Switched Mode Power Supplies

by Dr.Ing. Heinz Schmidt-Walter

Switched mode power supplies (SMPS) are used in nearly all electronic systems. Every
television set and computer is powered by an SMPS as are many state-of-the-art industrial
equipment. Battery powered equipment also uses SMPSs to provide a constant internal supply
operating voltage independent of the state of charge of the battery. SMPSs are also used to
achieve a higher supply voltage compared to the powering battery voltage. This is normally
the requirement for tape recorders, CD players, note books mobile phones and cameras.

SMPS have remarkable advantages in compared to linear regulated power supplies.

Theoretically SMPS work loss-free and in practice efficiencies of about 70% to 95% are
achieved, this results in low temperature operation and consequently high reliability. The
other major advantage is that SMPS operate at a high frequency which results in small low
weight components. Compared to linear power supplies SMPS are therefore inherently more
efficient, smaller, lighter and cheaper to manufacture.

In general all SMPS operate by the same principle whereby packets of energy are taken from
an input voltage by an electronic switch (transistor) which switches at a high frequency.
Switching frequencies are normally in the range of 20kHz to 300kHz, depending on the
required performance. The conditions between turn-on and turn-off time of the switch
determines the average energy flow. A low pass filter is placed at the output of all SMPSs to
smooth the discontinuous energy flow. The high efficiency of SMPSs is a direct result of the
theoretically loss free switching component and low pass filter.

There are a number of different types of switch SMPSs which are described below.

SMPSs can be configured as secondary or as primary switched power supplies. Secondary

switched power supplies have no isolation between input and output. They are used in
applications where isolation in respect to the mains already exists or where isolation is not
required, for example in battery supplied devices. Primary switched power supplies offer an
isolation between input and output, their switching transistors operate on the primary side of a
transformer. The energy is be transfered to the secondary side at a high frequency via a high
frequency transformer. The transformer can be relatively small, because of its high operating
frequency.

There are three basic SMPS configurations which are: flyback, forward and
resonant-converters. Flyback converters transfer their energy during the off-time of the
transistors. Forward converters transfer their energy during the on-time of the transistors.
Resonant converters use a resonant circuit for switching the transistors when they are at the
zero current or zero voltage point, this reduces the stress on the switching transistors.

A power factor pre-regulator is also a SMPS and is used to ensure that the mains current is
substantially sinusoidal.
 Page 2

Overview: Switch mode power supplies

Buck converter
V out ≤ V in
Vin Vout short-circuit and no load proof simply achievable
V GS has to float
Usage: Repacement of analoge voltage regulators

Boost converter
V out ≥ V in
Vin Vout
Not short cicuit proof
Not no load proof if not operating in a closed loop
Usage: Battery supplied devices as notebooks,
mobilphones, camera flashesPhotoblitze

Inverting converter
V out < 0V
Vin Vout short-circuit proof simply achievable
Not no load proof if not operating in a closed loop
usage: Achieve of a negative voltage out of a positive

Flyback converter
Several, isolated output voltages, regulated by one control
circuit, achievable
Vout1
Power up to some 100W
Vin
Wide range for input and output voltage (mains voltage
85...270VAC achievable)
Vout2
Transistor breakdown voltage V DS ≥ 2V in
Very good magnetic coupling necessary
Big core with air gap necessary

Single transistor forward converter

Only one output voltage
Vout Output power up to several 100W
Vin Transistor breakdown voltage V DS ≥ 2V in
Duty cycle onT ≤ 0, 5
t

Very good magnetic coupling necessary

Small core without an air gap

Two transistor forward converter

Only one output voltage
Vin Output power up to some kW
Vout Transistor breakdown voltage V DS = V in
Duty cycle onT ≤ 0, 5
t

Small core without an air gap

Not an extaordinary magnetic coupling necessary
 Page 3

Full-bridge push-pull converter

Only one output voltage
Output power up to many kW
Vin Vout Transistor breakdown voltage V DS = V in
Small core without an air gap
Not an extaordinary magnetic coupling necessary
Balancing problems

Half-bridge push-pull converter

Only one output voltage
Output power up to some kW
Vin
Transistor breakdown voltage V DS = V in
Vout
Small core without an air gap
Not an extaordinary magnetic coupling necessary
Balancing problems

ZCS push-pull resonant converter

Several, isolated output voltages achievable
Output power up to several kW
L
Transistor breakdown voltage V DS = V in
Vin Vout
Small core without an air gap
Not an extaordinary magnetic coupling necessary
Control with fixed pulse duration and variable frequency
C
If the output power is low compared with the rated power,
the frequency can be audible
 Seite 4

Buck-converter

The Buck-converter converts an input voltage into a lower output voltage, it is also called
step-down converter.

Iin T IL L Iout
+ + +
Vin VL
VPWM D V1 Vout Load
Cin Cout

Figure1.1.1: Buck-converter

Figure 1.1.1 shows the circuit diagram of a Buck-converter. The transistor T operates as the
switch, which is turned on and off by a pulse width modulated control voltage V PWM with
t
high frequency. The ratio 1 between on-time t 1 to the period time T is called the duty
T
cycle.

Vcont

t
t1 T
V1 Vin
Vout = V1
t
VL (Vin-Vout)

t
-Vout
∆ IL
IL Iout = I L

Figure 1.1.2: voltages and currents of the Buck-converter

In the following analysis it will be assumed that the conducting voltage drop of the transistor
and the diode is zero.

During the on-time of the transistor the voltage V 1 is equal to V in . When the transistor
switches off (blocking phase) the inductor L continues to drive the current through the load in
parallel with C out and the diode, consequently the voltage V 1 becomes zero. The voltage V 1
stays at zero during the off-time of the transistor provided that the current I L does not reduce
to zero. This mode of operation is called continous mode. In this mode V 1 is a voltage which
changes between V in and zero, corresponding to the duty cycle of V cont ,(see Figure 1.1.2).
 Seite 5

The low-pass filter, formed by L und C out , produces an average value of V 1 , i.e. V out = V 1 ,
therefore for continuous mode:

t1
V out = V in
T

For the continuous mode the output voltage is a function of the duty cycle and the input
voltage, it is independent on the load.

