Professional Documents
Culture Documents
Power Electronic System Design Linking Differential Equations Linear Algebra and Implicit Functions Keng C Wu Full Chapter PDF
Power Electronic System Design Linking Differential Equations Linear Algebra and Implicit Functions Keng C Wu Full Chapter PDF
https://ebookmass.com/product/linear-algebra-and-partial-
differential-equations-t-veerarajan/
https://ebookmass.com/product/transforms-and-partial-
differential-equations-t-veerarajan/
https://ebookmass.com/product/first-course-in-differential-
equations-11ed-differential-equations-and-boundary-value-
problems-9ed-solutions-manual-dennis-zill-roberto-martinez/
https://ebookmass.com/product/mathematical-physics-with-
differential-equations-yisong-yang/
https://ebookmass.com/product/advanced-differential-
equations-1st-edition-youssef-raffoul/
https://ebookmass.com/product/schaums-outline-of-differential-
equations-fifth-edition-bronson/
https://ebookmass.com/product/power-system-analysis-and-
design-6th-edition-j-duncan-glover/
https://ebookmass.com/product/algebra-linear-2nd-edition-alfredo-
steinbruch/
POWER ELECTRONIC
SYSTEM DESIGN
Linking Differential
Equations, Linear Algebra,
and Implicit Functions
KENG C. WU
Switching Power, Inc. Ronkonkoma, NY, United States
POWER ELECTRONIC
SYSTEM DESIGN
Linking Differential
Equations, Linear Algebra,
and Implicit Functions
Elsevier
Radarweg 29, PO Box 211, 1000 AE Amsterdam, Netherlands
The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, United Kingdom 50
Hampshire Street, 5th Floor, Cambridge, MA 02139, United States
No part of this publication may be reproduced or transmitted in any form or by any means,
electronic or mechanical, including photocopying, recording, or any information storage
and retrieval system, without permission in writing from the publisher. Details on how to
seek permission, further information about the Publisher’s permissions policies and our
arrangements with organizations such as the Copyright Clearance Center and the
Copyright Licensing Agency, can be found at our website: www.elsevier.com/permissions.
This book and the individual contributions contained in it are protected under copyright
by the Publisher (other than as may be noted herein).
Notices
Knowledge and best practice in this field are constantly changing. As new research and
experience broaden our understanding, changes in research methods, professional practices,
or medical treatment may become necessary.
Practitioners and researchers must always rely on their own experience and knowledge in
evaluating and using any information, methods, compounds, or experiments described
herein. In using such information or methods they should be mindful of their own safety
and the safety of others, including parties for whom they have a professional responsibility.
To the fullest extent of the law, neither the Publisher nor the authors, contributors, or
editors, assume any liability for any injury and/or damage to persons or property as a matter
of products liability, negligence or otherwise, or from any use or operation of any methods,
products, instructions, or ideas contained in the material herein.
ISBN: 978-0-323-88542-3
2 First-order circuits 19
2.1 RC network with periodic drive source 19
2.2 Sawtooth (triangle ramp) generator 30
2.3 Full-wave rectifier with RC load 33
2.4 A brushless DC Motor with permanent magnets rotor 38
2.5 A BLDC motor speed detector 45
References 47
3 Current source 49
3.1 Semiconductor diode equation 49
3.2 Simple current source 50
3.3 Bob Widlar current source 54
3.4 Improved current source 58
3.5 Source impedance 60
3.6 555 timer 64
3.7 Precision current loop 70
3.8 Current-mode laser driver 74
3.9 LED array driver 76
3.10 JFET current source 77
3.11 MOSFET current source 78
vii
viii Contents
4 Second order 81
4.1 Form 81
4.2 Root 83
4.3 Time domain 85
4.4 Frequency domain 89
4.5 Parallel and serial resonance 92
4.6 Eigen value approach 103
4.7 RC filters and Sallen–Key filters 104
4.8 Power filters 111
4.9 Oscillator 113
4.10 Implicit function 120
xi
Preface
Years ago, Prof. Emeritus Chi-Tsong Chen, the author of Linear System
Theory and Design, a very successful textbook (Oxford University Press), met
the author at his Flushing, New York residence. In the meeting, and in the
preface of Signals and Systems – A Fresh Look his last publication (PDF form
free to all globally), Prof. Chen lamented that “Feedbacks from graduates
that what they learned in university is not used in industry prompted me to
ponder what to teach in signals and systems.”
