Proceedings of the 6th JFPS International

Symposium on Fluid Power, TSUKUBA 2005

November 7-10, 2005

2A4-4

Model of a nozzle-flapper type pneumatic servo valve and

differential pressure control system design
Tao WANG*, Maolin CAI**, Kenji KAWASHIMA** and Toshiharu KAGAWA**
* Department of Mechano-Micro Engineering
Interdisciplinary Graduate school of Science and Engineering
Tokyo institute of Technology
4259 Nagatsuta-cho, Midori-ku, Yokohama, 226-8503 Japan
(E-mail: wangtao@k-k.pi.titech.ac.jp)
** Precision and Intelligence Laboratory
Tokyo institute of Technology
4259 Nagatsuta-cho, Midori-ku, Yokohama, 226-8503 Japan

ABSTRACT
In pneumatic position and force control systems, nozzle-flapper type servo valves are usually used for the pressure
control. This paper proposes a linear model for a 4-port nozzle-flapper type pneumatic servo valve, which also includes
the influence of the flow force, by deriving it from a full nonlinear model. The proposed linear model had a better
agreement with experimental dynamics response than the former one which neglects the flow force. Furthermore, the
proposed model was applied to achieve the differential pressure control between two symmetrical control chambers. A
two degree-of-freedom (TDF) controller was designed based on the proposed model. The experiment results showed
that differential pressure controllability could be improved for the reference input tracking and disturbance restraining
by using the proposed model.

KEY WORDS
Differential pressure control, Nozzle-flapper type servo valve, Linear model, Flow force, TDF controller

NOMENCLATURE
Bf
bo1, o2
bn1, n2
Co1, o2
Cn10, n20
Gc1, c2

damping coefficient of the

armature-flapper
critical pressure ratio of the left
and right fixed orifice
critical pressure ratio of the left
and right nozzle-flapper
sonic conductance of the left and
right fixed orifice
sonic conductance of the left and
right nozzle-flapper at zero input
current
mass flow rate of the left and right
control side

Go1, o2

[Nms]

Gn1, n2

[--]

i
J
Ka

[--]
[sm4/kg]

Kf1, f2

[sm4/kg]

Km
Kt
lf

[kg/s]

mass flow rate through the left and

right fixed orifice
mass flow rate through the left and
right nozzle-flapper
input current
moment of the armature-flapper
rotational stiffness of the flexure
tube
flow force cofficient of the flapper
left anf right side
magnetic rotational stiffness
torque-motor gain
distance from the nozzle to the
pivot point of the armature-flapper

[kg/s]
[kg/s]
[mA]
[kgm2]
[Nm/rad]
[N/Pa]
[Nm/rad]
[Nm/mA]
[m]

100092 wang 1/6

322

Copyright 2005 by JFPS, ISBN 4-931070-06-X

mo1, o1
mn1,n2
Pa
Pc1, c2
Ps
R
V1, 2
Te
Tt
x
xf
P

subsonic index of the left and right

fixed orifice
subsonic index of the left and right
nozzle-flapper
atmosphere pressure
left and right side control pressure
supply pressure
gas constant
volume of the left and right
control chamber
time constant due to eddy currents
torque caused by electromagnetic
forces
flapper tip displacement
distance from flapper tip to each
nozzle at zero input current
differential pressure between both
control sides P=Pc2-Pc1
armature-flapper deflection
control chamber air temperauture
air density for standard conditions

[--]
Dynamics of servo valve and control chamber
[--]
The differential pressure control system shown in Figure
1 consists of a 4-port nozzle-flapper type
electronic-pneumatic servo valve with dual fixed
orifices and dual nozzles and two load chambers called
as Isothermal Chambers. The armature-flapper is moved
by a torque motor consisting of a coil and vertically
arranged magnets. Movement of the flapper changes the
distance between it and both nozzles. This creates
different pressures, Pc1 and Pc2, in the control sides. In
the steady state, the differential pressure, P=Pc2-Pc1, is
proportional to the input current The two Isothermal
Chambers, in which the copper wires with the diameter
of 50 m are stuffed uniformly, are connected to the two
control ports of the servo valve as load chambers
respectively.

