order system is stable iff all three denominator

coefficients have the same sign
 EE 3CL4, §4
22 / 55
Tim Davidson RH table, third order system
Stability
Condition in terms of
poles
Condition in terms of q(s) = a3 s3 + a2 s2 + a1 s + a0
denominator
coefficients

Routh Hurwitz
condition
Basics Row 3 a3 a1
Disk drive example
Dealing with zeros Row 2 a2 a0
Zeros in first column
Zero rows
Row 1 b1 0
Using Routh
Row 0 c1 0
Hurwitz for
design
Turning control of a    
tracked vehicle
−1  a3 a1  −1  a2 a0 
b1 = c1 = = a0
a2  a2 a0  b1  b1 0 

Therefore, if a3 > 0, necessary and sufficient condition for

third-order system to be stable is that a2 > 0, b1 > 0 and a0 > 0.
b1 > 0 is equiv. to a2 a1 > a0 a3 , and this implies a1 > 0.
 EE 3CL4, §4
23 / 55
Tim Davidson Disk drive read control
Add velocity feedback (switch closed)
Stability
Condition in terms of
poles
Condition in terms of
denominator
coefficients

Routh Hurwitz
condition
Basics
Disk drive example
Dealing with zeros
Zeros in first column
Zero rows

Using Routh
Hurwitz for Using block diagram manipulation
design
Turning control of a
tracked vehicle

5000 1
G1 (s) = G2 (s) =
s + 1000 s(s + 20)
 EE 3CL4, §4
24 / 55
Tim Davidson Closed loop
Stability
Condition in terms of
poles
Condition in terms of
denominator
coefficients

Routh Hurwitz
condition
Basics
Disk drive example
Dealing with zeros
Zeros in first column
Zero rows

Using Routh
Hurwitz for
design
Turning control of a Y (s) Ka G1 (s)G2 (s)
tracked vehicle T (s) = =
R(s) 1 + Ka G1 (s)G2 (s)(1 + K1 s)

Hence, char. eqn:

s3 + 1020s2 + (20000 + 5000Ka K1 )s + 5000Ka = 0
 EE 3CL4, §4
25 / 55
Tim Davidson Stabilizing values of K1 and Ka
s3+ 1020s2 + (20000 + 5000Ka K1 )s + 5000Ka = 0
Stability
Condition in terms of
poles Routh table
Condition in terms of
denominator
coefficients
Row 3 1 20000 + 5000Ka K1
Routh Hurwitz
Row 2 1020 5000Ka
condition Row 1 b1
Basics
Disk drive example
Row 0 5000Ka
Dealing with zeros
Zeros in first column 1020(20000 + 5000Ka K1 ) − 5000Ka
Zero rows b1 =
Using Routh
1020
Hurwitz for
design
Turning control of a For stability we require b1 > 0 and Ka > 0. That is,
tracked vehicle

Ka − 4080
Ka > 0 and K1 >
1020Ka

Note that for 0 < Ka < 4080, any positive K1 will stabilize
the loop, and some negative ones will, too.
For example, Ka = 100 and K1 = 0.05.
That pair gives a 2% settling time of 260ms
 EE 3CL4, §4
26 / 55
Tim Davidson Reminder: Construction
Stability procedure
Condition in terms of
poles
Condition in terms of
Row k + 2 p1 p3 p5 ...
denominator
coefficients Row k + 1 q1 q3 q5 ...
Routh Hurwitz Row k r1 r3 r5 ...
condition
Basics
Disk drive example To compute r3 , multiply
Dealing with zeros
Zeros in first column • first element −1 −1
of previous row = q1 by
Zero rows

Using Routh • determinant of 2 × 2 matrix formed in the following way:

Hurwitz for
design • The first column contains the first elements of the two
Turning control of a
tracked vehicle rows above the element to be calculated
• The second column contains the elements of the two
rows above that lie one column to the right of the
element to be calculated
• Therefore
 