Sadly,and based on long professional career serving RCA/GE/Lockheed
Martin space sector, the author can definitively confirm the fact Prof.
Chen was sad about. The less-than-desirable state had existed, and is still
present,in the form that many degree-holding engineers including electrical,
electronic, mechanical, and other specialties are falling short in applying
mathematical tools they were taught in college. Given electrical schematic
drawings, they were unable to formulate and express systems’ dynamics in
state variables and state transition using the first-order differential equations
and linear algebra technique. As a result, they were unable to boost their
productivity using software such as MATLAB.
This book intends to bridge the gap—what is taught in college and how
it is being applied in industry. In essence, this writing shall be considered
didactic.
It begins with Chapter one giving capacitors and inductors, two indis-
pensable energy storage components, an in-depth examination from the
view point of the first-order derivative, its corresponding integral form,
and its physical implications. Chapter two covers RC- and RL-type net-
works governed by a single differential equation. Key steps moving system
differential equations to Laplace transform in a frequency domain and to
a state-space transition form are introduced. Along the way, unconven-
tional approaches deriving Fourier series, explaining orthogonal property,
or treating boundary value problems are also explored. Chapter three covers
current sourcing circuits including current mirror, the workhorse of analog
integrated circuits, and precision current generator loops critical to instru-
mentation. Chapter four extends Chapter two to networks of second order
governed by two first-order differential equations. Procedures transforming
multiple differential equations to Laplace form, to state-transition form, and
to state-transition solution are shown. Chapter five examines circuit blocks
xiii
xiv Preface
was said true in the past may not be true in the future when new discoveries
see the daylight.
On the backdrop of the above conviction, this author took additional
efforts to make this writing also available in Chinese language;thanks to pub-
lisher Elsevier for granting such translation right. Thanks are also extended
to Mr. , at ITRI (Industrial Technology Research Institute, Hsinchu Taiwan),
who had performed the translation, a very demanding task considering the
limitations of Chinese language in handling technical subjects.
With the advance of miniaturized electronic hardware and supercom-
puter equipped with mathematical co-processors, engineering design tasks
are now mostly carried out by the simulation and computation. The
implementation of both always requires design formulation in the form
of analytical expressions based on, in most cases, systems of differential
equations with coefficients depending on components/parts values.
In the course of almost four decades‘Ł‘™professional career in aerospace
industries, the author had definitely derived significant benefits from follow-
ing the path outlined above.
You, readers, can certainly do the same.
Keng C. Wu
Princeton, NJ.
Dec. 2020
CHAPTER 1
i
dv(t ) +v
i(t ) C (1.1)
dt
Power electronic system design. Copyright © 2021 Elsevier Inc.
DOI: 10.1016/B978-0-32-388542-3.00004-2 All rights reserved. 1
2 Power electronic system design
v
i
t
t
v
By the same token, inductor and its electrical symbol was always intro-
duced by the following
i
di(t ) +v
v(t ) L (1.4)
dt
In this form, one important property of inductor stands out. That is, the
device allows a DC current, IDC , which however does not contribute to its
voltage.
3. The allowed DC current is however constrained within a limit; the
magnetic core saturation and winding wire Ampere rating.
i1 i2
v1 v2
N1 N2
B-H curve
0.4 0.4
0.2
B( H)
Tesla
0 0
−0.2
−0.4
−0.4
−400 −200 0 200 400
−400 H 400
Oersted
N1 i1 Aw1
λ1 = N1 1 = N1 Aw1 B = N1 Aw1 μoμr = N12 μoμr i1 = L1 i1
Lm Lm
(1.7)
where Aw1 stands for winding 1 cross-sectional area.
Under the same driving condition, Eq. (1.8), a cross-coupled flux linkage,
holds for winding 2, assuming Aw2 < Aw1 .
