Chapter 17

Winding Capacitance and Leakage Inductance

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.

 Table of Contents

1. Introduction
2. Parasitic Effects
3. Leakage Flux
4. Minimizing Leakage Inductance
5. Winding Capacitance
6. Winding Capacitance Turn-to-Turn
7. Winding Capacitance Layer-to-Layer
8. Capacitance Winding-to-Winding
9. Stray Capacitance
10. References

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.

 Introduction
Operation of transformers at high frequencies presents unique design problems due to the increased
importance of core loss, leakage inductance, and winding capacitance. The design of high frequency power
converters is far less stringent than designing high frequency, wide-band audio transformers. Operating at
a single frequency requires fewer turns, and consequently, there is less leakage inductance and less
capacitance with which to deal. The equivalent circuit for a two-winding transformer is shown in Figure
17-1.
Lr
R

Figure 17-1. Equivalent Transformer Circuit.

High frequency designs require considerably more care in specifying the winding specification. This is
because physical orientation and spacing of the windings determine leakage inductance and winding
capacitance. Leakage inductance and capacitance are actually distributed throughout the winding in the
transformer. However, for simplicity, they are shown as lumped constants, in Figure 17-1. The leakage
inductance is represented by, Lp for the primary and, Ls for the secondary. The equivalent lumped
capacitance is represented by, Cp and Cs for the primary and secondary windings. The dc winding
resistance is, Rp, and Rs is for the equivalent resistance for the primary and secondary windings. Cw is the
equivalent lumped, winding-to-winding capacitance. R,, is the equivalent core-loss shunt resistance.

Parasitic Effects
The effects of leakage inductance on switching power supplies' circuits are shown in Figure 17-2. The
voltage spikes, shown in Figure 17-2, are caused by the stored energy in the leakage flux and will increase
with load. These spikes will always appear on the leading edge of the voltage switching waveform.

L,'(Leak
Energy = - [watt-seconds] [17-1]

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.

 Transformers designed for switching applications are normally designed to have minimum leakage
inductance, in order to minimize the voltage spikes, as shown in Figure 17-2. Also, leakage inductance can
be observed by the leading edge slope of the trapezoidal current waveform.

'on
Actual

Figure 17-2. Switching Transistor Voltage and Current Waveforms.

Transformers designed for power conversion are normally being driven with a square wave, characterized
by fast rise and fall times. This fast transition will generate high current spikes in the primary winding, due
to the parasitic capacitance in the transformer. These current spikes, shown in Figure 17-3, are caused by
the capacitance in the transformer; they will always appear on the lead edge of the current waveform and
always with the same amplitude, regardless of the load. This parasitic capacitance will be charged and
discharged every half cycle. Transformer leakage inductance and capacitance have an inverse relationship:
if you decrease the leakage inductance, you will increase the capacitance; if you decrease the capacitance,
you increase the leakage inductance. These are trade-offs that the power conversion engineer must make to
design the best transformer for the application.

T
P Light Load Heavy Load

Current Spike »•

^
2 3 4
ton ton
^
ton
•4 »- ton
-4 »•
^
Ideal Actual Ideal Actual

Figure 17-3. Transformer Capacitance Induced Current spike.

Leakage Flux

Leakage inductance is actually distributed throughout the windings of a transformer because of the flux set-
up by the primary winding, which does not link the secondary, thus giving rise to leakage inductance in
each winding without contributing to the mutual flux, as shown in Figure 17-4.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.

 However, for simplicity, leakage inductance is shown as a lumped constant in Figure 17-1, where the
leakage inductance is represented by Lp.

/
c=:^^^^^^=
^^^^
Leakage Flux — —*"Ls
-^ •s* -i
-r-^s r=- Primary
•^ c E5 ~
—
3
c
^ > Secondary
C >
EE Core ^ |=ii=^^^l
^

Figure 17-4. Leakage Flux.