The inductor current I L has a triangle shape, its average value is determined by the load. The
peak-to-peak current ripple ∆I L is dependent on L and can be calculated with the help of
Faraday's Law:

V = L di → ∆i = 1 ⋅ V ⋅ ∆t → ∆I L = 1 (V in − V out ) ⋅ t 1 = 1 V out (T − t 1 )
dt L L L
t1
For V out = V in and a switching frequency f it follows that for the continous mode:
T

V
∆I L = 1 (V in − V out ) ⋅ out ⋅ 1
L V in f

The current ripple ∆I L is independent of the load.

The average of the current I L is equal to the output current I out .

∆I L
At low load current, in case that I out ≤ , the current I L becomes zero in every switching
2
cycle. This mode is called discontinous mode and for this mode the calculations above are
not valid.

Calculation of L and C out :

To the calculate the value of L a realistic value of ∆I L has to be selected. The problem is as
follows: If ∆I L is selected at a very low value, the value of L has to be relatively high and
this would require ae very heavy and expensive inductor. If ∆I L is selected at very high level
the switch-off current of the transistor would be very high (this would result in high losses in
the transistor). A good and usual compromise between these effects is: ∆I L ≈ 0.2 I out

For L it follows:

V
L = 1 (V in − V out ) ⋅ out ⋅ 1
∆I L V in f

The maximum value of the inductor current is:

 Seite 6

I L = I out + 12 ∆I L

Assuming that the inductor ripple current is small compared to its dc current the RMS value
of the current flowing through the inductor is given by:

I L(RMS) ≈ I out

The capacitor C out is chosen usually for a cut-off frequency of the LC out -low-pass filter,
which is approximately 100 to 1000 times lower than the switching frequency. An exact
calculation of the capacitor depends on its maximum rating of the AC current and its serial
equivalent impedance Z max , both can be verified from the relevant data sheet.
The current ripple ∆I L causes a voltage ripple ∆V out at the output capacitor C out . For normal
switching frequencies this voltage ripple is determined by the equivalent impedance Z max .

The output voltage ripple is given by Ohm's law:

∆V out ≈ ∆I L ⋅ Z max

The choice of the output capacitor depends not on its capacitance, but on its series equivalent
impedance Z max at the switching frequency which can be verified from the capacitor data
sheet.
 page 7

Boost converter

The boost converter converts an input voltage to a higher output voltage. The boost
converter is also called a step-up converter.
Boost converters are used in battery powered devices, where the electronic circuit requires a
higher operating voltage than the battery can supply, e.g. notebooks, mobile phones and
camera-flashes.

Iin IL L D Iout
+ + +
Vin VL T Vout Load
V1
Cin Vcont Cout

Figure 1.2.1: Boost converter

Figure 1.2.1 shows the basic circuit diagram of the boost converter. The transistor T operates
as a switch, which is turned on and off by a pulse-width-modulated control voltage V cont .

Vcont

t
t1 T
V1
Vout

t
VL
Vin
t
-(Vout-Vin)
∆ IL
IL Iin = IL

t
ID

Iout = I D
t

Figure 1.2.2: voltages and currents of the boost converter

In the following analysis it will be assumed that the conducting voltage drop of the transistor
and the diode is zero.

During the on-time of the transistor, the voltage across L is equal to V in and the current I L
increases linearly. When the transistor is turned off, the current I L flows through the diode
and charges the output capacitor.
The function of the boost converter can also be described in terms of energy balance. During
the on-time of the transistor the inductance is charged with energy and during the off-time of
 page 8

the transistor this energy is transfered from the inductor through the diode to the output
capacitor.

If the transistor switch is not turned on and off by the clock the output capacitor charges via L
and D to the level V out = V in . When the transistor is switched the output voltage will increase
to higher levels than the input voltage.

As with the buck converter (see chapter 1.1: "buck converter") discontinous and continous
mode is dependent on whether the inductor current I L reduces to zero during the off-time of
the transistor.

With the help of Faraday's Law the continous mode and steady state conditions (see also
figure1.2.2) can be established:

∆I L = L1 V in ⋅ t 1 = L1 (V out − V in ) ⋅ (T − t 1 ) .
From which:

V out = V in T
T − t1

For the continous mode the output voltage is a function of the duty cycle and the input
voltage, it is independent of the load.

The boost converter is not short circuit proof, because there is inherently no switch-off
device in the short-circuit path.

NOTE:
If the boost converter is not regulated in a closed loop but is controlled by a fixed
duty cycle of a pulse generator (this could be the case for a laboratory set up), the
boost converter is not no-load proof. This is because each switching cycle results in
energy in the choking coil being tranfered to the output capacitor. This will result in
the output voltage continously increasing until the devices are eventually destoyed.

Calculation of L and C out :

As with the buck converter the starting point of calculating L is to select a current ripple ∆I L
of about 20% of the input current: ∆I L ≈ 0.2 I in . The input current I in can be calculated by
assuming zero losses (input power = output power), therefore:

V out
V in ⋅ I in = V out ⋅ I out → I in = I out
V in
L can be calculated as follows:

V
L = 1 (V out − V in ) in ⋅ 1
∆I L V out f

The peak value of the inductor current is (Ref: Fig. 1.2.2):

 page 9

I L = I in + 12 ∆I L .

Assuming that the inductor ripple current is small compared to its dc current the RMS value
of the current flowing through the inductor is given by:

I L(RMS) ≈ I in

The output capacitor is charged by pulses (Ref Fig. 1.2.2). The ripple ∆V out of the output
voltage results from the pulsating charge current I D and is mainly determined by the
impedance Z max at the switching frequency of capacitor C out . Z max can be verified from the
the capacitor data sheet.

The output voltage ripple is given by Ohm's law:

∆V out ≈ I D ⋅ Z max
 page 10

Buck-Boost Converter

The buck-boost converter converts a positive input voltage to a negative output voltage.
Iin T D Iout
+
Load
Vin Vcont L VL Cout Vout
Cin
IL +

Figure 1.3.1: Buck-Boost converter

Fig 1.3.1 shows the basic circuit of the buck-boost converter. The transistor T works as a
switch, which is turned on and off by the pulse-width-modulated voltage V cont . During the
on-time of the transistor, the inductor current I L increases linearly. During the off-time the
current I L continous and charges the output capacitor C out . Note the polarity of the output
voltage in Fig1.3.1.

