m1 0 0 z1 (k1 k 2 ) l2 k 2 k2 z1 F1 Fc
0 J1 0 z2 l2 k 2 l22 k 2 l2 k 2 z2 T2 Fc l1 (2.7)
0 0 m2 z3 k2 l2 k 2 k 2 z3 0 0
F1 Fc (2.8)
T2 Fc l1 (2.9)
2.3 Single Degree of Freedom (sdof) System Transfer Function
and Frequency Response
2.3.1 sdof System Definition, Equations of Motion
The sdof system to be analyzed is shown below. The system consists of a
mass, m, connected to ground by a spring of stiffness k and a damper with
viscous damping coefficient c. Since the mass can only move in the z
direction, a single degree of freedom is sufficient to define the system
configuration. Force F is applied to the mass.
z F
k
m
c
Figure 2.5: Single degree of freedom system.
The equation of motion for this system is given by:
mz cz kz F (2.10)
2.3.2 Transfer Function
Taking the Laplace transform of a general second order differential equation
(DE) with initial conditions is:
Second Order DE: L z(t) s 2 z(s) sz(0) z(0) , (2.11)
where z(0) and z(0) are position and velocity initial conditions, respectively,
and z(s) is the Laplace transform of z(t). See Appendix 2 for more on Laplace
transforms.
Because we are taking a transfer function, representing the steady state
response of the system to a sinusoidal input, initial conditions are set to zero,
leaving
L z(t) s 2 z(s) (2.12)
The Laplace transform of the sdof equation of motion (2.10), where F(s)
represents the Laplace transform of F, is:
ms 2 z(s) csz(s) kz(s) F(s) (2.13)
Solving for the transfer function:
z(s) 1 1/ m
(2.14)
F(s) 2
ms cs k 2 c k
s s
m m
We can simplify the equation above by applying the following definitions:
2 k
1) n , where n is the undamped natural frequency,
m
rad/sec
2) c cr 2 km , where c cr is the critical damping value
3) is the amount of proportional damping, typically
stated as a percentage of critical damping
4) 2 n is the multiplier of the velocity term, z ,
developed below:
c
2 n
m
c k
2
ccr m
(2.15)
2c k
2 km m
c
m
Rewriting, using the above substitutions:
z(s) 1/ m
2 2
(2.16)
F(s) s 2 ns n
2.3.3 Frequency Response
Substituting j for s to calculate the frequency response, where j is the
imaginary operator:
z( j ) 1/ m
2 2
F( j ) (j ) 2 n(j ) n
1/ m
2 2
2 nj n
1/(m 2 )
2
(2.17a,b,c,d,e)
2 nj
1 n
2
2
1/(m )
2
2 j
n
2
1 n
2
1/(m )
2
n
1 j2 n
The frequency response equation above shows how the ratio (z/F) varies as a
function of frequency, . The ratio is a complex number that has some
interesting properties at different values of the ratio n / .
2 2
At low frequencies relative to the resonant frequency, n n , and
the transfer function is given by:
z( j ) 1/ m
2 2
F( j ) 2 nj n
1/ m 1 1 1 (2.18)
2
m 2n k k
n
m
m
Since the frequency response value at any frequency is a complex number, we
can take the magnitude and phase.
z( j ) 1
F( j ) k
(2.19a,b)
z( j )
0
F( j )
Thus, the gain at low frequencies is a constant, (1/k) or the inverse of the
stiffness. Phase is 0 because the sign is positive.
2 2
At high frequencies, n n , the transfer function is given by:
z( j ) 1/ m
2 2
F( j ) 2 nj n
(2.20)
1/ m 1
2 2
m
Once again, taking the magnitude and phase:
z( j ) 1 1
2
F( j ) m m 2
(2.21a,b)
z( j )
180
F( j )
2
At high frequencies, the gain is given by 1/(m ) and the phase is 180
because the sign is negative.
At resonance, n , the transfer function is given by:
z( j ) 1/ m
2 2
(2.22)
F( j ) 2 nj n
1/ m 1/ m 1 1 1 1/ k j/ k
2 n j 2 n2 j 2 2
n mj 2 kmj 2 kj 2 j 2
m
Taking magnitude and phase at resonance:
z( j ) j/ k 1/ k
F( j ) 2 2
(2.23a,b)
z( j )
90
F( j )
The magnitude at resonance is seen to be the gain at low frequency, 1/k,
divided by 2 . Since is typically a small number, for example 1% of
critical damping or 0.01, the magnitude at resonance is seen to be amplified.
At resonance the phase angle is 90 .
SDOF frequency response magnitudes for zeta = 0.1 to 1.0 in steps of 0.1
1
10
0
10
-1
10
-2
10
-3
10
-1 0 1
10 10 10
frequency, rad/sec
Figure 2.6: sdof magnitude versus frequency for different damping ratios.
The MATLAB code sdofxfer.m, listed in the next section, is used to plot the
frequency responses from (2.17) for a range of damping values for
m = k = 1.0, shown in Figures 2.6 and 2.7. These m and k values give a n
value of 1.0 rad/sec.
Since n is 1.0 rad/sec, the resonant peak in Figure 2.6 should occur at that
frequency. The low frequency magnitude was shown above to be equal to
1/k = 1.0. The curves for all the damping values approach 1.0 ( 100 1.0 ) at
2
low frequencies. At high frequencies the magnitude is given by 1/ m ,
and since m = 1, we should have magnitude of 1/ 2 . Checking the plot
above, at a frequency of 10 rad/sec, the magnitude should be 1/100 or 0.01.
