Stability analysis– Bode Plot
Frequency Domain Specifications
Resonant Peak 𝑴𝒓 : This is the maximum value of the
transfer function magnitude.
• 𝑴𝒓 depends on the damping ratio ξ only and indicates the
relative stability of a stable closed loop system.
• The higher the peak, the poorer the relative stability.
𝟏
𝑴𝒓 =
𝟐ξ 1 − ξ2
Resonant Frequency 𝝎𝒓 : This is the frequency at which the
resonant peak is obtained.
𝝎𝒓 = 𝝎𝒏 𝟏 − 2ξ𝟐 Figure 1: Resonant Peak &
Resonant Frequency
Frequency Domain Specifications
Bandwidth : The range of frequencies for
which the system normalized gain is more
than – 3dB.
• Bandwidth is proportional to 𝝎𝒏
• Larger is the bandwidth faster is the
response.
• It is also an indicator of noise filtering
characteristics of the control systems.
Figure 2: Bandwidth
Frequency Domain Specifications
Cut-off rate: It is the slope of the magnitude
function at higher (above resonant) frequencies.
• It indicates the ability of the system to
distinguish between signals and noise.
• Two systems having same bandwidth can
have different cut-off rates.
Figure 3: Cut-off rate
Stability analysis using Bode Plot
𝐴
• Let us consider a general closed loop system, 𝐺𝐶𝐿 =
1+𝐴𝛽
• This formula assumes that Aβ is a positive number (because positive Aβ means
that the feedback is negative).
• What happens when Aβ is negative?? Consider the case Aβ= -1:
𝐴
𝐺𝐶𝐿 = =∞
1 + 𝐴𝛽
• In this context, a closed-loop gain of infinity corresponds to an oscillator—even
with zero input the output is saturated.
• The critical quantity in stability analysis is the loop gain.
• For stability Aβ must be less than unity at the high frequencies where phase shift
reaches 180°
Aβ 𝒂𝒕 𝒇𝟏𝟖𝟎 < 𝟏
Gain Margin & Phase Margin
• Similarly let us apply the concept discussed in previous slide for the control
system analysis
• Consider a closed loop transfer function,
𝐺(𝑠)
C s =
1 + 𝐺 𝑠 𝐻(𝑠)
𝐺(j𝜔)
C j𝜔 =
1 + 𝐺 j𝜔 𝐻(j𝜔)
• If 1 + 𝐺 j𝜔 𝐻 j𝜔 = 0, then C j𝜔 = ∞, under this condition system oscillates at
a constant.
• 𝐺 j𝜔 𝐻 j𝜔 = −1 + 𝑗 0 = 1∠−180° → condition for marginal stability.
• There will be a frequency 𝝎 = 𝝎𝒑𝒄 (phase crossover frequency)at which
∠𝐺 j𝜔 𝐻 j𝜔 = −180°
• At this frequency 𝝎𝒑𝒄 if 𝐺 𝑗𝜔 𝐻 𝑗𝜔 𝝎=𝝎𝒑𝒄 = 𝟏, then there will be oscillations
Gain Margin & Phase Margin
Conditions
• if 𝐺 𝑗𝜔 𝐻 𝑗𝜔 𝝎=𝝎𝒑𝒄 > 𝟏 → then system is unstable.
• if 𝐺 𝑗𝜔 𝐻 𝑗𝜔 𝝎=𝝎𝒑𝒄 = 𝟏 → then system is marginally stable.
• if 𝐺 𝑗𝜔 𝐻 𝑗𝜔 𝝎=𝝎𝒑𝒄 < 𝟏 → then system is stable.
Gain Margin & Phase Margin
• Gain cross over frequency 𝝎𝒈𝒄 :
frequency at which magnitude of
𝐺 𝑗𝜔 𝐻 𝑗𝜔 is unity.
• In dB , 𝐺 𝑗𝜔 𝐻 𝑗𝜔 𝝎=𝝎𝒈𝒄 = 0𝑑𝐵
• Phase cross over frequency 𝝎𝒑𝒄 :
frequency at which phase angle of
𝐺 𝑗𝜔 𝐻 𝑗𝜔 is −180°
Gain Margin & Phase Margin
• Gain Margin: Margin in gain allowable by which gain can be increased till
system reaches the verge of instability.
• Positive gain margin means ,increase ‘K’ is possible before system becomes
unstable hence system is stable.
• Negative gain margin means, ‘K’ as to be decreased to make system stable.
• Gain margin is measured at phase cross over frequency
1
𝐺𝑀 =
𝐺 𝑗𝜔 𝐻 𝑗𝜔 𝝎=𝝎𝒑𝒄
1
In dB, 𝐺𝑀 = 20 log = 0 − 20 log 𝐺 𝑗𝜔 𝐻 𝑗𝜔 𝝎=𝝎𝒑𝒄 𝑑𝐵
𝐺 𝑗𝜔 𝐻 𝑗𝜔 𝝎=𝝎𝒑𝒄
Gain Margin & Phase Margin
• Phase Margin: defined as amount of phase lag that can be introduced without
affecting the magnitude plot of 𝐺 𝑗𝜔 𝐻 𝑗𝜔 or additional phase lag that can be
introduced till the system reaches the instability.
𝑃𝑀 = ∠𝐺 𝑗𝜔 𝐻 𝑗𝜔 𝑎𝑡𝝎=𝝎𝒈𝒄 − −180°
𝑃𝑀 = 180° + ∠𝐺 𝑗𝜔 𝐻 𝑗𝜔 𝑎𝑡𝝎=𝝎𝒈𝒄
• Positive phase margin : negative angle can be introduced without making system
unstable.
• Negative phase margin: positive angle should be introduced to make system
stable.
