Sunteți pe pagina 1din 12

Downloaded from SAE International by Li Sun, Tuesday, January 20, 2015

892483

The Control of Semi-Active Dampers Using


Relative Feedback Signals
Mark R. Jolly and Lane R. Miller
Lord Corp.

Reprinted from SP-802


Advanced Truck Suspensions

Truck and Bus Meeting


and Exposition
Charlotte, North Carolina
November 69, 1989

Downloaded from SAE International by Li Sun, Tuesday, January 20, 2015

No part of this publication may be reproduced in any form, in


an electronic retrieval system or otherwise, without the prior
written permission of the publisher.
ISSN 01487191
Copyright 1989 Society of Automotive Engineers, Inc.
Positions and opinions advanced in this paper are those of the
author(s) and not necessarily those of SAE. The author is
solely responsible for the content of the paper. A process is
available by which discussions will be printed with the paper
if it is published in SAE Transactions. For permission to
publish this paper in full or in part, contact the SAE Publications Division.

Persons wishing to submit papers to be considered for presentation or publication through SAE should send the manuscript or a 300 word abstract of a proposed manuscript to:
Secretary, Engineering Activity Board, SAE.
Printed in U.S.A.

Downloaded from SAE International by Li Sun, Tuesday, January 20, 2015

892483

The Control of Semi-Active Dampers Using


Relative Feedback Signals
Mark R. Jolly and Lane R. Miller
Lord Corp.

composed of sensors, actuators, and


microcontrollers.
However, in some
applications, cost, simplicity, and ease of
installation may be overriding factors. For
these applications, a semi-active control
approach called "relative control" has been
developed which may have the potential
for eliminating sensors, electronics, and
actuators [2]. This concept is principally
based on improving vibration isolation.
Thus, while it may be applied to primary
suspensions, secondary suspensions such as
truck cab isolators and equipment
mounting may benefit the most.
Semi-active suspensions have been
shown to provide superior vibration
isolation over conventional suspensions
[3,4,5,6,7,8]. The active damper, shown
schematically in Figure 1, is the key element
of a semi-active suspension system. An
active damper is similar to a passive
damper in that both are only capable of

Abstract
An algorithm has been developed for
the control of semi-active dampers which
uses the feedback of displacement and
velocity signals measured across the
damper. This algorithm has been called
"relative control." Experimental and
computer simulation data show that the
performance achieved with relative control
can be superior to that of passive dampers.
The motivation for examining this
approach is that relative control can be
implemented without electronics. Thus, an
all-mechanical damper device, which in
a p p e a r a n c e looks identical to a
conventional shock absorber, can be
designed to implement relative control.
Introduction
In recent years, controlled suspension
systems have begun to evolve into the
market place. Technologies ranging from
manually adjustable to fully active
suspensions have been introduced or are
being developed [1]. Obviously, the cost of
the suspension system varies greatly with
its level of sophistication. Many automotive
and trucking applications may require
active suspension systems which are
9

Downloaded from SAE International by Li Sun, Tuesday, January 20, 2015

dissipating energy. However,


the
force/velocity characteristics of the active
damper can be varied by modulation of a
valve.
A conventional
single-degree-offreedom (SDOF) suspension model is
shown in Figure 2. This system is capable of
isolating the mass M from base inputs by a
parallel combination of a spring K and

The transmissibility, in this case, is the


ratio of the output motion to the input
motion. This ratio can be of displacement,
velocity or acceleration. Note that a family
of curves is generated by varying the
damping ratio, .

As damping is increased, isolation


improves at low frequencies, but degrades at
high frequencies. This trade-off is an
inherent disadvantage of conventional
suspension systems.
While semi-active suspensions can be
designed to eliminate this performance
trade-off [5], they tend to be
more
complicated and costly. Typically, the semiactive suspension will include sensors,
microprocessors and valves. Currently,
semi-active control policies require absolute
velocity of the suspended mass as a feedback
signal. Because absolute velocity is
impossible to measure directly, special
filtering techniques [93 have been developed

damper C. The transmissibility of this


conventional system is shown with solid
lines in Figure 3.

