Scribd
Upload
56 views8 pages

Ehsan 506 Project Report

This document describes optimizations made to a 2-degree of freedom mass-spring-damper system apparatus. Improvements included adding safety features, fixing code issues, and properly accounting for effects of the motor and drive system to obtain accurate force response functions. Methods to characterize the dynamic properties and optimize excitation of the system were explored.

Uploaded by

Ehsan Moosavi Mehr
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
56 views8 pages

Ehsan 506 Project Report

This document describes optimizations made to a 2-degree of freedom mass-spring-damper system apparatus. Improvements included adding safety features, fixing code issues, and properly accounting for effects of the motor and drive system to obtain accurate force response functions. Methods to characterize the dynamic properties and optimize excitation of the system were explored.

Uploaded by

Ehsan Moosavi Mehr
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 8

1

MECH 506 Term Project





Optimizing the Excitation and Characterizing the Dynamic Properties of a 2-DoF Mass-Spring-Damper
System




Ehsan M. Mehr
86464120







Dec 2013
2

1. Overview of the system

The 2-DoF mass-spring-damper system is designed and fabricated as a collaborative effort of the Dynamics
and Applied Mechanics Laboratory (DAL) and the Control Engineering Laboratory at the University of
British Columbia (Figure 1.)

Figure 1. MSD apparatus
It consists of a brushless servomotor AKM23E (driven through S20360 servo drive) which applies an
oscillatory external force on a 2-dof mass-spring system. The drive is operated under constant torque which is
supposed to apply a current proportional to its input voltage by a factor called command gain to our servo
motor. The command gain is adjustable through the software interface of the servo drive (its current value is
0.7500133 A(rms)/V. Low-friction carts with adjustable loads are used as masses. Two encoders are used to
measure the displacements of the carts. Currently, the system is programmed with MATLAB to do a sine-
dwell test and plot the response as a function of frequency without considering the effect of the rotors
dynamic stiffness or the transfer function of the electric driving system. The code had some bugs that caused
the cart to drift even in its rest time and as a result overheated the servo drive. Also, the wiring was unsafe at
some points and there was not any safety toggle switch.
In this project we consider optimizing the current MATLAB code available for driving the MSD system and
improve the safety reliability and accuracy of this procedure.
2. Improvements on the Excitation Unit

Hardware: The wiring was cleaned up and a safety toggle switch was inserted.
Software: The drifting and overheating problem was fixed by re-adjusting the offset voltage of the servo
motor to -0.050 instead of -0.054 and rewriting some parts of the code with more care. The problem was
mainly because of the wrong offset voltage and the wrong waiting times commanded by pause.
The method the previous code used to calculate the FRFs was not also exactly right. It used the wrong
parameter as its input and the FRFs were not calibrated. This will be discussed in the next chapter.
3

Another modification was made for the data analysis part. Previously, all the data in each cycle was used and a
sine function and its harmonics were fitted to it; however, it takes the system some time to reach steady-state
from rest. Therefore, in the new code we use only the final half of each interval for sine fitting. Figure 2
shows an example of the fitted data and the error.

Figure 2. An example of the sine fit used
3. Correcting the FRFs Calculated

In the system we have at hand the input we can keep track of is the voltage output from AO0 channel. The
input to the 2DoF mass-spring-damper (MSD) system, however, is equal to the vector sum of the tensions of
the cables connected to the pulley. This force is not directly related to the torque on the rotor and is affected
by the inertia of the rotor. Moreover, the inertia of the motor may not be proportional to the output voltage
from DAQ as the data sheets for servo motor and the servo drive suggest. First, we will consider the simple
2-DoF MSD system and then we will account for the effects caused by the inertia of the rotor. The next step
is to assume no proportionality in the drive and motor and find the transfer function for the system of drive
and motor.
3.1. Simple 2DoF System

The excitation used for calculating the FRF in the current code is the input Voltage to the motor. According
the data sheets for the servo motor and servo drive, this voltage itself can be assumed to be proportional to
the current that the Drive outputs to the motor which itself is proportional to the torque that the motor
applies on the system.
The equation of motion for this system is:
4


} [

] {

}
[

] [

]
Taking the terms with X to the right side and solving for steady-state conditions where


where is the frequency of excitation:
[

]
The matrix

is called the dynamic stiffness matrix. Taking the inverse of this matrix to find X in terms of
:


The matrix

is the dynamic compliance matrix also known as the position FRF matrix. Knowing F and
X we can find this matrix. However, for the system at hand we know X but we do not know F and even
assuming the proportionality we can only find the torque applied on the rotor which cannot be directly
related to F without writing the equations of motion for the whole system. Therefore, we have to consider
the full system to be able to find F.

