Modeling Electrical-

Mechanical Systems
(DC Motors)
MECH 3140
Lecture #

1
Class Update
• We will start grading Exam #2 this week
– It will probably take a couple of weeks as we try
and stay on top of other on-line items
• HW #9 is out and contains the last matlab
assignment you will have to do
– There will be a HW #10 which will be problems
to prepare for the final exam
• We should be posting the final project this
week
– It will be due the last week of classes and teams
will present their power-point slides remotely
2
Multi-DOF Modeling

• We are starting to looking at modeling multi-

DOF systems
• One example will be the DC electric motor.
• This will also be our first example of
modeling a combined electrical-mechanical
system
– Note that solenoids and other systems are very
analogous

3
DC Electric Motors

• This is one of the examples where we get to

practice modeling Step #0
– We have to determine which modeling elements
to include
– Here is a picture of the outside and inside of a
typical DC motor. Which of our modeling
elements do you think we need to use?

4
“General” Fundamental Operation

• DC motors work by generating opposing

magnetic fields that produce a force (or
torque) on the shaft.
• In a permanent magnet motor, the opposing
field is generated by “pushing” current
through a set of coils in the permanent
magnetic field
– The change in “opposition” is created by
changing the set of coils that are magnetized
(powered) in the armature through the split
commutator.
5
Inputs and Outputs

• Lets start by defining the input and output

– Input could be voltage or current. We will use
voltage (but will see later why many in robotics
use a current source or current input)
– The output of the “system” is the position or
speed of the motor shaft
• We will use angular velocity as the fundamental
output

6
Step #0: Electrical Side
• Lets start with modeling the electrical side
• What components do you think we need?
– Voltage supply (since this is our input)
– Resistance
• Why? There is no resistor.
• The wire inside the motor certainly has resistance (energy loss)
that needs to be modeled
– Inductance
• Same reasons as above, it is not possible to instantly generate
current through all of the wire inside the motor.
– Voltage Output
• This is known as the back emf (electromotive force). A voltage is
generated in the coils as the travel through the magnetic field
– This is how a generator works (it’s a motor driven in reverse).
7
Step #0: Electrical Side

• So putting our components together in an

electrical schematic to model we get:
L
– 1 DOF (I)
– 1st Order
R
Vs

Vemf

– Notice this schematic is similar to a previous

homework problem where Vemf was a
disturbance voltage!
• This is where that problem came from.
8
Step #0: Mechanical Side
• Now to the mechanical side
• What components do you think we need?
– Torque on the shaft (this is what cause the motor to spin)
– Damping
• Why? The bearings are not perfect and certainly would have
some energy loss that is proportional to velocity
– Friction
• There probably is some constant friction force the motor must
overcome, but we will ignore this
– We will assume good bearings or that this force is negligible

– Inertia
• For sure the armature has inertia that has to be accounted for.
– Shaft compliance
• Possibly, but we will ignore it here (assume the armature is rigid)
9
Step #0: Mechanical Side
• So putting our components together in an
mechanical schematic to model we get:
– 1 DOF (ω)
– 1st Order Jm

τm
bearings

10
Modeling Step #1 (DOF and Order)

• How many Degrees of Freedom is the

system?
– Its 2 DOF: Current and Angular Velocity
• These are not directly relatable (if I know one, I don’t
necessarily know the other)
• This means I should expect 2 coupled differential
equations
• What is the total system order?
– Its 2nd Order
• There are two independent storage elements
– Inductor
– Armature Inertia
11
Step #2 Electrical Side
• Assigning current loop and the voltage drops:
+ L -

+
I R
+
Vs -
-
- +
Vemf

• Defining the voltage drops:

– 𝑉𝑉𝑒𝑒𝑒𝑒𝑒𝑒 = 𝐾𝐾𝑏𝑏 𝜔𝜔
• Back emf is proportional to the motor speed.
• Kb is a motor constant
– 𝑉𝑉𝑅𝑅 = 𝑅𝑅𝑅𝑅
– 𝑉𝑉𝐿𝐿 = 𝐿𝐿𝐼𝐼 ̇ 12
Step #3 Electrical Side
• Using Kirchoff’s Voltage Law:
+ L -

+
I R
+
Vs -
-
- +
Vemf

∑ 𝑉𝑉𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿 = 0
−𝑉𝑉𝑠𝑠 + 𝑉𝑉𝐿𝐿 + 𝑉𝑉𝑅𝑅 + 𝑉𝑉𝑒𝑒𝑒𝑒𝑒𝑒 = 0