N1 i1 Aw2
λ2 = N2 2 = N2 Aw2 B = N2 Aw2 μoμr = N2 N1 μoμr i1
Lm Lm (1.8)
λ2 = M21 i1
Capacitor and inductor 7
B-H curve
0.4 0.4
0.2
B(H, 50)
Tesla 0
0
B(H, −50)
−0.2
−0.4
−0.4
−400 −200 0 200 400
−400 H 400
Oersted
Va
flux
−Va
Va Va
0
−Va −Va
(a) (b)
Fig. 1.8 Lumped model of real (a) capacitors and (b) inductors.
i1 M i2
v1 v2
N1 N2
1:1
i1 L 1 - M L2 - M i2
v1 M v2
Ideal
1:1
i1
(L1L2 - M 2 )/ i2
v1 M
v2
(L1L2 - M 2 )/ (L1L2 - M 2 )/ Ideal
(L2-M ) (L1-M )
Taking Laplace transform for the left most parts of Eqs. (1.21) and (1.22),
we reach a new equation set
v1 = sL1 I1 (s) + sMI2 (s) v2 = sMI1 (s) + sL2 I2 (s) (1.23)
Eq. (1.23) enables us to express I1 and I2 in terms of v1 and v2 .
v1 sM
v2 sL2 sL2 v1 − sMv2 (L2 − M )v1 + M (v1 − v2 )
I1 = = 2 =
sL1 sM s L 1 L2 − M 2 s L 1 L2 − M 2
sM sL2 (1.24)
v1 (v1 − v2 )
= +
−M
s L1 L2M−M
2 2
s L1LL22−M
sL1 v1
sM v2 sL1 v2 − sMv1 (L1 − M )v2 + M (v2 − v1 )
I1 = = 2 =
sL1 sM s L 1 L2 − M 2 s L 1 L2 − M 2
sM sL2 (1.25)
v2 (v2 − v1 )
= +
L1 L2 −M 2
s L1 L2M−M
2
s L1 −M
1:1
i1 (1-k 2)L 1 i2
v1 k 2L 1 v2
Ideal
Ac
Aw
Fig. 1.13 Half of a ferrite core; Ac = center post core area; Aw . (dotted line) = winding
window area, small filled circles = coil wires.
Here, Bmax is often chosen to be less than Bsat . Kf is a scaling factor relating
the RMS value of a periodic voltage and its time-domain magnitude.
Next, for each winding with peak current ij , a winding area (Aw )j is
conceptually assigned. Due to wire shape and unavoidable stacking in actual
build, only k(Aw )j is utilized. Given a desired inverse current density J [unit,
length2 /Amp],the winding peak current and its corresponding winding area
is associated by
k · (Aw ) j
ij = (1.28)
JN j
Eqs. (1.27) and (1.28) allow, for jth winding, its area product
(Vrms ) j JN j i j J (Vrms ) j i j
Ac (Aw ) j = · = (1.29)
K f f N j Bmax k kK f f Bmax
The total core cross-sectional area and winding area product covering all
winds is then
J
Ac A w = A c (Aw ) j = (Vrms ) j i j (1.30)
j
kK f f Bmax j
The summation on the right-hand side hints the total power handling
capacity of the device. The mathematical formulation may not be exactly
right, but it does give the flavor.
Therefore, referring to Fig. 1.13 and ignoring unit discrepancy, the
volumetric size of a magnetic device, in numerical term, may be considered
Capacitor and inductor 15
almost twice of Eq. (1.30). Designers must check with core manufacturers
as to the accounting of the area2 number to avoid over, or under, count;
therefore over, or under, sizing a device.
1.11.1 Capacitor
A c B–L m H curve
0.4
Weber(volt.second)=AcB
Ψ(H, 50∙500)
Ψ(H, −50∙500)
−0.4
−2×105 H 2×105
AmpereTurn=LmH
1.11.2 Inductor
The intricacy for specifying magnetic devices is, in this writer’s view, orders
of magnitude harder than that for capacitors. Here, we will begin with an
important reexamination of the B-H curves given in Fig. 1.4 and Fig. 1.5.
As far as this writer has been able to reach, the most existing literature
including journalistic articles and textbooks dealing with the subject cover
it solely in terms of B(flux density)-H(field strength) parameters. What was
not clearly mentioned in those presentations is the underlying significance
of presenting in B-H form. In a single statement, we proclaim that the
curve in B-H parameter form is “material specific.” It is characterizing a
specific ferrous material in terms of per unit volume. In other words, it is
independent of core geometry.