[Pa]
[Pa]
[Pa]
[m2/s2K]
[m3]
[s]
[Nm]
[m]
[m]
[Pa]
[rad]
[K]
[kg/m3]

COIL
ARMATURE
MAGNET

INTRODUCTION

In many pneumatic position and force control systems,

nozzle-flapper type servo valves are normally used for
controlling small flow rates. For example, the air-spring
type horizontal vibration isolation is mainly achieved by
differential pressure control using a 4-port
nozzle-flapper type pneumatic servo valve due to its
simple construction, high sensitivity and wide frequency
band [1]. Therefore, modeling the servo valve is
important for designing a good performance control
system. However, there was no general mathematic
model describing a nozzle-flapper type pneumatic servo
valve considering the flow force on the flapper. The
nozzle-flapper type servo valve is usually treated as a
proportional element with the torque motor dynamitic
and the flow force being neglected. The pneumatic
control system with a nozzle-flapper element is
generally approximated to a first order lag system
assuming an isothermal state change for the air in the
load chamber [2]. The conventional control approach
for differential pressure control uses a PID controller
regarding the system model as a first-order model, in
which parameters are recognized by experiment
directly.
In this research, a full nonlinear model of the
nozzle-flapper type servo valve was derived including
the torque motor dynamics and the flow force. The
linear model is derived through linearizing the nonlinear
model around it equilibrium state. Through the
simulation and experiment for the dynamics response,
the influence of the flow force is confirmed. Using the
proposed linear model a two-degree-of-freedom control
system is designed for the differential pressure between
the two load chambers.

YOKE
FLEX-TUBE
FLAPPER
NOZZLE

Ps

Ps
Pa
Pc1

V1

ORIFICE

x
Pc2

ISOTHERMAL
CHAMBER

V2

Figure 1 differential pressure control system

The toque Tt, driving the armature-flapper generated by
the coil current considering the influence of the eddy
current, is given as the following [3]:
Te T&t + Tt = K m + K i i

(1)

The moment around the pivot of the armature-flapper

around is obtained by

J&x& l f = Tt K a + K f 1 Pc1 K f 2 Pc 2 l f B f x& l f (2)

The first term is toque of the torque motor as Eq. (1).

The second term is the torque on the armature-flapper
around the pivot, caused by the flex-tube which acts as
an elastic support. The third term is the torque on the
flapper around the pivot, caused by flow forces which
are generated by air jet from the two nozzles.

100092 wang 2/6

323

Copyright 2005 by JFPS, ISBN 4-931070-06-X

Air temperature inside the chamber is considered

constant and isothermal state during pressure and flow
rate change. It is because that the high heat capability of
copper wire than air and large heat area between the
copper wires and air inside the chambers. So air
dynamics in two load chambers are obtained from air
state equation [4].

In Eq. (2), the left and right side flow force coefficients
are determined by the experimental approach using the
following approximant equations [5].

K f1

a14

x f +x

a13

K
x f
= i a11 (a11 a12 )e

lj

(3)

Kf2

a 24

x f x

a23

Ki
x
f

=
a 21 (a 21 a 22 )e

lj

(4)

Gc =

G c = Go G n

(5)

Pc

Linear model for differential pressure model

The linear model is derived from the above nonlinear
model at an equilibrium state. The equilibrium states are
obtained for zero input, i.e., i=0 as follows:
0=0, x0=0,
Pc10 and Pc20 can be calculated from Eqs. (5)-(7) and
(11) when the flow rate through the fixed orifice equals
that through the nozzle-flapper; the flow state at the
fixed orifice is subsonic and that at the nozzle-flapper is
sonic.

mo

Pc

Ps

Ps

bo

> bo

Go, Co, Pc, bo, mo and o, for the left and right side of the
flapper are identified by attaching the subscripts 1 and 2
respectively.
The flow rates through the two nozzle-flappers are
given by
G n1 = C n10
G n 2 = C n 20