−1  p1 p5  −1

r3 = = p1 q5 − q1 p5
q1 q1 q5
  q1
 EE 3CL4, §4
27 / 55
Tim Davidson RH table, dealing with zeros
Stability
Condition in terms of
poles
Condition in terms of
denominator
coefficients • The Routh-Hurwitz table encounters trouble when there
Routh Hurwitz
condition
is a zero in the first column
Basics
Disk drive example
Dealing with zeros
• The next row involves (−1/0) times a determinant
Zeros in first column
Zero rows
• When some other elements in that row are not zero, we
Using Routh
Hurwitz for can proceed by replacing the zero by a small positive
design
Turning control of a number , and then taking the limit as  → 0 after the
tracked vehicle
table has been constructed.
• When a whole row is zero, we need to be a bit more
sophisticated (later)
 EE 3CL4, §4
28 / 55
Tim Davidson RH table, zero first element in
Stability non-zero row
Condition in terms of
poles As an example, consider
Condition in terms of
denominator

q(s) = s5 + 2s4 + 2s3 + 4s2 + 11s + 10

coefficients

Routh Hurwitz
condition
Basics
Disk drive example
Dealing with zeros
Zeros in first column
Routh table
Zero rows

Using Routh
Row 5 1 2 11
Hurwitz for Row 4 2 4 10
design
Turning control of a Row 3 0← 6 0
tracked vehicle
Row 2 c1 10 0
Row 1 d1 0 0
Row 0 10 0 0

4 − 12 −12 6c1 − 10

c1 = = d1 = →6
  c1

Two sign changes, hence unstable with two RHP poles

 EE 3CL4, §4
29 / 55
Tim Davidson Zero row
Stability
• It is possible that the Routh Hurwitz procedure can
Condition in terms of
poles produce a zero row
Condition in terms of
denominator
coefficients
• While this complicates the procedure, it yields useful
Routh Hurwitz
condition information for design
Basics
Disk drive example
Dealing with zeros • Zero rows occur when polynomial has roots that are
Zeros in first column
Zero rows radially symmetric.
Using Routh
Hurwitz for
design
• Since the roots must also occur in conjugate pairs, this
Turning control of a
tracked vehicle
means that there is at least one pair of roots that is
symmetric in the imaginary axis.
(All roots are symmetric in the real axis.)
 EE 3CL4, §4
30 / 55
Tim Davidson Zero row
Stability
Condition in terms of
poles
Condition in terms of
denominator
coefficients
• Common examples include:
Routh Hurwitz
• equal and opposite roots on the real axis,
condition • a pair of complex conjugate roots on the imaginary axis.
Basics
Disk drive example
Dealing with zeros
• The latter is more common, and more useful in design
Zeros in first column
Zero rows

Using Routh
Hurwitz for
design
Turning control of a
tracked vehicle

• So how can we deal with the zero row?

 EE 3CL4, §4
31 / 55
Tim Davidson Dealing with a zero row
Stability
• Routh Hurwitz procedure provides an “auxiliary polynomial”,
Condition in terms of
poles
a(s), that contains the roots of interest as factors
Condition in terms of
denominator
coefficients • The auxiliary polynomial is a factor of the original polynomial;
Routh Hurwitz i.e., q(s) = a(s)b(s); b(s) can be found by polyn. division
condition
Basics
Disk drive example • Given the symmetry of the roots, the auxiliary polynomial is
Dealing with zeros
Zeros in first column
of even order (or will have all its roots at the origin)
Zero rows