Therefore, those material property curves can be easily modified to be
“core specific,” Fig. 1.14, in which core geometry is now included; Ac core
cross-sectional area and Lm magnetic path length. The plot x-axis coordinate
is HLm (Ampere) while the y-axis is Ac B (flux, per turn, in Weber) (Readers
should ignore numerical figures in Fig. 1.14. It just shows that the inclusion
of core geometry will alter plot coordinate scales.)
By the same token, it can be further modified to be “device specific” or
“winding specific,” “terminal specific” in which the winding turn number,
N, is included. With that, the plot x-axis coordinate is Ni (Ampere turn)
while the y-axis is NAc B (total flux linkage in volt-second = ∫vdt).
In summary, three aspects are involved in specifying a magnetic device:
core material B-H, core geometry Ac B–HLm , and winding terminal NAc B-
Ni.
Capacitor and inductor 17
Keywords
Capacitor; Current lagging; Current leading; Differential form; Inductor; Integral form;
Voltage lagging; Voltage leading; Zero-crossing
CHAPTER 2
First-order circuits
In Chapter 1, specifically Fig. 1.8, capacitor and inductor with multiple
parasitic elements included exhibit second-order behavior. However, for
the frequency range, less than 500KHz, we are to deal with in this writing,
the serial inductance and leakage resistance for capacitor and the parallel
winding capacitance for inductor may be ignored. By so doing, both reduce
to a first order circuit. Much as a first-order sounds simple, its role in
stitching together inexpensive, component-saving applications cannot be
underestimated. The fact merely reflects a plain truth that simplicity is
equivalent neither to uselessness nor to ineffectiveness.
This chapter attempts to uphold the belief.
MATH. NOTE: Textbooks on differential equations and boundary value
problems define “linear first-order differential equations” as
F (x, y, y ) = 0 (M2.1)
where x is the independent variable, and dependent variable y(x) and its first
derivative y = dy/dx are in the form of the first degree; no term of square,
or higher, power {y2 , (y´)2 ..}.
By habitually using the symbol x and y, Eq. (M2.1) tends to lead readers
to the impression of dealing solely with geometry, or space. Of course,
“linear first-order differential equations” are not limited to dealing merely
with space variables. It can as well treat many other physical variables and
involve “time” as an independent variable. In other words, Eq. (M2.1) can
be modified to
F (t, x, x ) = 0 (M2.2)
where x is any time-dependent variable. In this writing, x mostly stands for
current, voltage, or power. ♣
0 T/2 T
a0 + ∑ (a
n =1.2..
n sin n 2Tp t + bn cos n 2Tp t )
1
1
2 0.5
sin( t )
0 0
0 1.57 3.14
0 t π
1
1
sin(t)∙cos(t)
0
0
−1 −1
0 3.14 6.28
0 t 2∙π
The fact, in reference to Fig. M2.2, also gives a very interesting geomet-
rical beauty: the sin2 θ , or cos2 θ , curve is also one of many curves bisecting
the 1 × π rectangle. Another view is: under the curve sin2 θ , above the curve
cos2 θ .
Anyway, if we extend the integration boundary, the following results:
2π π
(sin θ ) dθ = 2
2
(sin θ )2 dθ = π (M2.5)
0 0
or
2π
1
(sin θ )2 dθ = 1 (M2.6)
π 0
1
1
sin(t)∙cos(2·t)
0
0
−1 −1
0 3.14 6.28
0 t 2∙π
If we add more cosine terms at a higher frequency, the integral does not
change either; as a product of sin(t) and cos(nt) again generates equal positive
and negative area.
1 2π
sin θ + cos nθ · sin θ dθ = 1 (M2.9)
π 0 n=1,2..
sinusoid and an unknown, but periodic function under study has the effect
of accentuating the sinusoidal element embedded in the latter and with the
same frequency and phase.
The operation has a hidden effect, the suppression of elements that do
not match the template sin θ .
From a viewpoint of physical effects, the act of forming a product
and performing integration signifies the search for commonality among
two identities; one as a reference. Nonmatching, or nothing in common,
is orthogonal. Orthogonal of course means zero projection too. It also means
mutually inclusive. The presence of one does not exclude the other. They
also coexist simultaneously.
The above operation can be extended to find a cosine component hidden
in an unknown function. That is, if f(θ ) contains cos θ
2π
1
f (θ ) · cos θ dθ = 1 (M2.13)
π 0
2π
One may be tempted to think that π1 0 f (θ ) · (sin θ + cos θ ) dθ will
find both the sine and the cosine component simultaneously. But, it does
not work as the integral gives a single value. There is no indication how the
single value is partitioned among the two components.