(9)

Gc, Kc, V, Pc, Go and Gn for the left and right side of
flapper are expressed by attaching the subscripts 1 and 2
respectively.

where

o ( Pc ) = Pc bo
Ps
1
1 bo

(8)

Applying the continuity equation to the 4-port

nozzle-flapper servo valve, the following equation can
be obtained:

With respect to the compressibility of air, the flow state

through the fixed orifice and the nozzle-flapper can be
sonic or subsonic depending upon the upstream/
downstream pressure ratio. Firstly, the flow rate through
the fixed orifice is given by
Go = Co Ps o ( Pc )

V &
Pc
R

xf + x
xf
xf x
xf

Pc1 n1 (Pc1 )

(6)

Pc 2 n 2 (Pc 2 )

(7)

Pc10
bo1

Ps
Co1 Ps 1

1 bo1

Pc 20
bo 2

Ps
Co 2 Ps 1
1 bo 2

Pa

Pa

Pc

Pc

= C n10 Pc10

(10)

mo 2

= C n 20 Pc 20

(11)

Because the two control pressures at equilibrium are not

zero, the pressure variations and flow rate variations
dynamics to the input current variation derived in
Appendix A, have the forms shown in Figure 2. This
block diagram shows that a differential pressure control
system for two fixed chambers with a nozzle-flapper
type servo valve can be validly represented by a fourth
order model. Furthermore, the flow force can be
considered as the disturbances to flapper movement.
The specific servo valve used in the experiment system
is part AS121005 as manufactured by the PSC Company.
The control pressure change range is 300~400 kPa for
the rated current of 100 mA and air supply pressure of
400 kPa. The rated differential pressure is 8.6 kPa for an
input current of 10 mA. The two chambers with the

where

2 n
Pa

n ( Pc ) =
b
Pc n
1

1 bn

mo1

bn

> bn

In the above equation, n Pc, bn, and mn, for the left and
right side of the flapper are identified by attaching the
subscripts 1 and 2 respectively. The nozzle-flapper
critical pressure ratio and subsonic index are
considering constant with the flapper displacement
change.

100092 wang 3/6

324

Copyright 2005 by JFPS, ISBN 4-931070-06-X

effective volume of 95 cm2, in which the copper wires

with the diameter of 50 m and weight of 0.4 kg are
stuffed uniformly, are connected to the two control ports

of the servo valve closely neglecting connecting section.

The experiments were performed at a constant ambient
temperature of 20 C and a supply pressure of 400 kPa.

Nozzle-flapper type servo valve

Control chamber

K10l 2f (Te s + 1)

ax1

Ki lf

+
-

1
3

Gc1

Gn1 -

+
+

+ Go1

ap1

ao1

Kc1

1
s

Pc1

a3 s + a2 s + a1s + a0

+
ap2
ax2

ao2

+
+
+ Gn2 -

Go2
Gc2

Kc2

1
s

Pc2

K 20l 2f (Te s + 1)

Figure 2 Schematic of a linear model for a combination of servo valve and control chambers

P [kPa]

35
30
25
20
15
10
5
0
-5
-10
-15
-20
-25
-30
-35
0.0

Gain [dB]

To validate the derived linear model, the step responses

and frequency responses for the proposed model and
experiment were achieved. Various step input currents
were applied to the model and experimental setup. The
input current was varied using step-up signals with
magnitudes of 10%, 20% and 30% of the rated
current. There is good agreement between the results of
the simulation and experiment, as shown in Figure 3.
Furthermore, comparison between the results of the
experiment and the simulation which neglects the flow
force yields a steady state error of about 30%. This
shows that flow force has the effect of reducing the
pressure gain of the servo valve.