For the continous mode and steady state conditions the output voltage is given by:

t1
V out = V in
T − t1

The inductor current I L is given by:

I L = I out T = I out  out + 1 

V V V
and ∆I L = 1 V in t 1 = 1 ⋅ in out ⋅ 1
T − t1 V in L L V in + V out f

Vcont

t
t1 T
VL
Vin
t
-Vout
∆ IL
IL
IL

t
ID

Iout = I D
t
Figure 1.3.2: Voltages and currents of the buck-boost-converter
 page 11

Flyback converter

The Flyback converter belongs to the primary switched converter family, which means there
is isolation between in and output. Flyback converters are used in nearly all mains supplied
electronic equipment for low power consumption, up to approximately 300W. Examples of
which are televisions, personal computers, printers, etc..

Flyback converters have a remarkably low number of components compared to other SMPSs,
they also have the advantage that several isolated output voltages can be regulated by one
control circuit.

L1, L2,
I1 N1 N2
+ +
V1 V2 Vout Load
Cout
Vin Cin

VDS
Vcont

Figure 2.1.1: Flyback converter

Fig. 2.1.1 shows the basic circuit of a flyback converter. The transistor works as a switch,
which is turned on and off by the pulse-width-modulated control voltage V cont . During the
on-time of the transistor the primary voltage of the transformer V 1 is equal to the input
voltage V in which results in the current I 1 increasing linearly. During this phase, energy is
stored in the transformer core. During the on-phase the secondary current is zero, because the
diode is blocking. When the transistor is turned off the primary current I 1 is interrupted and
the voltages at the transformer invert due to Faraday's Law (v = L didt
), the diode conducts and
the energy moves from the transformer core via the diode to the output capacitor C out .

During the on-phase of the transistor the drain-source voltage V DS is equal to zero. During
the off-time of the transistor, the output voltage V out will be transformed back to the primary
side and the drain-source voltage theoretically steps up to V DS = V in + V out ⋅ N 12 . If a mains
N

voltage of 230V/50Hz is used V DS will jump up to approximately 700V. In practice this

voltage will be even higher due to the self induction of the leakage inductance of the
transformer. To allow for this effect the minimum rated drain-source breakdown voltage of
the transistor must be 800V.

The transformer is not a "normal" transformer, because its function is to store energy during
the on-time of the transistor and to deliver this energy during the off-time via the diode to the
output capacitor. In effect the transformer is a storage inductor (often called a choke) with a
primary and secondary winding. To store energy the transformer core needs an air gap
(normal transformers do not have an air gap). An important consideration for this transformer
is, that primary and secondary windings are closely coupled to achieve a minimum leakage
inductance. It should be noted that the energy of leakage inductance cannot be transfered to
the secondary side and is therefore disipated as heat on the primary side.
 page 12

Vcont

t1 T
t
N1
VDS Vin+Vout( )
N2

t
V1 Vin

t
-Vout N1
I1 N2
∆ I1
t

I2 Î2
∆ I2
t
secondary current calculated to the primary side:
IL N2
I1 I2
N1 IL = I1+I2 N2
∆ IL N1
t

Figure 2.1.2: Voltages and currents at the flyback converter

Design of the flyback converter:

For the primary voltage of the transformer V 1 the average V 1 must be equal to zero for steady
state conditions (if not, the current will increase to infinity).
N
This leads to: V in ⋅ t 1 = V out ⋅ 1 ⋅ (T − t 1 ) and:
N2

N2 t1
V out = V in ⋅ ⋅
N1 T − t1

The turns ratio of the transformer should be choosen so that for the rated output power the
on-time (energy charge time) t 1 is equal to the off-time (energy discharge time) T − t 1 . This
leads to the turns ratio:

N 1 V in
=
N 2 V out
 page 13

The breakdown voltage of the transistor and the reverse voltage of the diode must be for this
case:
Transistor: V DS = V in + V out ⋅ N 12 ≈ 2V in
N

V R = V out + V in ⋅ N 21 ≈ 2V out
N
Diode:

It should be noted that the rated breakdown voltage of the transistor must be chosen
significantly higher, because at the turn-off instant the energy of the leakage inductance L s
will not be taken over by the secondary winding. To keep the overvoltage in an acceptable
range a snubber circuit is required, see Fig 2.1.3. At the instant of turn-off the current of the
leakage inductance L s is diverted through by the diode D and charges the capacitor C . The
power is dissipated in resistor R .
If R and C are required to operate at 230VAC,, a value of R has to be determined
experimentally to ensure that the dc voltage across C falls within the region of 350V to
400V.

C R D
Ls

+ +
Vout
Vin

Figure 2.1.3: Snubber circuit to limit the peak voltage across the transistor

To design the transformer the primary inductance L 1 has to be calculated first.

L 1 has to store energy during the on-time of the transistor, which is the energy required at the
output. This energy is given by: W = P out ⋅ T , where T is the periodic time of the switching
frequency and P out is the rated power. This energy is stored in the primary inductance during
the first half of the period time and is transfered to the output capacitor during the second half
of the switching period. As before the switching period is divided into two equal parts, one
part to store the energy and the other part to transfer the energy.

During the on-time of the transistor the voltage across the primary inductance is equal to V in
and the current I 1 is a ramp waveform. For every cycle of the input energy it follows that :

I1 T
W = V in (see Fig. 2.1.4)
22

This energy is stored in L 1 and can be calculated as:

W = 1 L 1 I 12
2
For the size of the primary inductance this leads to:

V 2in
L1 ≈ .
8 P out ⋅ f
 page 14

The calculation above assumes an efficiency of 100 %. If we consider an efficiency of η , it

means that we have to store more energy in L 1 and not all of this energy is delivered to the
output, then L 1 can be calculated as follows:

V 2in
L1 ≈ ⋅η
8 P out ⋅ f

η has to be estimated because its value is not known at this point of calculation. (η ≈ 0.75 is
normally a good estimate.)