Conclusions
• If system gain is high, gain margin will be low , system step response will have
high overshoots and long settling time.
• Very low gain results in high gain margin and phase margin but causes higher
steady state error, higher values of rise and settling time.
• So we should keep gain as high as possible to reduce steady state error within
acceptable range & to maintain adequate GM & PM
Example 1
• For a control system having
𝐾(1 + 0.5𝑠)
𝐺 𝑠 =
𝑠(1 + 2𝑠)(1 + 0.05𝑠 + 0.125𝑠 2 )
Draw Bode plot, with K=4 and find GM & PM
Example 1
• Factors
• K=4 , 20 log 4 = 12dB
1
• , pole at the origin → straight line of slope -20dB/dec
s
1 1
• , simple first order pole 𝑇1 = 2 𝑠𝑒𝑐 , 𝜔𝑐1 = = 0.5 𝑟𝑎𝑑/𝑠𝑒𝑐
1+2s 𝑇1
Straight line of 0dB till 𝜔=0.5 rad/sec and then a line of slope -20dB/dec
1
• 1 + 0.5s , simple first order zero 𝑇2 = 0.5 𝑠𝑒𝑐 , 𝜔𝑐2 = = 2 𝑟𝑎𝑑/𝑠𝑒𝑐
𝑇2
Straight line of 0dB till 𝜔=2 rad/sec and then a line of slope +20dB/dec
1 2𝜁 𝑠2
• 2 , quadratic factor compare denominator with 1 + 𝑠+ 2
1+0.05𝑠+0.125𝑠 𝜔𝑛 𝜔𝑛
𝑟𝑎𝑑
We get 𝜔𝑛 = 𝜔𝑐3 = 2.83 , so straight line of 0 dB till 𝜔𝑐3 = 2.83
𝑠𝑒𝑐
and then line of slope -40dB/dec
Example 1
• For phase plot ,
𝐾(1 + 0.5𝑗𝜔)
𝐺 𝑗𝜔 =
𝑗𝜔(1 + 2𝑗𝜔)(1 + 0.05𝑗𝜔 + 0.125(𝑗𝜔)2 )
Example 1
𝜔 1 tan−1 0.5𝜔 − tan−1 2𝜔 −1
0.05𝜔 𝜙𝑅𝑒𝑠𝑢𝑙𝑡𝑎𝑛𝑡
− tan
𝑗𝜔 1 − 0.125𝜔 2
0.2
2
2.2
2.4
2.83
27
∞
Example 2
• For a unity feedback systems
𝐾
𝐺 𝑠 =
𝑠(𝑠 + 2)(𝑠 + 10)
Determine the marginal value of K for which system will be marginally stable.
𝐾
20 𝐾′ ′ 𝐾
Solution: 𝐺 𝑠 = 𝑠 𝑠 = 𝑠 𝑠 𝑤ℎ𝑒𝑟𝑒 𝐾 =
𝑠(1+2)(1+10) 𝑠(1+2)(1+10) 20
Example 2
• Factors:
• 𝐾 ′ unknown
1
• , pole at the origin
𝑠
1 1
• Simple pole, 𝑠 , 𝑇1 = , 𝜔𝑐1 = 2 𝑟𝑎𝑑/𝑠𝑒𝑐
1+2 2
1 1
• Simple pole, 𝑠 , 𝑇2 = , 𝜔𝑐2 = 10 𝑟𝑎𝑑/𝑠𝑒𝑐
1+10 10
• Sketch the magnitude plot without considering the effect of the unknown 𝐾 ′
• Effect of 𝐾 ′ will be addition of 20 log 𝐾 ′ dB at all frequencies i.e, shifting the
magnitude plot drawn upwards or downwards without changing the slope.
Example 2
Phase plot:
𝐾 ′ is not going to contribute to phase angle table constructed.
Example 2
𝜔 1 −tan−1 0.5𝜔 − tan−1 0.1𝜔 𝜙𝑅𝑒𝑠𝑢𝑙𝑡𝑎𝑛𝑡
𝑗𝜔
0.1
10
20
∞
Example 3
𝐾(𝑠+2)
• A system has 𝐺 𝑠 𝐻(𝑠) = . Find K to get a PM =+30°
𝑠(𝑠+4)(𝑠+10)
𝐾 𝑠 𝑠
(1+ ) 𝐾′ (1+ ) 𝐾
Solution : 𝐺 𝑠 𝐻(𝑠) = 20
𝑠
2
𝑠 = 𝑠
2
𝑠 𝑤ℎ𝑒𝑟𝑒 𝐾′ =
𝑠(1+4)(1+10) 𝑠(1+4)(1+10) 20
Example 3
• Factors:
• 𝐾 ′ unknown
1
• , pole at the origin
𝑠
𝑠 1
• Simple zero, 1 + , 𝑇1 = , 𝜔𝑐1 = 2 𝑟𝑎𝑑/𝑠𝑒𝑐
2 2
1 1
• Simple pole, 𝑠 , 𝑇2 = , 𝜔𝑐2 = 4 𝑟𝑎𝑑/𝑠𝑒𝑐
1+ 4
4
1 1
• Simple pole, 𝑠 , 𝑇3 = , 𝜔𝑐3 = 10 𝑟𝑎𝑑/𝑠𝑒𝑐
1+10 10
• Sketch the magnitude plot without considering the effect of the unknown 𝐾 ′
Example 3
• Phase plot:
𝐾 ′ is not going to contribute to phase angle table constructed
Example 3
𝜔 1 −1
𝜔 −1
𝜔 −1
𝜔 𝜙
+ tan − tan − tan 𝑅𝑒𝑠𝑢𝑙𝑡𝑎𝑛𝑡
𝑗𝜔 2 4 10
0.1
1
2
4
10
20
∞