Using Routh • The coefficients of the auxiliary polynomial appear in the row
Hurwitz for
design above the zero row
Turning control of a
tracked vehicle
• Let k denote the row number of the row above the zero row,
and let ck ,1 , ck ,2 , ck ,3 denote the coefficients in that row. The
auxiliary polynomial is constructed as
a(s) = ck ,1 sk + ck ,2 sk −2 + ck ,3 sk −4 + . . .
• Finally, we replace the zero row by the coefficients of the
derivative of the auxiliary polynomial; i.e., the zero row is
replaced by kck ,1 , (k − 2)ck ,2 , (k − 4)ck ,3
 EE 3CL4, §4
32 / 55
Tim Davidson Zero row example
Stability • q(s) = s5 + 2s4 + 24s3 + 48s2 − 25s − 50 = 0
Condition in terms of
poles
Condition in terms of
denominator
• Construct table
coefficients
Row 5 1 24 −25
Routh Hurwitz
condition Row 4 2 48 −50
Basics
Disk drive example Row 3 0 0
Dealing with zeros
Zeros in first column
Zero rows
• Auxiliary polynomial: a(s) = 2s4 + 48s2 − 50
Using Routh
Hurwitz for This is actually a factor of q(s).
design
Turning control of a
tracked vehicle
Using quadratic formula, roots of a(s) are s2 = 1, −25
Hence roots of a(s) are s = ±1, ±j5
da(s)
• ds = 8s3 + 96s.
Replace zero row by these coefficients
Row 5 1 24 −25
Row 4 2 48 −50
Row 3 8 96
 EE 3CL4, §4
33 / 55
Tim Davidson Zero row example, cont
Stability • Now complete the table in the usual way
Condition in terms of
poles
Condition in terms of
Row 5 1 24 −25
denominator
coefficients Row 4 2 48 −50
Routh Hurwitz
condition
Row 3 8 96
Basics
Disk drive example
Row 2 24 −50
Dealing with zeros
Zeros in first column
Row 1 112.7 0
Zero rows Row 0 −50
Using Routh
Hurwitz for
design • One sign change in first column.
Turning control of a
tracked vehicle Indicates one root in right half plane.
• Recall a(s) is a factor of q(s).
Indeed, by polyn division q(s) = (s + 2)a(s)
We have seen that roots of a(s) are ±1 and ±j5.
• Hence q(s) does indeed have one root with a positive
real part.
 EE 3CL4, §4
35 / 55
Tim Davidson Turning control of a tracked
Stability vehicle
Condition in terms of
poles
Condition in terms of
denominator
coefficients

Routh Hurwitz
condition
Basics
Disk drive example
Dealing with zeros
Zeros in first column
Zero rows

Using Routh
Hurwitz for
design
Turning control of a
tracked vehicle

Select K and a so that

• the closed-loop is stable, and
• the steady-state error due to a ramp is at most 24% of
the magnitude of the command
 EE 3CL4, §4
36 / 55
Tim Davidson Deal with stability first
Stability
Condition in terms of Gc (s)G(s)
poles Transfer function: T (s) = 1+Gc (s)G(s)
Condition in terms of
denominator

Char. equation: s4 + 8s3 + 17s2 + (K + 10)s + Ka = 0

coefficients

Routh Hurwitz
condition
Basics Routh table
Disk drive example
Dealing with zeros
Zeros in first column
Row 4 1 17 Ka
Zero rows Row 3 8 K + 10
Using Routh Row 2 b3 Ka
Hurwitz for
design Row 1 c3
Turning control of a
tracked vehicle Row 0 Ka

126 − K b3 (K + 10) − 8Ka

b3 = c3 =
8 b3

For stability we require b3 > 0, c3 > 0 and Ka > 0

 EE 3CL4, §4
37 / 55
Tim Davidson Stability region
For positive K , these constraints can be rewritten as
Stability
Condition in terms of
poles K < 126
Condition in terms of
denominator
coefficients a>0
Routh Hurwitz
condition (K + 10)(126 − K )
Basics
a<
Disk drive example 64K
Dealing with zeros
Zeros in first column
Zero rows
Region of stable parameters is between blue curves
Using Routh
Hurwitz for
design
Turning control of a
tracked vehicle
 EE 3CL4, §4
38 / 55
Tim Davidson Steady-state error to ramp
Stability
Condition in terms of
• For a ramp input r (t) = At, we have ess = A
Kv , where
poles
Condition in terms of
denominator
coefficients
Ka
Routh Hurwitz Kv = lim sGc (s)G(s) = 10
condition s→0
Basics
Disk drive example
Dealing with zeros
Zeros in first column
• Therefore, ess = A 10A
Zero rows Kv = Ka
Using Routh
Hurwitz for
• To obtain ess < 0.24A ' A
4.167 , we need Kv > 4.167
design
Turning control of a
tracked vehicle

• That means we need Ka > 41.67.