10

Downloaded from SAE International by Li Sun, Tuesday, January 20, 2015

to produce estimates of the absolute velocity


from accelerometer signals.
This paper will present a new control
scheme, which has been called relative
control, that lends itself to a simpler, less
expensive suspension system. The
performance of such a suspension system is
predicted to fall between that of a passive
system and that of current semi-active
systems.
'Skyhook' Control
Currently,
most
semi-active
suspension research efforts investigating
real-time control use control schemes
which are, in part, derived from what is
often called skyhook control [5]. Because, in
essence, relative control can be viewed as a
derivative of skyhook control, it is
appropriate to first review skyhook control
theory.
Figure 2 shows a conventional baseexcited SDOF system. The total force on the
mass provided by the passive linear
elements K and C is given by

reasoned. The 'skyhook' damper tends to


resist the absolute velocity of the mass,
while the conventional damper tends to
resist the relative velocity between the mass
and the base. Hence, for high frequency
input, the conventional damper tends to
stiffen the suspension where a soft
suspension is desirable.
It is obvious that in most practical
cases, such as vehicle suspensions, the
'skyhook' damper configuration is
impossible to achieve. It is also impossible
for a passive device located between the
mass and the base to always mimick
'skyhook' damper force. For example, a
conventional passive damper is only
capable of resisting relative velocity, while a
'skyhook' damper generates forces
independent of the relative velocity.
However, with the use of a semi-active
damper, as shown in Figure 1, it may be
possible to create any damping force Fd
between the mass and the base, as long as
the power associated with that force is
dissipated. In other words, it may be
possible to vary damping C such that the
force in Equation (1) looks like the force in
Equation (2) when

where X1 is the velocity of the mass, X2 is


the relative displacement between the mass
and the base, and U is the base velocity
input. Through the study of optimal
control, it can be seen that the damper C
should be located between the mass and an
inertial reference such that the damper
force is proportional to the absolute velocity
of the mass [5]. The force on the mass
becomes

This configuration, known as the 'skyhook'


damper system, is shown in Figure 4.
Transmissibility curves for the
'skyhook' damper system are shown in
dotted lines in Figure 3. It can be seen that,
indeed, as damping increases, isolation
improves at both high and low frequencies.
The results of Figure 3 can be intuitively
11

Downloaded from SAE International by Li Sun, Tuesday, January 20, 2015

Since the semi-active damper cannot supply


power to the system, the best it can do is to
supply no damping force at all when
Equation (3) is not satisfied. With the above
considerations, a control policy can be
constructed that will allow a semi-active
damper located between the mass and the
base to behave as close to a 'skyhook'
damper as possible. This policy, given in
Equations (4), will herein be referred to as
skyhook control.

Assuming imperfect suspension devices,


the in Equation (4) might imply "attempts
to equal" or "ideally equals". It should be
noted that the above considerations apply
only to linear SDOF systems.
A version of skyhook control may also
be implemented with an on/off type valve
instead of a continuously variable valve.
This means of control, often referred to as
on/off control, is capable of creating a high
relative velocity damping coefficient or a
near zero damping coefficient, as dictated by
the logic in Equations (4). On/off control
generally performs worse than skyhook
control, but is much easier to implement.
Currently, most forthcoming semi-active
suspension systems for automobiles use
on/off control.

Relative control was developed by


means of intuitive reasoning. Once again,
semi-active control will be applied to a
linear SDOF system as shown in Figure 2.
The intuitive basis for relative control can
be realized by considering the suspension
force transmitted to the mass. In Equation
(5), the suspension force FS is a combination
of the spring force KX2 and the damper force
C(X1-U).

Intuitively, for good isolation, we


would like to minimize the suspension
force FS on the mass. During oscillation,
there are times when the spring force and
damper force act in opposite directions.
This situation occurs when the relative
dynamic displacement X2 and the relative
velocity X1-U act in opposite directions.
Since the damper force opposes the spring
force, the orifice area can be set such that the
damper force and spring force tend to cancel
each other. The spring force and the damper
force act in the same direction when the
relative dynamic displacement and the
relative velocity act in the same direction.
Since the spring force and damper force are
additive in this case, the damping
coefficient is set to zero to minimize the
overall suspension force. Such control,

Relative Control
At the onset, the development of
relative control was spurred by the desire to
eliminate early difficulties in estimating the
absolute velocity of the sprung mass.
Further investigation and development
uncovered more inherent advantages, one
of which is the possibility of all-mechanical
implementation [2]. The following
development will outline the basic theory
of relative control giving potential
advantages and disadvantages as compared
to skyhook control.
12