𝐿𝐿𝐼𝐼 ̇ + 𝑅𝑅𝑅𝑅 = 𝑉𝑉𝑠𝑠 − 𝐾𝐾𝑏𝑏 𝜔𝜔

13
Step #2 & #3: Mechanical Side
• Assigning our coordinate (θ)
• Drawing our FBD
Jm

τm
θ
– 𝜏𝜏𝑏𝑏 = 𝑏𝑏𝜔𝜔
bearings
– 𝜏𝜏𝑚𝑚 = 𝐾𝐾𝐼𝐼 𝐼𝐼
• The motor torque is proportional to motor current
• This is why robotics use current sources
– Want to directly input motor torque
• KI is a motor constant
– KI=Kb (in SI units)
14
Step #4: Mechanical Side
• Using Newton’s Law to sum moments:

� 𝑀𝑀 = 𝐽𝐽𝜃𝜃̈ = 𝐽𝐽𝜔𝜔̇

𝐽𝐽𝜔𝜔̇ = −𝜏𝜏𝑏𝑏 + 𝜏𝜏𝑚𝑚

𝐽𝐽𝜔𝜔̇ = −𝑏𝑏𝜔𝜔 + 𝐾𝐾𝐼𝐼 𝐼𝐼

𝐽𝐽𝜔𝜔̇ + 𝑏𝑏𝜔𝜔 = 𝐾𝐾𝐼𝐼 𝐼𝐼

15
Final Equations of Motion

• We get 2 EOMs 𝐽𝐽𝜔𝜔̇ + 𝑏𝑏𝜔𝜔 = 𝐾𝐾𝐼𝐼 𝐼𝐼

– Recall it was 2 DOF 𝐿𝐿𝐼𝐼 ̇ + 𝑅𝑅𝑅𝑅 = 𝑉𝑉𝑠𝑠 − 𝐾𝐾𝑏𝑏 𝜔𝜔
• 1 EOM per DOF
• The equations are coupled
– The speed differential equation has current
– The current different equation has speed
• The system is two 1st order coupled
differential equations
– So the system order is 2nd order!

16
Now What?

• How do we solve or analyze?

– We need….the eigenvalues (of course)
– So how do we get them?
• We can’t just set the “right side” to zero
– This would set outputs to zero!
– Must use our new tools!
• Transfer Function
• State Space

17
Transfer Function
• Take the Laplace of the two LTI equations and setting
initial conditions to zero (since we are solving for the
transfer function):
𝐽𝐽𝜔𝜔 𝑠𝑠 𝑠𝑠 + 𝑏𝑏𝜔𝜔 𝑠𝑠 = 𝐾𝐾𝐼𝐼 𝐼𝐼 𝑠𝑠

𝐿𝐿𝐼𝐼 𝑠𝑠 𝑠𝑠 + 𝑅𝑅𝑅𝑅 𝑠𝑠 = 𝑉𝑉𝑠𝑠 𝑠𝑠 − 𝐾𝐾𝑏𝑏 𝜔𝜔(𝑠𝑠)