2.0

2.5

3.0

Phase [deg]

-270
0

10

10
Fequency [Hz]

10

10

The frequency response was obtained by superimposing

a 5 mA swept sinusoidal current on a biased zero input
current and is summarized in the bode diagram shown
in Figure 4. From this bode diagram we may see that the
gains for both the experiment and model descended to 3
dB at about 0.65 Hz. At about 400 Hz a peak appears in
the bode diagram and corresponds to the resonant
frequency of the armature-flapper. The influence of the
flow force on frequency response is almost negligible.
The dynamics of the system is mainly dominated by the
air dynamics in the control chamber in the low
frequency field. Many electro-pneumatic servo
applications demand closed loop bandwidth far below

Experiment
Simulation including flow force
Simulation neglectinging flow force
t [s]

-180

Figure 4 Frequency response of the differential pressure

-10%10%

1.5

-90

-360 -1
10

-20%20%

1.0

Experiment
Simulation including flow force
Simulation neglecting flow force

-30%30%

0.5

0
-10
-20
-30
-40
-50
-60
-70
-80

3.5

Figure 3 Step response of the differential pressure

100092 wang 4/6

325

Copyright 2005 by JFPS, ISBN 4-931070-06-X

the servo valve bandwidth, therefore the dynamics of

the combination of servo valve and control chambers
can approximately be considered as a first order loop for
the servo control design.

We introduced the parameters of the above two TDF

controllers to the digital experimental control system
respectively and compare the control results to the step
input and step disturbances respectively.

Control system control design

Table 1 FB compensator parameters

To obtain high performance control results for reference

tracking and disturbance restraining simultaneously, a
two degree-of-freedom (TDF) controller is designed for
the differential pressure control. It comprises two
components, a prefilter and a feedback compensator.
The control system is shown as Figure 5.

Gc2 (s )

Prefilter

Gc1 (s )

G p (s )

Figure 5 Differential pressure control system with a

TDF controller

0.238

3.375

0.270

P [kPa]

20
15
10

Reference
Model including flow force
Model neglecting flow force
Servo valve +Chamber

(12)
5

The prefilter is designed as follows:

1 + c1c 2 Ti s
1 + c 2 Ti s

4.5

25

Gp(s) is the transfer function of the combination of

servo valve and control chambers shown in Figure 2.
The FB compensator is designed as a PI controller
consisting of a proportional coefficient Kp and an
integral coefficient Ti:

G c 2 (s ) =

Ti [s]

Figure 5 shows the experiment step response using each

TDF controller for a reference amplitude of 20 kPa.
Better differential pressure response is achieved with the
more accurate control model including the influence of
the flow force and the rising time to 64% of the
reference is 53 ms. Compared with it the rising time of
the combination model and the control model neglecting
the flow force of the servo valve are 257 ms and 73 ms
respectively.

FB compensator

Gc1 (s ) = K p 1 +
T
is

Model including the

flow force
Model neglecting the
flow force

Servo valve + chamber D

Pref

Kp

No

0
0.0

(13)

0.2

0.4

t [s]

0.6

0.8

1.0

Figure 6 Step reference tracking responses

Different outputs results for models, with and without

flow force, with respect to input current show the flow
force is mainly influencing the gain of the transfer
function for servo valve show in Figure 3. Therefore the
different controllers need to be designed for different
models with respect to the same reference input.
In order to confirm control effect, using the models
including and neglecting the flow force we designed the
different controllers. The controller design is performed
as the following. Firstly because the model has been
established clearly shown as Figure 2, the parameters of
the FB compensator can be tuned using simulating the
model, no depending upon the experiment investigation,
by the Critical Gain Method of Ziegler-Nichols shown
as Table 1. Next the parameters of the prefilter are
usually recommended as c1=0.4 and c2=1.35 referring to
[6].