In the layer-wound coil, a substantial reduction in leakage inductance, Lp and Ls, is obtained by
interweaving the primary and secondary windings. The standard transformer, with a single primary and
secondary winding, is shown in Figure 17-5, along with its leakage inductance, Equation [17-2]. Taking
the same transformer and splitting the secondary on either side of the primary will reduce the leakage
inductance, as shown in Figure 17-6, along with its leakage inductance, Equation [17-3]. The leakage
inductance can be reduced even more, by interleaving the primary and secondary, as shown in Figure 17-7,
along with its leakage inductance, Equation [17-4]. Transformers can also be constructed using the side-
by-side, sectionalized bobbin as shown in Figure 17-8, along with its leakage inductance, Equation [17-5].
The modified three section, side-by-side bobbin is shown in Figure 17-9, along with its leakage inductance
Equation [ 17-6].

Leakage flux
EE Core *- \
^ \ ^
££ t
Winding Area

Core Mating Surface— *- ->-*-

1
I
a = Winding length, cm
a b = Winding build, cm
Mutual Flux »-
Window — —<*- ^••••••••J
I1 \ '
c = Insulation thickness, cm

..., \
\v
H Secondary
Primary ~~*" J\ Insulation
-».

Figure 17-5. Conventional Transformer Configuration.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.

 EE Core »• ,- Primary
^"-
H~ i

|11
\Vinding Area |\
s
\
\ a = Winding length, cm
s,
Core Mating Surface —*- s a b = Winding build, cm
c = Insulation thickness, cm

W////M/A
Mutual Flux *-
Window — —\^*- 1EB^^^ \ _ Insulation
V \

b!-,
c
b2
c
•4 -b3 ^ N /2 Secondary each
s

Figure 17-6. Conventional, Transformer Configuration with Simple Interleave.

2
n(MLT\N
{ X
Lp= > " , [henrys] [17-3]

EE Core *- Np/2 Primary each

^

II
i

f^
a = Winding length, cm
Core Mating Surface —*-
\ a b = Winding build, cm
c = Insulation thickness, cm
Mutual Flux »•
\ Insulation
1
\

Window — —n*. l_^___£

\• V

Ib b b b \
Ns/2 Secondary each
c cc

Figure 17-7. Sectionalized, Transformer Configuration Primary and Secondary Interleave.

EE Core —*• ' Secondary

Bobbin ? \
/
y i
Winding Area • a
\r ± a = Winding length, cm
Core Mating Surface
Mutual Flux
Window
»•
-**•
s
< ... V..
I
1
f
=*=

^—
\> N
JT-
a
c

^ Insulation
x
Primary
b - Winding build, cm
c = Insulation thickness, cm

-H b
Figure 17-8. Pot Core, Sectionalized Transformer Configuration.

2
4x(MLT)N
V
^ " (l(T 9 ), [henrys] [17-5]

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.

 EE Core Secondary
Bobbin
Winding Area 0
a = Winding length, cm
Core Mating Surface a b = Winding build, cm
c = Insulation thickness, cm
Mutual Flux
Window Insulation
Np/2 Primary

Figure 17-9. Modified, Pot Core Sectionalized, Transformer Configuration.

7r(MLT}Nl(
V Za\ «A
Lp= ' ' Zc + — v(KT 9 ), [henrys] [17-6]
b I 3 j

Minimizing Leakage Inductance

Magnetic core geometry has a big influence on leakage inductance. To minimize leakage inductance, the
primary winding should be wound on a long bobbin, or tube, with the secondary wound as close as
possible, using a minimum of insulation. Magnetic cores can have identical rating, but one core will
provide a lower leakage inductance than the other. A simple comparison would be two cores with the same
window area, but one core has twice the winding length. Only half the winding build is shown in Figure
17-10.

EE Cores. Bobbin Secondary

\ / Bobbin
EE Cores
Primary /

Special Long Tongue EE Core Standard EE Core Configuration

Figure 17-10. Comparing a Standard EE Core and a Special Long Tongue Core.

If layers must be used, the only way to reduce the leakage inductance is to divide the primary winding into
sections, and then to sandwich the secondary winding between them, as shown in Figure 17-7. This can
pose a real problem when designing around the European VDE specification, because of the required
creepage distance and the minimum insulation requirements between the primary and secondary.
Minimizing the leakage inductance on a push-pull converter design could be a big problem. A special
consideration is required symmetry in both the leakage inductance and dc resistance; this is in order to get a
balanced winding for the primary switching circuit to function properly.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.