• Now What?
• We have one input, but two outputs (I, ω)
– Remember a Transfer Function is between a single input and a
single output
• We can not just set the output we don’t want to zero – This
would change the system!
– You can set additional inputs to zero
18
Transfer Function
• Since we have 2 equations, we can solve one for one
output and substitute into the other equation
𝐽𝐽𝜔𝜔 𝑠𝑠 𝑠𝑠 + 𝑏𝑏𝜔𝜔 𝑠𝑠 = 𝐾𝐾𝐼𝐼 𝐼𝐼 𝑠𝑠
𝐿𝐿𝐼𝐼 𝑠𝑠 𝑠𝑠 + 𝑅𝑅𝑅𝑅 𝑠𝑠 = 𝑉𝑉𝑠𝑠 𝑠𝑠 − 𝐾𝐾𝑏𝑏 𝜔𝜔(𝑠𝑠)
𝐼𝐼(𝑠𝑠) 𝜔𝜔(𝑠𝑠)
• Which transfer function do we want? or
𝑉𝑉𝑠𝑠 (𝑠𝑠) 𝑉𝑉𝑠𝑠 (𝑠𝑠)
• For the eigenvalues, it doesn’t matter – both transfer
functions will have the EXACT same denominator (i.e. same
characteristic equation, same eigenvalues)
• Only the numerator will be different
• Depends on which output we are interested in
– I if we want the current (maybe since torque is proportional to I).
– ω if we want the motor speed
19
Transfer Function: 𝜔𝜔(𝑠𝑠)/𝑉𝑉𝑠𝑠 𝑠𝑠
• First, lets collect terms: 𝜔𝜔 𝑠𝑠 (𝐽𝐽𝐽𝐽 + 𝑏𝑏) = 𝐾𝐾𝐼𝐼 𝐼𝐼 𝑠𝑠
𝐼𝐼 𝑠𝑠 (𝐿𝐿𝐿𝐿 + 𝑅𝑅) = 𝑉𝑉𝑠𝑠 𝑠𝑠 − 𝐾𝐾𝑏𝑏 𝜔𝜔(𝑠𝑠)
• Then use your favorite algebraic trip to eliminate I(s).
• I am going to solve the 2nd equation for I(s) and substitute
into the 1st equation:
𝑉𝑉𝑠𝑠 𝑠𝑠 − 𝐾𝐾𝑏𝑏 𝜔𝜔(𝑠𝑠)
𝜔𝜔 𝑠𝑠 (𝐽𝐽𝑠𝑠 + 𝑏𝑏) = 𝐾𝐾𝐼𝐼
𝐿𝐿𝑠𝑠 + 𝑅𝑅
• Then cleaning up:
𝜔𝜔 𝑠𝑠 𝐽𝐽𝑠𝑠 + 𝑏𝑏 𝐿𝐿𝑠𝑠 + 𝑅𝑅 + 𝐾𝐾𝐼𝐼 𝐾𝐾𝑏𝑏 𝜔𝜔 𝑠𝑠 = 𝐾𝐾𝐼𝐼 𝑉𝑉𝑠𝑠 𝑠𝑠
• Then solving for the transfer function:
𝜔𝜔(𝑠𝑠) 𝐾𝐾𝐼𝐼
=
𝑉𝑉𝑠𝑠 (𝑠𝑠) 𝐿𝐿𝐿𝐿𝑠𝑠 2 + 𝐽𝐽𝐽𝐽 + 𝐿𝐿𝐿𝐿 𝑠𝑠 + (𝐾𝐾𝐼𝐼 𝐾𝐾𝑏𝑏 + 𝑅𝑅𝑅𝑅)
20
Transfer Function: 𝜔𝜔(𝑠𝑠)/𝑉𝑉𝑠𝑠 𝑠𝑠

• Notice the denominator is 2nd order!

𝜔𝜔(𝑠𝑠) 𝐾𝐾𝐼𝐼
=
𝑉𝑉𝑠𝑠 (𝑠𝑠) 𝐿𝐿𝐿𝐿𝑠𝑠 2 + 𝐽𝐽𝐽𝐽 + 𝐿𝐿𝐿𝐿 𝑠𝑠 + (𝐾𝐾𝐼𝐼 𝐾𝐾𝑏𝑏 + 𝑅𝑅𝑅𝑅)

• The system characteristic equation is:

𝐿𝐿𝐿𝐿𝑠𝑠 2 + 𝐽𝐽𝐽𝐽 + 𝐿𝐿𝐿𝐿 𝑠𝑠 + (𝐾𝐾𝐼𝐼 𝐾𝐾𝑏𝑏 + 𝑅𝑅𝑅𝑅) = 0
– So there are two eigenvalues
– You can use your 2nd order tools!
• Do you think it is over damped or under damped?
• You can also see the DC Gain is:
𝐾𝐾𝐼𝐼
(𝐾𝐾𝐼𝐼 𝐾𝐾𝑏𝑏 + 𝑅𝑅𝑅𝑅)
21
State Space

• The other method to find the eigenvalues from multi-

DOF systems is state space
– The eigenvalues are the eigenvalues of the A Matrix
𝐽𝐽𝜔𝜔̇ + 𝑏𝑏𝜔𝜔 = 𝐾𝐾𝐼𝐼 𝐼𝐼
𝐿𝐿𝐼𝐼 ̇ + 𝑅𝑅𝑅𝑅 = 𝑉𝑉𝑠𝑠 − 𝐾𝐾𝑏𝑏 𝜔𝜔
• Placing the EOMs into state space representation:
– Note that the equations were already 1st order (so they
were almost already essentially in the correct form)