• Any (K , a) pair in stable region with a > 41.67

K will satisfy
design constraints
 EE 3CL4, §4
39 / 55
Tim Davidson Set of parameters with desired
Stability performance
Condition in terms of
poles
Condition in terms of
denominator
coefficients

Routh Hurwitz
condition
Basics
Disk drive example
Dealing with zeros
Zeros in first column
Zero rows

Using Routh
Hurwitz for
design
Turning control of a
tracked vehicle

For positive K ,
• stability region is between the blue solid curves
• desired steady-state error region is above the black
dashed curve and between the blue solid curves
• Design example: K = 70, a = 0.6, marked by + sign
 EE 3CL4, §4
40 / 55
Tim Davidson Ramp response
Stability
Condition in terms of
poles
Condition in terms of
denominator
coefficients

Routh Hurwitz
condition
Basics
Disk drive example
Dealing with zeros
Zeros in first column
Zero rows

Using Routh
Hurwitz for
design
Turning control of a
tracked vehicle

• Steady-state error criterion satisfied

(Kv = 4.200 > 4.167)
• Transient time is quite long
 EE 3CL4, §4
41 / 55
Tim Davidson What about settling times?
Stability • That system takes a long time to settle.
Condition in terms of
poles
Condition in terms of • What does the step response look like?
denominator
coefficients

Routh Hurwitz
condition
Basics
Disk drive example
Dealing with zeros
Zeros in first column
Zero rows

Using Routh
Hurwitz for
design
Turning control of a
tracked vehicle

• Could we have predicted the long settling time and large overshoot?
• We know that the settling time of each component of the response
is related to the real parts of the poles
 EE 3CL4, §4
42 / 55
Tim Davidson Closed-loop poles and zeros
Stability
Condition in terms of
poles
Condition in terms of
denominator
coefficients

Routh Hurwitz
condition
Basics
Disk drive example
Dealing with zeros
Zeros in first column
Zero rows

Using Routh • One “fast” real pole

Hurwitz for
design • One “reasonably slow” real pole that is close to a zero
Turning control of a
tracked vehicle
• Complex conjugate pair with small real parts (≈ 0.17), and large
angles with negative real axis (≈ 87◦ ). Conjugate pair dominates
• 2% settling time of response to conjugate pair is 4 time constants,
1
4( 0.17 ) ≈ 23
• Damping ratio of conjugate pair ζ = cos(87◦ ) ≈ 0.053.
Corresponds to overshoot of ≈ 85% = 0.85
• Actual performance of the fourth-order system with one zero is quite
close to this guidance from second-order system with no finite zeros
 EE 3CL4, §4
43 / 55
Tim Davidson Managing settling times
Stability
• It is disappointing that using Routh-Hurwitz for design gives
Condition in terms of
poles
us control over stability, but does not allow us to manage
Condition in terms of
denominator
settling times
coefficients
• Particularly disappointing, given “shape” of region to left of a
Routh Hurwitz
condition certain real number is same as that to left of origin
Basics
Disk drive example • Can we do anything?
Dealing with zeros
Zeros in first column
Zero rows
• Recall closed-loop denominator polynomial is:
Using Routh n
Y
Hurwitz for
design q(s) = an (s − (−pcl,j ))
Turning control of a
tracked vehicle
j=1

where −pcl,j are the closed-loop poles

• If we want q(s) to have closed-loop poles to left of −σ,
then we need
Yn


q̃(s) = an s − (−pcl,j + σ)
j=1

to be stable
 EE 3CL4, §4
44 / 55
Tim Davidson Managing settling times, II
Stability • That means we want
Condition in terms of
poles
Condition in terms of
denominator
q̃(s) = q(s − σ)
coefficients