 The best way to minimize the leakage inductance, and to have a balanced dc resistance in a push-pull or
center-tapped winding, is to wind bifilar. Bifilar windings will drastically reduce leakage inductance. This
condition also exists on the secondary, when the secondary is a full-wave, center-tapped circuit. A bifilar
winding is a pair of insulated wires, wound simultaneously and contiguously, (i.e., close enough to touch
each other); Warning: do not use bifilar wire or the capacitance will go out of sight. Each wire constitutes a
winding; their proximity reduces leakage inductance by several orders of magnitude, more than ordinary
interleaving. This arrangement can be applied to the primary, to the secondary, or, it can be applied to the
primary and secondary together. This arrangement will provide the minimum leakage inductance.

Winding Capacitance

Operating at high frequency presents unique problems in the design of transformers to minimize the effect
of winding capacitance. Transformer winding capacitance is detrimental in three ways: (1) winding
capacitance can drive the transformer into premature resonance; (2) winding capacitance can produce large
primary current spikes when operating from a square wave source, (3) winding capacitance can produce
electrostatic coupling to other circuits.

When a transformer is operating, different voltage gradients arise almost everywhere. These voltage
gradients are caused by a large variety of capacitance throughout the transformer, due to the turns and how
they are placed throughout the transformer. When designing high frequency converters, there are several
factors that have a control over the turns: (1) the operating flux density or core loss; (2) the operating
voltage levels in the primary and secondary; (3) the primary inductance.

Keeping turns to a minimum will keep the capacitance to a minimum. This capacitance can be separated
into four categories: (1) capacitance between turns; (2) capacitance between layers; (3) capacitance
between windings; and (4) stray capacitance. The net effect of the capacitance is normally seen by the
lumped capacitance, Cp, on the primary, as shown in Figure 17-1. The lumped capacitance is very difficult
to calculate by itself. It is much easier to measure the primary inductance and the resonant frequency of the
transformer or inductor, as shown in Figure 17-11. Then, calculate the capacitance using Equation [17-7].
The test circuit, in Figure 17-11 functions as follows: The input voltage, VI, is held constant while
monitoring the voltage, V2, sweep through the frequency with the power oscillator. When the voltage, V2,
rises to a peak, and starts to decay at this peak voltage, the transformer or inductor is in resonance. At this
point the phase angle is also 0 degrees at resonance when looking at both the curves of VI and V2.

Ca = —~ = — — — , [farads] [17-7]
' '

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.

 Power Audio Oscillator Dual Channel Oscilloscope

Vert. #1, VI
Vert. #2, V2

UUT

Vert #2 Input Voltage

Figure 17-11. Circuit for Measuring either a Transformer or Inductor Self Resonates.

For transformers designed to operate with a square wave, such as dc-to-dc converter, leakage inductance,
Lp, and the lumped capacitance, Cp, should be kept to a minimum. This is because they cause overshoot
and oscillate, or ring, as shown in Figure 17-12. The overshoot oscillation, seen in Figure 17-12A, has a
resonant frequency, f, that is controlled by, Lp and Cp. This resonant frequency could change and change
drastically after potting, depending on the material and its dielectric constant, as shown Figure 17-12B.

(17-12-A) (17-12-B)

Figure 17-12. Primary Voltage with Leading Edge Ringing.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.

 Winding Capacitance Turn-to-Turn

The turn-to-turn capacitance, Cb, shown in Figure 17-13, should not be a problem if you are operating at
high frequency, low voltage power converters, due to the low number of turns. If the turn-to-turn
capacitance is important, then change the magnet wire insulation to one with a lower dielectric constant.
See Chapter 4.

<w> ' o
C
t

Primary
Cp
f
C
s~ -~^

t
f^ ~> N p
s s
cs
Secondary
J^
s~ ->
r» , r>

^
Figure 17-13. Capacitance Turn-to-Turn.