𝐼𝐼 ̇ = −𝑅𝑅/𝐿𝐿 −𝐾𝐾𝑏𝑏 /𝐿𝐿 𝐼𝐼

+
1/𝐿𝐿
𝑉𝑉𝑠𝑠
𝜔𝜔̇ 𝐾𝐾𝐼𝐼 /𝐽𝐽 −𝑏𝑏/𝐽𝐽 𝜔𝜔 0
22
Eigenvalues (State Space)

• Now give the A matrix, we can find the

characteristic equation (and eigenvalues)
−𝑅𝑅/𝐿𝐿 −𝐾𝐾𝑏𝑏 /𝐿𝐿
𝐴𝐴 =
𝐾𝐾𝐼𝐼 /𝐽𝐽 −𝑏𝑏/𝐽𝐽
𝑒𝑒𝑒𝑒𝑒𝑒 𝐴𝐴 ⇒ det 𝑠𝑠𝑠𝑠 − 𝐴𝐴 = 0

– Or in matlab: >>eig(A)

−𝑅𝑅/𝐿𝐿 −𝐾𝐾𝑏𝑏 /𝐿𝐿 𝑠𝑠 + 𝑅𝑅/𝐿𝐿 𝐾𝐾𝑏𝑏 /𝐿𝐿

det 𝑠𝑠𝑠𝑠 − = det =0
𝐾𝐾𝐼𝐼 /𝐽𝐽 −𝑏𝑏/𝐽𝐽 −𝐾𝐾𝐼𝐼 /𝐽𝐽 𝑠𝑠 + 𝑏𝑏/𝐽𝐽
𝑅𝑅 𝑏𝑏 𝐾𝐾𝑏𝑏 𝐾𝐾𝐼𝐼
𝑠𝑠 + 𝑠𝑠 + + =0
𝐿𝐿 𝐽𝐽 𝐿𝐿𝐿𝐿
23
Eigenvalues (State Space)

𝑅𝑅 𝑏𝑏 𝑅𝑅𝑏𝑏 + 𝐾𝐾𝑏𝑏 𝐾𝐾𝐼𝐼

𝑠𝑠 2 + + 𝑠𝑠 + =0
𝐿𝐿 𝐽𝐽 𝐿𝐿𝐽𝐽

• Multiplying by the prior equation by LJ we

get:
𝐿𝐿𝐿𝐿𝑠𝑠 2 + 𝐽𝐽𝐽𝐽 + 𝐿𝐿𝐿𝐿 𝑠𝑠 + (𝐾𝐾𝐼𝐼 𝐾𝐾𝑏𝑏 + 𝑅𝑅𝑅𝑅) = 0

– Note this is exactly the same characteristic

equation we saw in the transfer function
– Now use quadratic formula to solve for the
two eigenvalues and use our 2nd order tools
24
Eigenvalues Analysis

• Most motors are heavily overdamped

– This means their response can be
approximated as first order
• Due to the inductance being negligible (in terms of
dynamics, i.e. eigenvalues)
– Results in the following approximate
characteristic equation:
𝐾𝐾𝑏𝑏 𝐾𝐾𝐼𝐼
𝐽𝐽𝑠𝑠 + (𝑏𝑏 + )=0
𝑅𝑅
– Or the following differential equation:
𝐾𝐾𝑏𝑏 𝐾𝐾𝐼𝐼 𝐾𝐾𝐼𝐼
𝐽𝐽𝜔𝜔̇ + (𝑏𝑏 + )𝜔𝜔 = 𝑉𝑉𝑠𝑠
𝑅𝑅 𝑅𝑅
25
Steady State Motor Speed Curve
• DC Electric motors have
a linear torque speed
relationship at steady
state
– i.e., for 𝐼𝐼 ̇ = 𝜔𝜔̇ = 0
• You get max torque at
zero speed
– Called stall torque
– Different from an IC
engine
• You can only generate
the max speed when
there is no load
– Called no load speed http://lancet.mit.edu/motors/motors3.html/
– The net torque at this
speed is zero! 26
Motor Speed Curve

• Speed-Torque is shifted as more voltage is applied

– Optional lecture will discuss how to vary the voltage

https://robotics.stackexchange.com/questions/4492/dc-motor-control-speed-torque-curve
27
Maxon DC16 Spec Sheet

https://www.maxongroup.com/medias/sys_master/root/8833376059422/19-EN-94.pdf
28