Downloaded from SAE International by Li Sun, Tuesday, January 20, 2015

which will herein be called relative control,


is given by Equations (6)

to the results of three other suspensions; 1)


conventional, 2) skyhook controlled, and 3)
fully active. Figure 6 gives a plot of the
transmissibilities of the four different
systems. It can be seen that the best
performing system is the fully active, which
essentially replaces the suspension spring
and damper with an active force generator.
With such a system, it is theoretically
possible to mimick the forces that would be
generated by a 'skyhook' damper. Relative
control performs better than the
conventional (passive) system, but slightly
worse than skyhook control. Note that at

Simulations have shown that optimal


isolation is obtained when the gain is
near unity. Further investigation has
shown that relative control is not too
unlike skyhook control. Equations (6) differs
from Equations (4) in that the damping
force is varied between zero and some
varying finite, value depending upon the
sign of the product of relative velocity (X1U) and relative displacement X2, rather than
absolute velocity X1. For a broad frequency
range, the relative displacement between
the base and the mass X2 is in phase with
and of proportional magnitude to the
absolute velocity of the mass X1. The
transfer function between the two states is
given in Figure 5. Indeed, it can be seen that
the two states behave similarly above the
natural frequency of the SDOF system.
With the above considerations, a
potential problem of relative control
becomes evident from Figure 5. There is
very poor correlation between relative
displacement and absolute velocity at low
frequencies. While this may have very little
effect on the ability of relative control to
isolate vibration, the system's performance
may degrade at low frequency inputs.
Once again, the relative control logic
may also be implemented with an on/off
valve with a slight degradation in
performance. It will become apparent that
implementation of relative control with an
on/off valve may only be easier in
electromechanical systems, and not allmechanical systems.

high frequencies, relative


control
performance is very similar to that of
skyhook control. As predicted, at very low
frequencies, relative control performs
slightly worse than the passive system. In
fact, Figure 6 suggests that the use of relative
control may result in an amplitude ratio
greater than 1.0 at 0.0 hz. Additional
simulations were conducted at extremely
low frequencies to study this effect. These
simulations showed that the amplitude
ratio approaches 1.0 at frequencies below
0.05 hz.

The Performance of Relative Control


Computer simulations were conducted
of various types of suspension systems. To
demonstrate the performance, simulation
results of relative control will be compared
13

Downloaded from SAE International by Li Sun, Tuesday, January 20, 2015

Note that in Figure 6, relative control


might perform similarly to a passive system
which is tuned to a lower natural frequency.
This performance, of course, occurs without
any reduction in the spring constant and
w i t h o u t the a d d e d
problems
of
precompression and vehicle height control.
From a vibration isolation standpoint,
relative control allows the suspension
system to have the vibration isolation
quality of a system which is tuned to a lower
natural
frequency
without
the
disadvantages that a corresponding passive
system would have.
While the simulation results of
relative control may look promising,
putting it to practice will be the ultimate
judge of its worthiness. Inherent
disadvantages of relative control may make
it unsuitable for many suspension
applications. For instance, a relative control
system, optimally tuned for higher
frequency isolation requires
more
suspension travel than competing means of
control. Advantages of theoretical or actual
vehicle implementation of vehicle primary
suspensions, which generally require
optimization of suspension travel, wheel
dynamics and isolation, have yet to be
demonstrated. Simulations have also
shown that relative control does not
respond favorably to disturbance forces
applied directly to the suspended mass.
It is likely that relative control may
perform better in applications where most
of the disturbance energy is transmitted at
higher frequencies. Such applications might
include truck cab suspensions.

sent to a microprocessor which implements


the control policy. In the case of relative
control, the control policy given in
Equations (6) would be implemented. The
result of this control policy implementation
would be a force command which would be
conditioned and sent to the valve of an
active damper similar to the one shown in
Figure 1. Note that an accelerometer is not
needed and that the feedback quantities are
all measured in the rattle space volume
between the base and the mass.
The
term
relative
dynamic
displacement is used to describe the
measure of deflection between the mass and
the base which does not include static
deflections. Therefore, the LVDT would

essentially have to be "zeroed" once


installed into the suspension system while
the car is at rest. If the mass M does not
change, this procedure will work. However,
if the mass M changes by fairly large
amounts, then some scheme to "re-zero"
the LVDT output may be necessary.
An alternative configuration may be
used to implement relative control. In this
configuration, the LVT and the LVDT
would be replaced with load cells located in
series with the damper and spring,
respectively. Hence, load cells are used to
produce signals of the actual spring and
damper forces. According to the relative
control policy of Equations (6), the direction
of the relative velocity (damper force) is