The right way to find both the sine and the cosine component in a single
operation is
2π
1
f (θ ) · (cos θ − i sin θ )dθ (M2.14)
π 0
That is by utilizing the quadrature of the real and the imaginary parts of
a complex number.
The above is equivalent, and extended, to
2π
1
f (θ )e− jnθ dθ (M2.15)
π 0
v i(t)
R v o(t)
Va
C
DT T
Table 2.1 Input Fourier harmonic, the first 10, amplitudes for Fig. 2.1 drive source.
n= an = bn =
1 −0.895 2.76
2 0.729 0.525
3 −0.482 0.355
4 0.219 0.689
5 −3.516 ×10-3 −1.868 ×10-4
6 −0.147 0.457
7 0.207 0.147
8 −0.183 0.135
9 0.095 0.305
10 −3.518 × 10-3 −3.741 × 10-4
6
6
5
vi ( t ) 4
3
vir ( t )
2
0 1
0
−1 −1
0 t T
Fig. 2.2 Reconstructed drive source for Fig. 2.1 using Eq. (2.3); (Dot, drive source; solid,
reconstruction with limited terms).
10
vir (t ) = a0 + [an cos(nω0t ) + bn sin(nω0t )] (2.3)
n=1
Fig. 2.2 shows that Eq. (2.2) does decompose properly the rectangular
drive into a series of individual, contributing harmonics.
With that, we will next work out the output in a steady state.
First, the RC network gives a transfer function
1
H (s) = , H (s) = H ( jω) arg(H ( jω)) (2.4)
RCs + 1
Based on the theory for AC (alternating current) network analysis,
Fig. 2.1 output, vo (t) is the superposition, or sum (Eq. 2.5), of all individual
26 Power electronic system design
6
6
5
vi ( t ) 4
3
vo ( t )
2
0
1
0
−1 −1
0 t T
Fig. 2.3 Fig. 2.1 approximated output; (Dot, drive source; solid, output with 10 harmon-
ics).
4
4
3.97
vi ( t ) 3.93
3.9
vo ( t )
3.87
0
3.83
3.8
3.767 3.77
5.208×10−3 t 6.458×10−3
Fig. 2.4 Zoom in near the peak of Fig. 2.3; (Dot, drive source; solid, output with
10 harmonics).
m
vo (t ) = [an H ( jnωo ) cos(nω0t + arg(H ( jnωo )))
n=0 (2.5)
+ bn H ( jnωo ) sin(nω0t + arg(H ( jnωo )))]
6 4.08
6 4.083
5 4.03
vi ( t ) 4 vi ( t ) 3.98
3 3.92
vo ( t ) vo ( t )
2 3.87
0 0
1 3.82
0 3.76
−1 −1 3.708 3.71
0 t T 5.486×10−3 t 6.389×10−3
Fig. 2.5 Fig. 2.1 approximated output; (Dot, drive source; solid, output with 20 harmon-
ics); normal view and zoom in near peak.
The key however is not in the integration. Rather, it is the initial inte-
gration constant C(t0 ), the initial value/condition so to speak in mathematics
books. The fact is that the time-dependent initial value can move a solution
all over the solution domain.
Actually, the nomenclature “initial value” is, in this writer’s view, not
an ideal choice of word. First of all, the integration constant is not really
depending on time alone. It cannot simply come from thin air. It must come
from somewhere and depend on many elements constituting the real world.
It must be the end state of some past process. And it exerts influence on the
future course of another process. In a sense, physical states propagate from
one instant to the other and are continuous. Niagara Falls, or Grand Canyon,
is extremely steep. Still, it is continuous.
In summary, there appear two critical properties of physical world: The
continuous nature of state transitions and the distinctive state contents at the
transition boundary. That could be the reason the word “boundary value” has
been instead invoked.
Furthermore, boundary values exert influences on the evolution of
physical world by two pathways: space (geometry) and time. The latter
actually is equivalent to energy (frequency). The impact of space boundary is
easily observed in the extraordinary efforts put into shaping propeller blades
for aircraft, ship, submarine, wind turbine, etc., to improve performance
efficiency. However cancer treatments based on systemic, nonspecific, some-
time toxic, chemotherapy are beginning to be questioned; more targeted
and location-specific approaches utilizing energetic particle beams, forms of
energy boundary conditions, are becoming better choices.