25

P [kPa]

20
Reference
Model including flow force
Model neglecting flow force
Servo valve +Chamber
Disturbance

15
10
5
0

t [s]

Figure 7 Disturbance restraining response

100092 wang 5/6

326

Copyright 2005 by JFPS, ISBN 4-931070-06-X

where:

As a disturbance, an additional pressure with the

amplitude of 3.7kPa and time interval of 1.74 s is added
to right control chamber after the control system has
settled to the steady state. Figure 7 shows that
disturbance restraining response of the control model
including the influence of the flow force and neglecting
it at the steady state of 20 kPa. Quicker disturbance
restraining effect with the control model including the
influence of the flow force shows the differential
pressure controllability could be improved.

a o1 =

dGo1
dPc1

A linear model for the differential pressure has been

derived from the nonlinear model for a 4-port
nozzle-flapper type pneumatic servo valve and two
control chambers including the flow force of the servo
valve. Simulation and experiment show that the flow
force influences the steady state error of the servo valve
dynamics. The differential pressure control system
design using a TDF controller has been performed based
on the two models including and neglecting the flow
force. As expected and verified by the experiment,
better control effect for reference input tracking and
disturbance restraining is achieved for the proposed
model including the flow force than the model
neglecting it. It is signified that the differential pressure
controllability can be improved by using the proposed
model.

where:
Gn1
a x1 =
x
ax2 =

K c1 =

(A4)

G n 2 = a x 2 x + a p 2 Pc 2

(A5)

x = x0
Pc1 = Pc10

, a p1 =

x = x0
Pc 2 = Pc 20

Gn1
Pc1

, a p2 =

x = x0
Pc1 = Pc10

Gn 2
Pc2

x = x0
Pc 2 = Pc 20

G c1 (s ) = K c1Pc1 (s )

(A6)

G c 2 (s ) = K c 2 Pc 2 (s )

(A7)

R
R
, K c2 =
V1
V2

REFERENCES

x (s )

1. Shearer, J. E., Study of Pneumatic Process in the

Continuous Control of Motion with Compressed Air-I,
II, Transactions of ASME. 1956, Feb. pp.233-249.
2. Zalmanzon, L. A., Components for Pneumatic
Control Instruments, 1965, Pergamon Press.
3. Urata, E., Influence of eddy current on torque-motor
dynamics, 4th IFK Workshop, 2004. Mar 24. pp.
71-82.
4. Kawashima, K. Fujita, T. and Kagawa, T., Flow Rate
Measurement of Compressible Fluid using Pressure
Change in the Chamber, Trans. of the Society of
Instrument and Control Engineering, 2001, Vo1.1,
No.1, 32, p.252-258.
5. Wang, T. Cai, M. Kawashima K and Kagawa, T.,
Modelling of Pneumatic Nozzle-flapper servo valves,
Spring Symposium of JFPS, 2004, May 26, pp. 70-72
6. Hiroi, K., US5105138, Two degree of freedom
controller, Kabushiki Kaisha Toshiba, Kawasaki,
Japan, 1992, April 14.

(A1)

a 3 s 3 + a 2 s 2 + a1 s + a 0

where
a 0 = K a K m , a1 = B f + K a Te
a 2 = J + B f Te , a 3 = JTe

K10 is Kf1 at the equilibrium position.

K20 is Kf2 at the equilibrium position.
By linearizing Eq. (7) at the equilibrium position, the
flow rate variation at the fixed orifice can be obtained
from the control pressure variation as follows:
Go1 = a o1Pc1
(A2)
Go 2 = a o 2 Pc 2

Gn 2
x

Gn1 = a x1 x + a p1 Pc1

where:

From the armature-flapper model in Eq. (1) and (2), a

transfer function, relating the displacement due to input
current and the control pressure can be derived as
follows:

Pc 2 = Pc 20

The transform function of the dynamics in the control

chamber shown as in Eq (10) can be written as follows:

Appendix A: Derivation of the linear model

Pc1 = Pc10

By linearizing Eqs. (8) and (9) at the equilibrium

position (x=x0=0, Pc1=Pc10, Pc2=Pc20), the flow rate
variation at the nozzle-flapper can be obtained from the
displacement variation and the control pressure
variation as follows:

Conclusion

K i l f i (s ) + (Te s + 1)l f (K10 Pc1 (s ) K 20 Pc 2 (s ))

dGo 2
dPc 2

, ao2 =

(A3)

100092 wang 6/6

327

Copyright 2005 by JFPS, ISBN 4-931070-06-X