Electromechanical Realization of Relative


Control
Figure 8 shows a schematic
representation of how relative control
might be implemented on a SDOF system.
A relative velocity signal can be provided by
a Linear Velocity Transducer (LVT) and a
relative dynamic displacement signal can be
provided by a Linear Variable Differential
Transformer (LVDT). These two signals are
14

Downloaded from SAE International by Li Sun, Tuesday, January 20, 2015

needed rather than the actual magnitude.


So, even though the damping coefficient is
constantly changing, the force transducer
connected to the damper can still determine
the direction of the relative velocity.
The system shown in Figure 8 was
demonstrated in the laboratory. The
laboratory configuration has an undamped
natural frequency of 1.5 Hz. A hydraulic
shaker provides random base input which
is approximately flat in the frequency
spectrum out to 10 hz and then decreases in
magnitude. The active damper [10] uses a
passive pressure relief valve in parallel
with an adjustable bypass orifice. A stepper
motor actuator is used to adjust the orifice
to achieve the desired damping force. This
arrangement essentially allows the damper
to be a force generator when relative
velocity is such that the desired force
dissipates energy. Under any other
condition, the bypass valve is commanded
to a wide open position, such that the
damping force is minimal.

but poor at attenuating higher frequency


input. The soft damper effectively
attenuates higher frequency noise, but is
unable to attenuate the resonant peak. This
results in excessive mass motion at the
system's damped natural frequency. It can
be seen that the use of relative control
allows for the attenuation of both the
resonant peak and high frequency input.
Here again, it can be seen that the use of
relative control shifts the resonant peak to a
lower frequency.
All-Mechanical Implementation of Relative
Control
One of the key aspects of relative
control, which warrants some investigation,
is the possibility of all-mechanical
implementation. This aspect gives relative
control the potential for being as
inexpensive as a conventional shock
absorber. In Figure 8, a basic schematic of an
all-mechanical configuration
which
implements the relative control policy is
shown [2]. This device uses direct
mechanical feedback of relative position X2
and relative velocity X1-U to implement the
relative control policy of Equations (6).
With the system at rest, the relative
dynamic displacement X2 is assumed to be
zero. Two check valves are in place between
the upper and lower chambers of the
damper such that flow, resulting from
relative motion, must pass through one of
the two spring loaded poppets. The flow can
then pass through a check valve into the
opposite damper chamber. It is assumed
that when the system is at rest no load is on
either sprung poppet. When the mass is
displaced up or down from the nominal
position, the mechanical feedback linkage
loads either the lower or upper poppet,
respectively. The poppets behave like
pressure relievers in that they produce a
pressure drop that is approximately
proportional to preload imposed on the
poppet springs.
The following discussion is intended to
show how the device of Figure 9

Three experimentally generated


transmissibility curves are shown in Figure
8. The three transmissibility curves are
representative of systems using a soft
passive damper, a stiff passive damper and
an active damper using relative control. It
can be seen that the stiff damper is very
effective at attenuating the resonant peak,
15

Downloaded from SAE International by Li Sun, Tuesday, January 20, 2015

implements the relative control policy of


Equations (6). For this argument, extended/
extending will be considered to be positive
and compressed/compressing will be
considered to be negative.
First suppose that the relative dynamic
displacement X2 is positive and the relative
velocity X1-U is positive. In this case, the
feedback linkage loads the lower sprung