(2.8)
vob (t ) = Vb0 e− RC [u(t − DT ) − u(t − T )]
t−DT
Va0 e− RC + Va 1 − e− RC = Vb0
DT DT
(1−D)T
(2.9)
Vb0 e− RC = Va0
Specifically, the end state of the first interval voltage at time DT serves as
the starting state for the dynamic output during the second time segment,
while the end state of the second interval voltage serves as the starting state
for the first interval voltage. This act, resembling passing the baton in a
10-km relay race (field and track competition), enables us to find both
starting conditions.
Va 1 − e− RC
DT
(1−D)T
Va0 = (1−D)T , Vb0 = Va0 e RC (2.10)
− DT
e RC − e RC
Once both cyclic starting condition are derived, Eq. (2.8) presents the
output in one cycle; sum of voa (t) and vob (t). If so desired for viewing purpose,
multiple cycle output is easily built by a periodic shift of a single cycle
expression.
m
vo1 (t ) = voa (t ) + vob (t ), vo (t ) = vo1 (t − nT ) (2.11)
n=0
For this current example, and using Eq. (2.10), Va0 = 1.677 Vb0 = 3.929.
Eqs. (2.8) and (2.10) then give Fig. 2.6.
Comparing Figs. 2.3–2.6 clearly shows the superiority of “continuity of
states and boundary condition approach.”
30 Power electronic system design
6 4.1
6 4.104
5 4.06
vi ( t ) 4 vi ( t ) 4.02
3 3.98
v( t) v( t)
2 3.94
0 0
1 3.9
0 3.85
−1 −1 3.813 3.81
0 t T 5.625×10−3 t 6.389×10−3
Fig. 2.6 Fig. 2.1 exact output; (Dot, drive source; solid, output); normal view and zoom in
near peak.
1
1
0.5
vi ( t )
0
iR ( t)
−0.5
−1 −1
0 t 1∙T
I
DT T
v o(t)
r
C
5
5
4
3
vo ( t )
2
1
0 0
0 5∙10−6 1∙10−5 1.5∙10−5 2∙10−5 2.5∙10−5 3∙10−5
0 t 3∙Ts
Eq. (2.15) yields r = 2.88 and I = 17 mA. Eq. (2.13) gives a sawtooth
output (Fig. 2.9).
This type of waveform served very important functions; in the CRT
(Cathode Ray Tube), now antiquated, display circuits as the beam scanning
controller and in pulse width modulation control for switch-mode power
converters and motor drivers. We will come to the latter later. ♠
It was mentioned in the previous MATH. NOTE that Eq. (M2.2) and
“di/dt + β r i = α r g(t)” or “dv/dt + β r v = α r g(t)” are the prevalent forms
of differential equations in this writing, and many other electric circuits
literature, rely on them. Much as the form looks simple, appearance can
be deadly deceiving. The trap lays in how the driving (source, generator)
function, g(t) on the right side, is specified.
There are more than one key points that must be properly observed,
otherwise attempts to solve the equation will end either in failure or in
error.
First of all, by sketching casually the driving function g(t) into a dif-
ferential equation without qualification, it seems to imply that the driving
function is valid starting at t = 0 and is actively applied all the time afterward.
These two notions are called into question.
Depending on the circumstance, the driving source time profile, the
properties of individual elements comprising the physical system, the cross-
interactions among elements, and the interaction between elements and
driving source, the active application of a driving source may start from
other than zero time. And, even more the important, the drive source may
kick in only sporadically.
Another random document with
no related content on Scribd:
would have been entirely lost had the two Russian admirals been
qualified for such a command. Captain Pélissier, who had served in
Holland, is said to have given Admiral Tchitchakov advice which he
ought to have followed, had he not been too obstinately attached to
his own opinions; Pélissier even pointed out to generals Suchtelen
and Soltikov the places where they ought to have erected their
batteries in order effectually to bar the egress of the Swedish fleet
from the bay; no attention, however, was paid to his advice. The
prince of Nassau-Siegen proved himself to be in no respect superior
as a commander to Tchitchakov. On the other hand, if the advice of
Duke Charles had been adopted, the Russians would have been
victorious without a battle; King Gustavus and Stedingk, however,
rescued the honour of the Swedish name.