I1
Î1

t
T/2 T
Fig. 2.1.4: Shape of the input current I 1 for rated power

4 ⋅ P out
The peak value of the current I 1 is: I1 =
V in ⋅ η
I
The RMS-value of the current I 1 is: I 1RMS = 1
6
The core of the transformer and the windings can now be calculated with the help of Chapter
5: "Calculation of inductors and high frequency transformers"

The output capacitor C out is charged by pulses (Ref Fig. 2.1.2). The ripple ∆V out of the output
voltage results from the pulsating charge current I 2 and is mainly determined by the
impedance Z max of the capacitor . Z max can be verified from the capacitor data sheet.

The magnitude of the ripple voltage is given as follows:

∆V out ≈ I 2 ⋅ Z max

The input capacitor C in can be calculated for 230V/50Hz-mains as follows:

µF
C in ≈ 1 ⋅ P in
W

A special feature of the flyback converter is the possilbility of controlling several isolated
output voltages with only one control circuit (Fig. 2.1.5).
 page 15

+
+
N2 Vout2
N1
Vin
+
N3 Vout3

control circuit

Fig. 2.1.5: Flyback converter for several output voltages

One output voltage is regulated (in Fig. 2.1.5 V out 3 ). Voltage V out 2 is coupled to V out 3 via the
V N
turns ratio: out2 = 2. The energy which is stored in L 1 (N 1 ) during the on-time of the
V out3 N 3
transistor moves during the off-time to the outputs. These output voltages maintain their
values in relationship to the turns ratio. The output voltages in relation to the the turns ratio
from the primary side appear to be in parallel. Therefore the energy from the primary side
transfers to the output where the lowest voltage appears.
 Page 17

Single Transistor Forward Converter

The single transistor forward converter belongs to the primary switched converter family
since there is isolation between input and output. It is suitable for output powers up to 1kW.
The single transistor forward converter is also called a single ended forward converter.

N1 N1' N2 D2 L I3 Ia
+ I1 I1' I2 +
Vout Load
V1 V2 V3 Cout
Vin D3
Cin

Vcont D1

Figure 2.2.1: Single transistor forward converter

The forward converter transfers the energy during the on-time of the transistor. During this
time the voltage V 1 is equal to the input voltage. The winding N 2 is in the same direction as
N
N 1 . When the transistor is on voltage V 2 at N 2 is given by V 2 = V in 2 . The voltage V 2
N1
drives the current I 2 through the diode D 2 which during this time is equal to I 3 through L
which charges the output capacitor C out .

Vcont

t1 T t
V1
Vin

t
-Vin
V3 N2
Vin N1
t

I1 } ∆ I1 ∆ I1: magnetizing-
} ∆ I3 N2
N1
current

t
I1'
∆ I1 }
t
∆ I3
I3 ID2 ID3
Iout = I3

Fig. 2.2.2: voltages and currents at the single transistor forward converter

During the off-time of the transitor, N 1 and N 2 are without current. The inductor L drawes its
current through the diode D 3 . The value of the voltage V 3 is equal to zero (neglecting the
forward voltage drop of D 3 ).
 Page 18

During the off-time of the transistor, the magnetic flux of the transformer has to reduce to
zero. The core is demagnetized with N 1 via D 1 to V in . N 1 has the same number of turns as
N 1 therefore the demagnetization needs an equal time intervall as the on-time. For this
reason the minimum off-time has to be as long as the on-time. This causes a maximum duty
cycle t 1 /T of 0.5 for the single transistor forward converter.

During the off-time, the voltage at N 1 is equal to the input voltage V in . This voltage will be
transformed back to the primary winding N 1 and for V 1 follows: V 1 = −V in . Due to this the
drain source voltage steps up to V DS ≥ 2V in when the transistor is turned off.

In comparison to the transformer of the flyback converter the transformer in this forward
converter is a "normal" transformer. Its job is not to store energy but to transfer energy. For
this reason the core has no air gap.

The breakdown voltage of the transistor has to be V DS > 2V in .

The windings N 1 and N 1 must be closely coupled. However a snubber circuit as shown
in Fig. 2.1.3, chapter 2.1. "flyback converter" is necessary.

In comparison to the flyback converter, the forward converter can only have one
regulated output voltage.

t1
The maximum duty cycle is = 0.5 .
T

Design of the single transistor forward converter:

The output voltage V out is equal to average of V 3 . The maximum duty cycle is 0.5.This leads
to (see also Chap. 1.1: "buck converter"):

N2 t1
V out = V in ⋅ ⋅
N1 T

For the turns ratio follows:

N2 V
= 2 ⋅ out and N 1 = N 1
N1 V in

For further calculation of the transformer see Chapter 5: "Inductors and high frequency
transformers"

To calculate L the method used for the buck converter is appropriate. Initially the current
ripple ∆I 3 of the inductor current I 3 has to be selected. A value for it is 20% of the output
current is normally acceptable: ∆I 3 ≈ 0, 2 ⋅ I out . Assuming a maximum duty cycle of 0.5, this
leads to:
 Page 19

V out ⋅ T/2
L=
∆I 3
The value of C out depends on the acceptable voltage ripple ∆V out of the output voltage. This
voltage ripple is mainly determined by the impedance Z max of the output capacitor C out :

∆V out ≈ ∆I L ⋅ Z max

Z max can be verified from the datasheet of C out .

The input capacitor C in for 230V/50Hz-mains should be:

µF
C in ≈ 1 ⋅ P in
W

2.2.1 Two-Transistor forward converter:

The two-transistor forward converter is a variant of the single transistor forward converter.

+ D2 N1 N2
T1 L Iout
+
Vout Load
Vin

D1 T2

Fig. 2.2.3: two-transistor forward converter

The transistors T 1 and T 2 are switching at the same time. During the on-time of the
transistors, the voltage at the primary winding is equal to the input voltage V in . During the
off-time of the transistors the transformer will be demagnetized via the diodes D 1 and D 2
into the input voltage V in . In comparison to the single transistor forward converter this
converter has the advantage that its transistors only have to block the input voltage and the
winding N 1 is not required. In addition to this the coupling of the transformer windings is no
longer critical. These advantages make this converter type suitable for significantly higher
output powers compared to the single transistor converter.

The calculation of the components is equivalent to the single transistor forward converter.

For the two transistor forward converter the breakdown voltage of the transistors is only
required to be V DS = V in .

The two transistor forward converter can be used for powers up to a few kWs. It is a
simple converter, which is not critical in regard to its physical design and its electrical
operation.
 Page 20

Push-Pull Converters

The push-pull converter can be designed for high power..