Note that the slope of the low frequency asymptote is zero, meaning it is not
changing with frequency. However, the slope of the high frequency asymptote
is 2 , meaning that for every decade increase in frequency the magnitude at
high frequency decreases by two orders of magnitude by virtue of the 2 term
in the denominator. The 2 slope on a log magnitude versus log frequency
plot comes from the following:
1 2
log high frequency log 2
log 2 log (2.24)
SDOF frequency response phases for zeta = 0.1 to 1.0 in steps of 0.1
0
-20
-40
-60
-80
-100
-120
-140
-160
-180
-1 0 1
10 10 10
frequency, rad/sec
Figure 2.7: sdof phase versus frequency for different damping ratios.
2.4 tdof Laplace Transform, Transfer Functions,
Characteristic Equation, Poles, Zeros
We now return to the original tdof model as shown in Figure 2.1. In order to
define transfer functions and understand poles and zeros of the system, we
need to transform from the time domain to the frequency domain. We do this
by taking Laplace transforms of the equations of motion.
2.4.1 Laplace Transforms with Zero Initial Conditions
Repeating (2.5) for the tdof system:
m1 0 0 z1 c1 c1 0 z1
0 m2 0 z2 c1 (c1 c 2 ) c 2 z2
0 0 m3 z3 0 c2 c2 z3
(2.25)
k1 k1 0 z1 F1
k1 (k1 k 2 ) k 2 z2 F2
0 k2 k2 z3 F3
Taking Laplace transforms assuming initial conditions of zero, where
z1 , z 2, z 3 now represent the Laplace transforms of the original z1 , z 2, z 3 :
m1 0 0 s 2 z1 c1 c1 0 sz1
0 m2 0 s2 z 2 c1 (c1 c 2 ) c 2 sz 2
0 0 m3 s 2 z3 0 c2 c2 sz 3
(2.26)
k1 k1 0 z1 F1
k1 (k1 k 2 ) k 2 z2 F2
0 k2 k2 z3 F3
Rearranging:
(m1s2 c1s k1 ) ( c1s k1 ) 0 z1 F1
2
( c1s k1 ) (m 2 s c1s c 2 s k1 k 2 ) ( c2s k 2 ) z2 F2
0 ( c2s k 2 ) (m 3s 2 c2 s k 2 ) z3 F3
(2.27)
2.4.2 Solving for Transfer Functions
In this section we solve for the nine possible transfer functions for all
combinations of degrees of freedom where force is applied and where
displacements are taken. Solving for the transfer functions for greater than a
2dof system is a task not to be taken lightly symbolic algebra programs such
as Mathematica, Maple or the MATLAB Symbolic Toolbox should be used.
z1 z1 z1
F1 F2 F3
z2 z2 z2
F1 F2 F3
z3 z3 z3
F1 F2 F3
Table 2.3: Nine possible transfer functions for tdof system.
The results below were obtained by use of a symbolic algebra program.
z1 s 4 m 2 m3 s3 m3 c1 m3 c2 m 2 c2
/ Den (2.28)
F1 s 2 c1c 2 m2k2 m 3 k1 m 3 k 2 s c1k 2 c 2 k1 k1k 2
z1
s3 m 3 c1 s 2 c1c2 m3 k1 s c1k 2 k1 c 2 k1k 2 / Den (2.29)
F2
z1
s 2 c1c 2 s c1k 2 c 2 k1 k1 k 2 / Den (2.30)
F3
z2
s 3 m3 c1 s 2 c1c 2 m 3 k1 s c1k 2 c 2 k1 k1 k 2 / Den (2.31)
F1
s 4 m1m3 s3 m1c 2 m 3 c1
z2
s 2 m1 k 2 c1c2 m3 k1 / Den (2.32)
F2
s c1k 2 c 2 k1 k1k 2
z2
s 3 m1c 2 s 3 m1k 2 c1c 2 s c1 k 2 c 2 k1 k1k 2 / Den (2.33)
F3
z3
s2 c1c 2 s c1k 2 c 2 k1 k1 k 2 / Den (2.34)
F1
z3
s3 m1c2 s 2 m1k 2 c1c2 s c1k 2 c2 k1 k1 k 2 / Den (2.35)
F2
s 4 m1 m2 s3 m1c2 m1c1 m 2 c1
z3
s 2 m 2 k1 m1k1 m1k 2 c1c2 / Den (2.36)
F3
s c 2 k1 c1k 2 k1 k 2
Where Den is:
s 4 (m1m 2 m3 ) s 3 (m 2 m3c1 m1m3 c1 m1m 2 c 2 m1m3 c 2 )
s 2 (m1m3 k1 m1m3 k 2 m1m 2 k 2 m 2 c1c 2 m3 c1c 2 m1c1c 2
Den s2 k1m 2 m3 )
s (m 3c1k 2 m 2 c 2 k1 m1c 2 k1 m1c1k 2 m 3c 2 k1 m 2 c1k 2 )
(m1k1k 2 m 2 k1k 2 m3 k1k 2 )
(2.37)
Note that all the transfer functions have the same denominator, Den, called the
characteristic equation.
To simplify the system for hand calculations, take:
m1 m2 m3 m
c1 c2 c (2.38)
k1 k2 k
z1
z11 m2s4 3mcs 3 c 2 3mk s 2 2cks k 2 / Den1 (2.39)
F1
z1
z12 mcs3 c2 mk s 2 2cks k 2 / Den1 (2.40)
F2
z1
z13 c2s2 2cks k 2 / Den1 (2.41)
F3