results. These two situations correspond to


the second equation of Equations (6).
When
the
relative
dynamic
displacement and the relative velocity have
opposite signs, the flow will be such that it
must pass through a sprung poppet which
has been loaded by the feedback linkage. In
this situation, substantial flow restriction
occurs and high damping force is created. It
can be seen that as relative displacement
increases, the feedback linkage provides
increasing load on one of the sprung
poppets. Hence, the flow resistance and the
damping force increase as the magnitude of
relative displacement increases. In fact, if
the poppet areas and the poppet springs are
designed correctly, damping forces as
described by the first equation of Equations
(6) can be approximated.
Practical designs for mechanically
implementing relative control are being
pursued. Most of the ways investigated
involve integrating the necessary valving
into the shock absorber piston head.
Conclusion
The performance of semi-active
relative control falls between the
conventional passive suspension and the
semi-active suspension which utilizes
skyhook control. Because of its inherent
lower cost, when implemented in a
mechanical design, relative control has the
potential for quick market introduction.
To be marketable, semi-active relative
control has to be less expensive than semiactive skyhook control and, at the same,
time it must perform better than a
conventional
system. This
report
demonstrates that these two criteria are
possible. In fact, the cost of a semi-active
relative control suspension may rival that
of a conventional suspension. This
possibility may lead to a sizeable retrofit
market. While superior vibration isolation
has been demonstrated, semi-active relative
control may require more suspension
travel. This requirement may limit the use
of relative control to vehicles that can

poppet and unloads the upper sprung


poppet. Since flow must pass from the
upper chamber to the lower chamber, it
must pass through the unloaded poppet.
This results in relatively free flow and
minimal damping force is created. When
the relative displacement and the relative
velocity are both negative, the opposite flow
situation is created and, again, free flow
16

Downloaded from SAE International by Li Sun, Tuesday, January 20, 2015

accommodate larger suspension travel.


Other vehicle dynamics, such as handling
and wheel hop, have not been investigated
thoroughly. Also, the use of relative control
in secondary suspensions, such as truck cab
suspensions, should be investigated. Semiactive relative control warrants continuing
development effort.

Damper," Shock and Vibration Bulletin,


Vol. 51, May 1981.
[9] Miller, L. R., "The Effect of Hardware
Limitations on an On/Off Semi-active
Suspension," Proc. IMechE, Paper No. C367020, Oct. 1988.
[10]
Ivers,
D.E.,
Miller,
L.R.,
"Experimental Comparison of Passive,
Semi-active On/Off, and Semi-active
Continuous Suspensions", 1989 SAE
Truck and Bus Meeting and Exposition,
Nov. 1989.

REFERENCES
[1] Wright, P. G. and D. A. Williams, "The
Application of Active Suspension to High
Performance Road Vehicles," Paper No.
C239/84, IMechE, 1984.
[2] Miller, L.R., "Control Method and Means
for Vibration Attenuating Damper," U.S.
Patent 4,821,849," April 1989.
[3] Crosby, M. J. and D. C. Karnopp, "The
Active Damper," The Shock and Vibration
B u l l e t i n , Vol. 43, Naval Research
Laboratory, Washington, D. C , 1973.
[4] Sharp, R. S. and S. A. Hassan, "The
Relative Performance Capabilities of
Passive, Active and Semi-Active Car
Suspension Systems," Proc. IMechE, Vol.
200, No. D3,1986.
[5] Miller, L. R. and Nobles, C. M., "The
Design and Development of a Semi-active
Suspension for a Military Tank," Proc. SAE,
No. 881133, Aug. 1988.
[6] Margolis, D. L., J. L. Tylee, and D. Hrovat,
"Heave Mode Dynamics of a Tracked Air
Cushion Vehicle with Semi-active Airbag
Secondary Suspension," Journal of Dynamic
Systems, Measurement, and Control, ASME
Publication, Dec. 1975.
[7] Krasnicki, E. J., "Comparison of
Analytical and Experimental Results for a
Semi-Active Vibration Isolator," Shock and
Vibration Bulletin, Vol. 50, Sept. 1980.
[8] Krasnicki, E. J., "The Experimental
Performance of an 'ON-OFF' Active
17

Downloaded from SAE International by Li Sun, Tuesday, January 20, 2015

Positions and opinions advanced in this paper are those of the


author(s) and not necessarily those of SAE. The author is
solely responsible for the content of the paper. A process is
available by which discussions will be printed with the paper
if it is published in SAE Transactions. For permission to
publish this paper in full or in part, contact the SAE Publications Division.

Persons wishing to submit papers to be considered for presentation or publication through SAE should send the manuscript or a 300 word abstract of a proposed manuscript to:
Secretary, Engineering Activity Board, SAE.
Printed in U.S.A.

S-ar putea să vă placă și