N1 N2 I3 L Iout
+
T1 T3 +

Vin V1 V2 V3 Vout Load

Cout
Cin
T2 T4

Fig. 2.3.1: Push-pull converter, here: full-bridge typ

V control of T1 and T4
GS

t
V control of T2 and T3
GS

t1 T t
V1
Vin

t
-Vin
V3
N2
Vin
N1 Vout = V3
t
I1

I3 ∆ I3
Iout = I3

t
Fig. 2.3.2: Voltages and currents at the push-pull converter

The push-pull converter drives the high frequency transformer with an AC-voltage, where the
negative as well as the positive half swing transferes energy. The primary voltage V 1 can be
+V in , −V in or zero depending on which pair of transistors (T 1 , T 4 or T 2 , T 3 ) are turned on or
off. At the secondary side the AC-voltage is rectified and smoothed by L und C out .

For continious mode follows (see also Chapter 1.1 "buck converter"):
 Page 21

N2 t1
V out = V in ⋅ ⋅
N1 T

t
The duty cycle T1 may theoretically increase to100%. This is not possible in practice because
the serial connected transistors T 1 , T 2 or T 3 , T 4 have to be switched with a time difference to
avoid a short-circuit of the input supply. The turns ratio of the transformer has to be:

N 2 V out
≥
N 1 V in

The transistors of the push-pull converter can be switched with the maximum duty
t
cycle of 0.5. This leads to the maximum duty cycle of 1 = 1 after rectification.
T

The calculation of L and C out follows those of the buck converter (chapter 1.1).

2.3.1 Half-Bridge Push-Pull Converter:

N1 N2 L Iout
+
T1 C1 +
V2
Vin V1 Vout Load
Cout
V2
T2 C2

Fig. 2.3.3: Half-bridge push-pull converter with full-wave rectifier

A variant of the push-pull converter is the half-bridge push-pull converter. The capacitors
C 1 and C 2 divide the input voltage V in into two. Therfore the magnitude of the primary
voltage is ±V in /2 . In comparison to the full-bridge push-pull converter follows for the
N V
half-bridge typ the turns ratio of the transformer to: 2 ≥ 2 out .
N1 V in
NOTE:
In Fig. 2.3.3 a two diode full-wave rectifier is used instead of a full-wave bridge
rectifier. The choice of rectifier type is dependent on the output voltage and current. The
difference between these two rectifier types is, that the current has to pass through two
diodes in the bridge type and only one diode in the full-wave type. Consequently the
full-wave type is used for high current to reduce the rectifier losses and the bridge type
is used for high voltage purpose to save one secondary winding of the transformer.
 Page 22

Resonant Converters

Resonant converters use a resonant circuit for switching the transistors when they are at the
zero current or zero voltage point, this reduces the stress on the switching transistors and the
radio interference. We distinguish between ZVS- and ZCS-resonant converters (ZVS: Zero
Voltage Switching, ZCS: Zero Current Switching).

To control the output voltage, resonant converters are driven with a constant pulse duration at
a variable frequency. The pulse duration is required to be equal to half of the resonant period
time for switching at the zero-crossing points of current or voltage.

There are many different types of resonant converters. For example the resonant circuit can be
placed at the primary or secondary side of the transformer. Another alternative is that a serial
or parallel resonant circuit can be used, depending on whether it is required to turn off the
transistor, when the current is zero or the voltage is zero.

The technique of resonant converters is described below giving the ZCS-push-pull resonant
converter as an example.

2.4.1 ZCS-Push-Pull Resonant Converter:

+
D1
T1
V'out
I L
Vin
Cin
D2 C
T2 Tr Vc

+
Vout Load
Cout

control circuit: constant pulse duration, variable frequency

Fig. 2.4.1: The ZCS-push-pull resonant converter

The converter works as follows:

The resonant circuit is formed by L and C . Assume an initial condition of the voltage V C
across C equal to zero. If now the transistor T 1 is turned on, a sinusoidal current half-swing
starts through T 1 , L, Tr, C and C in . This half-swing charges the capacitor C from zero to V in .
If this first half sinusoidal swing is finished, T 1 can be switched off without losses and after a
 Page 23

short delay T 2 can be switched and a next half sinusoidal swing starts, this discharges C
from V in back to zero Volts.
Every half sinusoidal swing transfers a certain amount of energy from the primary to the
secondary side of the transformer. The transformer Tr operates on its primary side as a
voltage source. For the duration of the current swing through the primary winding, the output
voltage V out will be transformed to the primary side: V out = V out N 21 . The energy which is
N

transfered by every half-swing is equal to W = V out ⋅ ∫ i(t) dt . This energy will be transfered
twice in each resonant period. This leads to the output power being given by
P out = W ⋅ 2 f switching (f switching : frequency of the converter). Fig. 2.4.2 shows an equivalent
circuit for one half-swing.

I C
S L V'out Î = (Vin -V'out)
L
I
+ T1 =Vout N1
N2 t
Vin C ton toff
Vc
To/2
Vc
Vin

t
Fig. 2.4.2: Equivalent circuit for one half sinusoidal swing of the ZCS-push-pull resonant converter

The resonant frequency is:

f0 = 1
2π LC

This leads to the minimum on-time of the transistors. The on-time should be a little higher
than half of the resonant period time to ensure that the current reduces to zero.

For maximum energy transfer, V out must be half of V in . This leads to the turns ratio of the
transformer:

N 1 1 V in
V out = 1 V in ⇒ = ⋅
2 N 2 2 V out

The maximum output power is achieved if one half current swing instantly follows the next.

The transfered energy of each half-swing further depends on the value of C and L . The higher
the value of C and the lower the value of L , to maintain a certain resonant frequency, the
higher the amount of energy transfer. (see also the peak value of the current in Figs. 2.4.2 and
2.4.3).
For a certain output power P out , considering V out = V in /2 , it can be shown that for L and C :
 Page 24

2
 V in  ⋅ 2 ⋅ f Switching
L = 2  π
2
f0 1  
⇒ C= und L=  L  ⋅C
C P out
2π ⋅ L
⋅ f0  C 
C

V
GS1
V t
GS2
t
1 C
I
Vin
2 L

Vc Vin

t
N1
V'out Vin
N2

t
N1
-Vout
N2

Figure 2.4.3: Voltages and currents at the ZCS-push-pull resonant converter

I addition to the general advantages of resonant converters, having lower switching losses and
lower radio interference, this particular resonant converter has two more additional
advantages:

The ZCS-push-pull resonant converter can regulate several output voltages using one
control circuit, as per the flyback converter. This because several output voltages seem
to be parallel connected from the view of the primary side. Due to this the energy
always passes to that output, having the lowest voltage value taking into consideration
the turns ratio.