Winding Capacitance Layer-to-Layer

The capacitance between layers on the primary or secondary is the best contributor to the overall, lumped
capacitance, Cp. There are three ways to minimize the layer capacitance: (1) Divide the primary and
secondary windings into sections, and then sandwich the other winding between them, as shown in Figure
17-7; (2) The foldback winding technique, shown in Figure 17-14, is preferred to the normal U type
winding, even though it takes an extra step before starting the next layer. The foldback winding technique
will also reduce the voltage gradient between the end of the windings; (3) Increasing the amount of
insulation between windings will decrease the amount of capacitance. But remember, this will increase the
leakage inductance. If the capacitance is reduced, then the leakage inductance will go up. There is one
exception to this rule, and that is, if the windings are sandwiched or interleaved, it will reduce the winding
capacitance, but, it will increase the winding-to-winding capacitance.

Foldback Winding U Type Winding

Figure 17-14. Comparing the Foldback to the U Type Winding.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.

 Transformers and inductors wound on toroidal cores can have capacitance problems, just as much if care is
not taken in the design at the beginning. It is difficult to control the winding capacitance on a toroidal core
because of its odd configuration, but there are ways to control the windings and capacitance. The use of
tape barriers to mark a zone for windings, as shown in Figure 17-15, offers a good way to control this
capacitance.
Tape Barrier

Winding area

- Toroidal Core

Figure 17-15. Tape Barrier for Winding Toroidal Core.

Another way to help reduce the capacitance effect on toroids is to use the progressive winding technique.
The progressive winding technique example is shown in Figure 17-16 and 17-17: Wind 5 turns forward and
wind 4 turns back, then wind 10 turns forward and keep repeating this procedure until the winding is
complete.

Continuous winding Back wind

Core

Figure 17-16. Progress Winding Top View.

Expanded View
Back Wind

Figure 17-17. Progress Winding Side View.

Capacitance Winding-to-winding

Balanced windings are very important in keeping down noise and common mode signals that could lead to
in-circuit noise and instability problems later on. The capacitance, from winding-to-winding, shown in
Figure 17-18, can be reduced, by increasing the amount of insulation between windings. This will decrease
the amount of capacitance, but again, this will increase the leakage inductance. The capacitance effect

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.

 between windings can be reduced, without increasing the leakage inductance noticeably. This can be done,
by adding a Faraday Shield or screen, as shown in Figure 17-19, between primary and secondary windings.

A Faraday Shield is an electrostatic shield, usually made of copper foil. The Faraday Shield is normally
added along with the insulation between primary and secondary. In some designs, the Faraday Shield can
consist of three independent insulated shields or just one. It all depends on the required noise rejection.

Primary Secondary

Figure 17-18. Capacitance, Cw, Winding-to-Winding.

Primary Secondary

Shield #1 Shield #2

Figure 17-19. Transformer with a Primary and Secondary Shield.

Stray Capacitance
Stray capacitance is very important to minimize because it too, can generate asymmetry currents and could
lead to high common mode noise. Stray capacitance is similar to winding-to-winding capacitance except
that the capacitance is between the winding next to the core, Cc, and the outer winding next to the
surrounding circuitry, Cs, as shown in Figure 17-20. Stray capacitance can be minimized by using a
balanced winding, or using a copper shield over the entire winding. A means for measuring leakage current
is shown in Figure 17-21. The winding-to-winding capacitance can be calculated, using Equations [17-8]
and [17-9].

C = -, [farads] [17-9]

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.

 Core -Surrounding Circuitry

Electrostatic

Figure 17-20. Transformer Winding with Stray Capacitance.

Power Audio Oscillator Digital Voltmeter

Frequency

1111IIIIIIII11111111
Voltage

Figure 17-21. Test Circuit for Measuring Primary and Secondary, ac Leakage Current.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.

 References
1. Grossner, N., "Transformer for Electronic Circuits." McGraw-Hill, New York, 1967.
2. Landee, R., Davis, D., and Albecht, A., "Electronic Designer's Handbook," McGraw-Hill, New York,
1957, p. 17-12.
3. Lee, R., "Electronic Transformer and Circuits," 2nd ed., John Wiley & Sons, New York, 1958.
"Reference Data for Radio Engineers," 4th ed., International Telephone and Telegraph Co., New York.
4. Richardson, I., The Technique of Transformer Design, Electro-Technology, January 1961, pp. 58-67.
Flanagan, W., "Handbook of Transformer Application." McGraw-Hill, New York, 1986.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.