Routh Hurwitz
= an (s − σ)n + an−1 (s − σ)n−1 +
condition
Basics
· · · + a1 (s − σ) + a0
Disk drive example
Dealing with zeros
Zeros in first column
to be stable
Zero rows

Using Routh
• Using binomial theorem, (x + y )n = nk =0 kn x n−k y k , we can
P

Hurwitz for
design
Turning control of a write
q̃(s) = ãn sn + ãn−1 sn−1 + · · · + ã1 s + ã0
tracked vehicle

where kn = (n−k n!


)!k !

• Now you can apply Routh-Hurwitz procedure to q̃(s)

• What happens if there is a pair of closed-loop poles with real

parts equal to −σ? A zero row in the table
 EE 3CL4, §4
45 / 55
Tim Davidson Application to tracked vehicle
Stability • Settling time of component due to complex-conjugate pair in current
Condition in terms of
poles
design is around 23 seconds
Condition in terms of
denominator • Can we reduce this? Let’s try to get it down to around 16 seconds.
coefficients

Routh Hurwitz • That means that we want poles to the left of −σ = −0.25
condition
Basics
Disk drive example
• q̃(s) ' s4 + 7s3 + 11.375s2 + (K + 2.9375)s + ã0 (K , a),
Dealing with zeros
Zeros in first column
where ã0 (K , a) = Ka − 0.25K − 1.3086
Zero rows

Using Routh
• Routh table
Hurwitz for
design Row 4 1 11.375 ã0 (K , a)
Turning control of a
tracked vehicle
Row 3 7 K + 2.9375
Row 2 b̃3 ã0 (K , a)
Row 1 c̃3
Row 0 ã0 (K , a)

76.6876 − K b̃3 (K + 2.9375) − 7ã0 (K , a)

b̃3 = , c̃3 =
7 b̃3

• Hence, we require b̃3 > 0, c̃3 > 0 and ã0 (K , a) > 0

 EE 3CL4, §4
46 / 55
Tim Davidson Settling time region
• For positive K , the constraints can be rewritten as
Stability
Condition in terms of
poles
K < 76.6876
Condition in terms of 1.3086
denominator a > 0.25 + K
coefficients

Routh Hurwitz (76.6896 − K )(K + 2.9375) + 12.25K + 64.12

condition
a<
49K
Basics
Disk drive example • Region between the red lines (inside blue region)
Dealing with zeros
Zeros in first column • For steady state error we need a > 41.67
K
; above dashed line
Zero rows

Using Routh
Hurwitz for
design
Turning control of a
tracked vehicle

• Not much choice this time. Example: K = 47, a = 0.9, marked by ∗

 EE 3CL4, §4
47 / 55
Tim Davidson Closed-loop poles and zeros
Stability
Condition in terms of
poles
Condition in terms of
denominator
coefficients

Routh Hurwitz
condition
Basics
Disk drive example
Dealing with zeros
Zeros in first column
Zero rows

Using Routh
Hurwitz for
design
Turning control of a
tracked vehicle

Since complex conjugate pair of poles still dominates, this suggests

new design (red) will have
• Somewhat reduced settling time (poles are to left of 0.25)
• Slightly reduced overshoot in step response (angle slightly reduced)
 EE 3CL4, §4
48 / 55
Tim Davidson Step response
Stability
Condition in terms of
poles
Condition in terms of
denominator
coefficients

Routh Hurwitz
condition
Basics
Disk drive example
Dealing with zeros
Zeros in first column
Zero rows

Using Routh
Hurwitz for
design
Turning control of a
tracked vehicle

• Settling time has come down; not far from 16 seconds

• Overshoot has been reduced a little bit
 EE 3CL4, §4
49 / 55
Tim Davidson Ramp response
Stability
Condition in terms of
poles
Condition in terms of
denominator
coefficients