The ZCS-push-pull resonant converter is no-load and short-circuit proof without any
electronic precaution. The output voltage cannot reach more than twice of the nominal
value, because then is V out = V in . The current cannot reach more than twice of the
nominal output current, because then is V out = 0 and I = V in C/L .
 Page 26

Control of switch mode power supplies

The output voltage of a switch mode power supply is kept constant with the help of closed
loop control. The value of the output voltage (actual value) is compared with a reference
voltage (nominal voltage). The difference between actual and nominal value controls the duty
cycle of the transistor drive. The function of the control loop is to regulate the variation of the
mains and of the change of the output current. This is called line regulation and load
regulation.

There are two different methods of regulation: voltage-mode and current-mode control.
The voltage-mode control is the "traditional" method of regulation. Most modern systems
use current-mode control which is the basis of nearly all IC current-mode controllers.

Both controller types can be explained using a boost converter shown in fig 4.1:

Voltage-mode control:

L
Vin Vout Load
Cout

PI-regulator
PWM R1
C1 R4 R3
+ -
t1/T - V2 +
R2 V'out
saw-tooth Vref
generator

Figure 4.1: voltage-mode-control for a boost converter

The output voltage V out is compared to the reference voltage V ref via a voltage divider R 1 , R 2
and amplified by the PI-regulator. A pulse width modulator (PWM, see Fig.4.1a) converts the
output voltage of the PI-regulator V 2 into a pulse width modulated voltage t 1 /T . The output
of the pulse width modulator (PWM) controls the transistor of the boost converter (see also
Chapter 1.2: "boost converter").
V Vspeak
V2 Vs
Pulse With Modulator
t
saw-tooth- Vs
signal - V3
V3
V2 +
t
t1 T
Figure 4.1a:Pulse width modulator
 Page 27

The closed loop operates as follows: If the output voltage V out is to low, the voltage V out will
be lower than the reference voltage V ref , this will cause the output voltage V 2 of the
PI-regulator to increase. In the PWM circuit V 2 is compared with a saw tooth signal and as it
increases the duty cycle t 1 /T also increases, this causes the output voltage to increase until
V out = V ref.

Current-mode control:

L IL ID
VL
Vin Vout Load
Ri ramp Cout

Q
Clock Flip-Flop R1
S
R C1 R4 R3
+ -
- V2 +
R2 V'out
comparator PI-regulator Vref

Figure 4.2: current-mode control for a boost converter

The output voltage V out is compared to a reference voltage V ref via the voltage divider R 1 , R 2
and amplified by the PI-regulator. The output voltage of the PI-regulator V 2 is compared with
ramp voltage across the current measuring resistor R i . When the voltage across Ri exceeds V2
the output of the comparator resets a RS-flip-flop and turns the transistor off. The
RS-flip-flop is set before by the clock. The transistor is turned on by the clock and turned off,
when the ramp voltage (which means the inductor current) reaches a certain value. In this way
the PI-regulator directly controls the inductor current.

The closed loop operates as follows: If the output voltage V out is too low, the voltage V out
will be lower than the reference voltage V ref . This causes the output voltage of the
PI-regulator V 2 increases. The comparator compares the voltage V 2 with the ramp voltage
across R i . In this way V 2 determines the value to which the ramp voltage across R i increases
(which means the value to which the inductor current I L increases) until the transistor is
turned off. If V 2 increases because the V out is lower than V ref , the inductor current will
increase until V out is exactly equal to the reference voltage.

Comparision of voltage-mode to current-mode control:

The PI-regulator of the current-mode control regulates the inductor current directly. This
current feeds the output capacitor C out and the load resistance R L . C out and R L form a first
order system and the step response is an an exponential function.

The voltage-mode control regulates the duty cycle t 1 /T , which means that the voltage across
L is controlled. This voltage operates on a second order system, formed by L, C out and R L .
The step response of such a system is a sinusoidal transient approaching a fixed value.
 Page 28

The current-mode control has therefore a better control response, for this reason most
controllers are current-mode types.

Vref + ID RL Vout Vref + t1/T k ID RL Vout

regulator regulator
- 1+sRL Ca - sL 1+sRL Ca

V'out R2 V'out R2
R1+R2 R1+R2
current-mode voltage-mode
Fig 4.3: Block-diagramms for current-mode- and voltage-mode control

Design of the PI-regulator:

The PI regulated system tends to oscillate, if the capacitance C 1 is selected at too small a
value and if the resistor R 4 is too high a value. To allieviate this problem C 1 should initially
be selected high ( A 1 F foil capacitor is normal in most control circuits). R 4 should be
selected so that the cut-off frequency of the PI-regulator stays well below the cut-off
frequency of L and C out :
1 ≥ 10 1
2π LC a 2πR 4 C 1

The controller should now operate in a stable mode (if not, internal interference or an unfit
architecture of the board might could be a problem). To improve the reaction of the closed
loop, C 1 can be decreased step by step with a parallel increase of R 4 . If the loop starts to
oscillate, C 1 can be increased by the factor of ten and R 4 decreased. Using these design
guides the loop will operate in a stable mode with sufficient regulation speed for most
applications.

HINT:
In many control circuits the operational amplifier (normally called the error amplifier)
is a transconductance amplifier. It supplies an output current (very high output
impedance), which is proportional to the input voltage. In this case R 4 and C 1 are
connected from the output to ground to achieve the PI-characteristic of the regulator.
 Page 29

Design of Inductors and High Frequency Transformers

Inductors store energy, transformers transfer energy. This is the prime difference. The
magnetic cores are significantly different for inductors and high frequency transformers:
Inductors need an air gap for storing energy, transformers do not. Transformers for flyback
converters have to store energy which means they are not a high frequency transformer but
they are in fact an inductor with primary and secondary windings. The material of the cores is
normally ferrite. In addition to this other marterials with high permeability and with a high
saturation point are used.