Figure 3. Free-body diagram for the motor and the first cart
5

2.1. The complete system

The free_body diagram of the rotor and pulley is shown in Figure 3.
The Newtons equation of motion for the rotary coordinate of the pulley is:


As the lower part of the figure shows the net force applied to mass 1 is:


Thus, the relation between the torque and the net force F assumed in the previous section is:


Substituting for

in the equations of motion of the simple 2DoF system we have:


[

] {

} {

}
[

] {

} {

]
The FRF that we used to find by dividing the output voltage of the encoders by the voltage of excitation is in
fact proportional to the FRF obtained by

. For now we are only concerned


with

.
The output amplitude of oscillations is known and is the same for the simple 2-DoF system and the full
system (including the rotor.) To obtain the correct FRF for the 2-DoF system (without the rotor) however,
we have to divide this amplitude by the force F instead of


We can find

from the FRF available for the full system:


Substituting for

in the equation we have:


6


The above equation relates the FRF for the full system and system without the rotor at each frequency.
Figure 4 shows the FRF that is corrected for rotor inertia versus the FRF that is not corrected.

Figure 4. Comparison of calculating the FRF with and without considering the effect of rotor inertia

3.3 System Identification for the Excitation Unit

Next, we assume the moment of inertia of the rotor and its torque constant, as well as the value for the motor
drives command gain given by the factory are not reliable. With this assumption we lose less generality in our
analysis, because we know that each DC motor has an internal impedance and we do not have much
information about the circuit inside the drive. So we look at the drive and motor and pulley system as a black
box for which the voltage is the input and the net force on the first cart applied by the cables under it is the
output and we try to find the transfer function for this black box:



DAQ
AO0
Voltage
Net External
Force
Motor & Drive
System
The Effect of Rotor Inertia not considered
The Effect of Rotor Inertia considered

P
o
s
i
t
i
o
n

F
R
F

(
m
/
N
)

7

We do not have any extensometers or any other means to measure the tension in the cables so in order to
simplify the problem we disconnect the first cart from the springs it is connected to. Considering the friction
forces to be negligible with respect to the inertial forces the free-body diagram for this cart will be:



The equation of motion for this system in steady-state harmonic motion is:


So, for this simplified 1-DoF system without any springs; knowing the X, force can be easily found. After
running a test with this configuration we can have an estimate for the transfer function between the DAC
output voltage and the actual excitation force on the MSD system. However, still this estimation is not
reliable before 3 Hz.

Figure 5. Position frequency response with respect to input voltage
F
N

W
8


Figure 6. Excitation force frequency response with respect to input voltage
We tried to fit rational fraction polynomials to this response (which is also the transfer function for the
excitation unit) but the simplest function which could acceptably follow the trend was 10
th
order in the
numerator and 5
th
order in the denominator. However, we can see that the for frequencies higher than 3 Hz
the response is quite constant and is equal to 0.0141 (N/V) Therefore, to find the frequency response with
respect to the input force to the 2-DoF system we first need to multiply the voltage Fourier transform by this
transfer function and get the force. This method is also implemented in the MATLAB code.
4. Conclusion

Changes to the hardware (wiring, toggle switch) and the software that operate the MSD setup made this setup
safer, more accurate and more reliable. Previously, the FRF was acquired without considering either the effect
that impedance of the motor and drive impose on the input voltage or the effect of the rotor inertia on the
resultant excitation force on the MSD system.
5. References

[1] S200 High Performance Compact Brushless Servo Drives, Reference Manual, Kollmorgen
[2] AKM Series Motors Selection Guide, Daner Motion
[3] Theoritical and Experimental Modal Analysis, Maia and Silva, Research Studies Press

You might also like