Routh Hurwitz
condition
Basics
Disk drive example
Dealing with zeros
Zeros in first column
Zero rows

Using Routh
Hurwitz for
design
Turning control of a
tracked vehicle

• Steady-state error criterion still satisfied

(Kv = 4.230 > 4.167)
• Transient time has been reduced
 EE 3CL4, §4
50 / 55
Tim Davidson Ramp response, zoomed in
Stability
Condition in terms of
poles
Condition in terms of
denominator
coefficients

Routh Hurwitz
condition
Basics
Disk drive example
Dealing with zeros
Zeros in first column
Zero rows

Using Routh
Hurwitz for
design
Turning control of a
tracked vehicle
 EE 3CL4, §4
51 / 55
Tim Davidson Discussion
• Transient times have been reduced, but only a little bit
Stability
Condition in terms of • Unfortunately, not much room to do any better
poles
Condition in terms of
denominator
coefficients

Routh Hurwitz
condition
Basics
Disk drive example
Dealing with zeros
Zeros in first column
Zero rows

Using Routh
Hurwitz for
design
Turning control of a
tracked vehicle

• Current compensator is Gc (s) = s+a

s+1
and gain is K
• To get significantly better performance, we need to be able to adjust
the pole of the compensator, as well as the zero and the gain
(s+a)
• For example, consider Gc (s) = (s+b) = s+1.9
s+6
with K = 133.
• Note Kv = lims→0 sGc (s)G(s) ' 4.212 > 4.167,
so ramp steady-state error criterion still satisfied
 EE 3CL4, §4
52 / 55
Tim Davidson Closed-loop poles and zeros
Stability
Condition in terms of
poles
Condition in terms of
denominator
coefficients

Routh Hurwitz
condition
Basics
Disk drive example
Dealing with zeros
Zeros in first column
Zero rows

Using Routh
Hurwitz for
design
Turning control of a
tracked vehicle

Since complex conjugate pair of poles still dominates, this suggests

new design (green) will have
• Significantly reduced settling time (dominant poles further to left)
• Somewhat reduced overshoot in step response (smaller angle)
 EE 3CL4, §4
53 / 55
Tim Davidson Step response
Stability
Condition in terms of
poles
Condition in terms of
denominator
coefficients

Routh Hurwitz
condition
Basics
Disk drive example
Dealing with zeros
Zeros in first column
Zero rows

Using Routh
Hurwitz for
design
Turning control of a
tracked vehicle

• Insight from pole and zero positions reasonably

accurate
 EE 3CL4, §4
54 / 55
Tim Davidson Ramp response, zoomed in
Stability
Condition in terms of
poles
Condition in terms of
denominator
coefficients

Routh Hurwitz
condition
Basics
Disk drive example
Dealing with zeros
Zeros in first column
Zero rows

Using Routh
Hurwitz for
design
Turning control of a
tracked vehicle

• Steady-state error criterion still satisfied for new design

(Kv = 4.212 > 4.167)
• Transient time reduced
 EE 3CL4, §4
55 / 55
Tim Davidson Discussion, cont.
Stability
Condition in terms of
poles
Condition in terms of
• New design does indeed do better in our chosen metrics
denominator
coefficients

Routh Hurwitz • However, design problem now has three parameters, not
condition
Basics
two; zero pos’n (−a), gain (K ), and also pole pos’n (−b)
Disk drive example
Dealing with zeros • Hence desired parameter region is a volume, not an area
Zeros in first column
Zero rows
• Routh-Hurwitz procedure and resulting equations for
Using Routh
Hurwitz for boundaries of desired parameter space get even more
design
Turning control of a
complicated
tracked vehicle

• Is there an easier way to gain insight into how to choose the

pole and zero positions, and to choose the gain?

• One option is the root locus procedure that we will develop in

the next section