Calculation of Inductors:

An inductor with certain inductance L and certain peak current I can be determined by the
following calculation:

Inductors should store energy. The stored energy of an inductor is: W = 12 LI 2 . This energy is
stored as magnetic field energy, within the ferrite core and within the air gap (see Fig.5.1.1).
The higher the required stored the energy the larger the required core.

The size of a inductor is approximately proportional to the stored energy.

A Φ=ΒΑ
I : inductor current
N: number of turns
I Hfe A: cross-section area of the core
lfe: magnetic length of the core
Hδ
N δ δ : air gap
Φ: magnetic flux
lfe B: magnetic flux density
Hfe: magnetic field strength within the ferrit
Hδ : magnetic field strength within the air gap

Fig. 5.1.1: inductor with its magnetic and mechanical sizes

The field energy in the inductor is:

W = 1 ∫ H ⋅ B dV ≈ 1 H Fe ⋅ B Fe ⋅ V Fe + 1 H δ ⋅ B δ ⋅ V δ (1)
2 2 2
energy in the ferrite energy in the air gap
The magnetic field density B is continuous and within the air gap and the ferrite is
approximately equal, i.e. B ≈ B Fe ≈ B δ . The magnetic field strength H is not continuous,
within the air gap it is increased by a factor µ r compared to that than within the ferrite. If this
is substituted into equation (1) and considering
B = µ 0 µ r ⋅ H, V Fe = l Fe ⋅ A and V δ = δ ⋅ A this leads to:
 Page 30

W≈ 1B  Fe + δ  ⋅ A
2 l

2 µ0  µr 

µ r of the ferrite amounts to 1000...4000. It should be noted that the magnetic length of the
ferrite is reduced by µ r in the above equation. Therefore it can be seen that the energy is
mainly stored within the air gap.
2
This leads to: W ≈ 1 B µ⋅A⋅δ
2 0

Inductors require an air gap to store energy.

Because the energy is stored within the air gap, an inductor requires a certain volume for the
air gap to store a certain amount of energy. The energy is given by 12 LI 2 . The core material
has a limit for the maximum magnetic flux density B , this limit is about B max = 0, 3 T for
usual ferrite materials. This leads to a minimum required volume V δ of the air gap:

L I 2 ⋅ µ0
Vδ = A ⋅ δ ≥ 2
where B max = 0, 3 T
B max

Knowing the required volume of the air gap, a core can be selected from a databook of ferrite
cores.

The number of turns N can be calculated with help of the magnetic conductance A L (often
simply called the A L -value):

N= L A L :magnetic conductance
AL

The A L -value can be verified from the databook of the ferrite cores.

The maximum flux density should not be higher than 0.3 Tesla. The maximum flux density
within the ferrite can be calculated using the data of the core datasheet.

B= L ⋅ I = N ⋅ A L ⋅ I ≤! 0, 3 T
N ⋅ A min A min

A min : Minimum cross-cut of the core. The flux density has its maximum. at A min . A min can be
verified from the datasheet.
 Page 31

Calculation of the wire:

The current density S of the wire can be chosen between 2 und 5 A/mm² (depending on the
size and the isolation, which determines the heat transport out of the inductor). This leads to
the diameter of the wire d :

4 ⋅ I RMS A
d= with S = 2…3…5
π⋅S mm 2

Calculation of High Frequency Transformers

A high frequency transformers transfer electric power. Its mechanical size depends on the
power to be transfered and on the operating frequency. The higher the frequency the smaller
the mechanical size. Usually frequencies are from 20 to 100kHz. The material of the core is
ferrite.

Databooks for appropriate cores provide information about the possible tranfer power for
various cores.

The first step to calculate a high frequency transformer is to choose an appropriate core with
the help of the databook, the size of the core is dependent on the transfer power and the
frequency. The second step is to calculate the number of primary turns. This number
determines the magnetic flux density within the core. The number of secondary turns is the
ratio of primary to secondary voltage. Following this the diameters of the primary and
secondary conductors can be calculated depending on the RMS-values of the currents.

Calculation of the minimum number of primary turns:

V1 I1 I2

t V1 V2 R
T/2 T

V2 N2 N1 N2
V1
N1
simple equivalent circuit:
t
I1 I2'
I2 V1 N2 2
R N1 V1 L1 V2' R N1
N2
t IM

2
I1 V1 N2
R N1 ∆ IM IM
IM : magnifiing current
t

Figure 5.2.1: Voltages and currents at a transformer

 Page 32

The voltage V 1 at the primary side of the transformers has a rectangle shape. This causes an
input current I 1 , which is the addition of the back transformed secondary current I 2 and the
magnetising current I M (see figure 5.2.1). To keep the magnetising current I M low, a
magnetic core without an air gap is used.

The rectangle voltage V 1 causes a triangle shape for the magnetising current I M . The
magnetising current is approximately independent of the secondary current I 2 (see the simple
equivalent circuit in figure 5.2.1). The magnifiing current is approximately proportional to the
magnetic flux or flux density. The input voltage V 1 determines the magnetic flux. The
physical correlations are given by Faradays law of induction: V = N ⋅ dΦ .
dt
V1 V1 ∆B
B
B
t
T/2 T

Figure 5.2.2: input voltage and magnetic flux density at the transformer

For the transformer in figure 5.2.1 follows:

V 1 ⋅ T/2
∆B =
N1 ⋅ A

The change ∆B of flux density depends on the frequency f = 1/T and the number of
turns N 1 . The higher the frequency and the number of turns the lower the change of
flux density.

The minimum number of turns N 1 min can be calculated to ensure that a certain change of flux
density ∆B is not exceeded. The saturation flux density of about B ≈ 0, 3 T (which means
∆B ≈ 0, 6 T ) cannot be used in high frequency transformers. In push-pull converters going
around the hysteresis loop with every clock would cause unacceptable losses i.e. heat
generation. If no further information on core losses and termal resistance are available, ∆B
should be limited to ∆B ≈ 0, 3…0, 2 T for operating frequencies from 20 to 100 kHz. The
lower ∆B the lower the core losses.