The Swedes had now been closely shut up in the bay of Viborg for
three weeks, and at the end of June were reduced to extremities; in
the beginning of July a grand council of war was held. Duke Charles
and many other members of the council recommended a
capitulation, but the king and Stedingk were in favour of making a
desperate effort to force their way through the enemy’s line. The
attempt was accordingly made on the 3rd of July, and through
Tchitchakov’s neglect it was so far successful, as it enabled the
Swedish fleet to bring the blockading squadron to an engagement.
But the Swedes lost in it not only seven ships of the line, three
frigates, and more than thirty galleys and gunboats, but almost the
whole of the royal guards, the queen’s regiment, and that of Upland,
amounting to six thousand or seven thousand men, which had been
put on board the fleet. Whilst the larger Swedish ships thus
endeavoured to gain the open sea, the flotilla had withdrawn for
safety into an arm of the gulf, which runs parallel to the shore and
stretches towards Friedrichsham. This inlet, called the sound of
Suenske, is extremely difficult of access on the side towards
Friedrichsham, in consequence of a group of rocky islands at its
mouth, but it may be safely reached through the open harbour of
Asph. By this way the prince of Nassau-Siegen determined to pass
into the sound with the Russian flotilla, and attack the Swedes in
their place of refuge.
The latter were well protected from the attack of the Russian fleet
by rocks, and when the prince gave orders for the assault, on the
9th, the sailors were so exhausted and his orders for battle were so
unskilful that the king of Sweden gained a splendid victory on that
and the following day. The loss of the Russians was so great as to
have surpassed any which they had suffered since the Seven Years’
War. Fifty-five vessels were captured, a number of others destroyed,
and fourteen thousand Russians either taken prisoners or slain. In
spite of this signal victory, the king of Sweden now awoke from his
dream of humbling the pride and glory of Russia; already he began
to cast his eyes towards France, and in the following year he
dreamed his monarchical dream in favour of the French émigrés.
The idea of becoming the Godefroy de Bouillon of the aristocratic
and monarchical crusade, which Burke at that time proclaimed in the
English parliament and in his work on the French Revolution, had
been awakened in his mind in 1790, and the empress of Russia
found means of confirming him in his visionary projects. Moreover
his means were exhausted, and he therefore lent a favourable ear to
the proposal of Galvez, the Spanish ambassador, who began to
mediate for a peace between Sweden and Russia.
This peace, concluded at Varela on the Kimmene on the 14th of
August, 1790, served to show how empty all Gustavus’ splendour
was, and how unreal and inefficient were all the efforts he had made.
It was now seen that all the blood had been shed to no purpose, and
all the treasures of his very poor kingdom mischievously
squandered, for everything remained on the footing on which it had
been in the spring of 1788.
We now return to the war in which Austria and Russia were jointly
engaged against Turkey. The whole Austrian army was ready to take
the field at the end of the year 1787: it formed an immense cordon
stretching from the mountains on the coast of the Adriatic Sea to the
Carpathians, and consisted of a main body and five divisions.
Unhappily, the emperor Joseph was desirous of commanding the
main army in person, under the unskilful direction of Lacy, his military
Mentor, who, like his pupil Mack, was a good drill-sergeant, but no
general. The main body consisted of 25,000 infantry and 22,000
horse, and the whole of the troops together amounted to 86,000
cavalry and 245,000 foot, accompanied by 898 pieces of artillery.
In February, 1788, Russia and Austria had simultaneously
declared war against the Turks; but in August of that year England
and Prussia entered into an alliance, the main object of which was to
place Prussia in a situation to prevent the aggrandisement of Austria,
if necessary, by force of arms. This, however, was superfluous in
1788, because the diversion effected by the king of Sweden
prevented the Russians from proceeding with their usual rapidity,
and the emperor Joseph by his presence with the army frustrated the
effect of his immense armaments. The dissatisfaction with the whole
conduct of the war became so general that Joseph was at length
obliged earnestly to entreat Laudon, who had been the popular hero
of the Austrians since the time of the Seven Years’ War, and whom
the emperor had hitherto neither employed nor consulted, to assume
the command of the army in Croatia.