This leads to a minimum number of turns for N 1 :

V 1 ⋅ T/2
N 1 min ≥ where ∆B ≈ 0, 2…0, 3 T
∆B ⋅ A min

A min : minimum cross-section area of the core. This is where the flux density is at a
maximum. A min can be checked from the datasheet.

HINT:
In single ended forward converters the core is magnetised into one polarity only. Iin
push-pull converters the core is magnetised alternating into both polarities.
 Page 33

B B
∆B ∆B

t t
single ended forward converters push-pull converters
The calculation of the minimum number of turns N 1 min is equal for these different
types of switch mode power supplies.

Calculation of the winding conductors:

The diameter of the conductors depends on the RMS-value of the current. The current can be
calculated with the power.

For the push-pull converter follows:

I1 I2
+Pout/Vin +Pout/Vout
P P
I 1RMS ≈ out and I 2RMS = out
V in V out
-Pout/Vin -Pout/Vout

For the single ended forward converter follows:

I1 I2
2 P out 2 P out
I 1RMS ≈ and I 2RMS = Pout/Vin Pout/Vout
V in V out

The magnetising current can be neglected in this calculation. The current density can be
chosen in a range of 2 to 5 A/mm, depending on the termal resistance of the choke. The
cross-section A wire and the diameter d wire can be calculated as follows:

A wire = I and d wiret = I⋅4 where S = 2… 3 …5 A 2

S S⋅π mm

HINT:
If good coupling is important, the primary and secondary winding should be placed
on top of each other. Improved coupling is achieved if the windings are interlocked.
The coupling is bad in a) good in b) and in c) about four times better than in b).

secondary
primary second. secondary primary
primary

primary
primary second. secondary primary
secondary
a) b) c)
 Page 34

HINT:
The primary number of turns should not be chosen significantly higher than N 1 min ,
otherwise the copper losses of the wire would increase needlessly due to the longer
conductor.

HINT:
For high frequencies and large diameter of the wire the skin effect should be
considered. For operating frequencies of more than 20kHz and diameters of more
than 1mm litz wire or copper foil should be used.
 Page 35

Power-Factor Control

The European standards EN61000-3-2 define limits for the harmonics of the line current.
This concerns appliances, which may be sold to public customers and have an input power of
≥ 75 W (special regulations see EN61000-3-2). Some limit values from this standard are
given in table 6.1. In practice this standard means that for many applications a mains rectifier
with smoothing is not allowed because of the amount of harmonics (see figure 6.1).

input power75 to 600W input power >600W

harmonic- maximum value of maximum value of
order harmonic current harmonic current
n per Watt (mA/W) / maximum (A) (A)
3 3,4 / 2,30 2,30
5 1,9 / 1,14 1,14
7 1,0 / 0,77 0,77
9 0,5 / 0,4 0,40
11 0,35 / 0,33 0,33
table 6.1: RMS limits for the harmonics of the line current

To keep the line current approximately sinusoidal, a boost converter can be used (see figure
6.2). In this case the boost converter is called Power Factor Pre-regulator (PFC). In
comparison to the boost converter the PFC is controled in a different way: The output voltage
is higher than the input voltage as for the boost converter, but the transistor is turned on and
off in a way that a sinusoidal input current is achieved instead of a exact constant output
voltage. The transistor is driven in such a way, that the inductor current I L (t) follows the
shape of the rectified mains V in (t) . The output voltage of the PFC is controled to
approximately U a ≈ 380 V .

Vout
Um
Im
t
Um Vout Load Im
C
mains t

Figure 6.1: Usual rectifieing and smoothing of the mains and its mains current
 Page 36

Vin Vout Vout

t t

Iin L D

mains EMI- Vin T Vout

Filter
Cout

Figure 6.2: boost converter as a power-factor preregulator

Currents, Voltages and Power of the PFC:

Vin
Iin Vin
Iein
Iin
∆I L

Pin t
Pout Pin Pout

t
Vout
∆Vout t
t
a) b)

Figure 6.3: Currents, voltages and Power of the PFC

For the following calculations, it is assumed that the output power is constant:

P out = V out ⋅ I out = const.

The input current should be controled to a sinusoidal shape and should be in phase with the
input voltage. The input power is now pulsating and can be calculated as follows:

V in ⋅ I in
P in (t) = ⋅ (1 − cos 2ωt)
2

V ⋅I V in ⋅I in
The input power consists of a DC-part P in = = in2 in and of an AC-part P in ∼ = 2
⋅ cos 2ωt .
The DC-part is equal to the output power P out , providing a loss-free PFC .
 Page 37

V in ⋅ I in
P in = = V out ⋅ I out = P out
2
P out
In practice an efficiency of about η = 95% is realistic which means that P in ≈ .
0.95

The output capacitor C out is charged by the pulsating input power P in and discharged by the
constant output power P out . This causes a voltage ripple ∆V out at C out , which depends on the
value of C out . For the 230V/50Hz-mains, providing V out = 380 V and ∆V out /V out = 10% , C out
can be calculated to:

µF
C out ≈ 0, 5;
W

The choke L determines the high frequency ripple of the input current ∆I L (figure 6.3b). The
higher the inductance and the higher the clock frequency f , the lower this current ripple. If
∆I L = 20% of the peak value of the input current I in and assuming that the
230V/50Hz-mains voltage is a minimum of V in min = 200V, it follows that:

L ≈ 50 ⋅ 10 ; L (H), f (Hz), P (W)

3

f ⋅ P in

and for the maximum inductor current:

2P in
I Lmax = I in max + 1 ∆I L = 1, 1 ⋅
2 U in min

Controlling the PFC:

Two feedback circuits are required:

One to control the input current to be sinusoidal (input current control)

and
one to keep the average output voltage constant, which means keeping it independent
from the load power consumption (output voltage control)

The input current control loop is lead by the input voltage. In this case the input current
aquires the same shape as the input voltage and consequently the power factor of the mains
current will be unity.

The output voltage is controled by comparing it to a constant reference voltage.

The multiplier links the two loops. The output of the multiplier is sinusoidal and its
magnitude depends on the output voltage control loop. If the output voltage decreases from its
nominal value, the output voltage of the voltage control amplifier increases which causes the
magnitude of the multiplier output to increase and consequently the RMS-value of the input
current also increases.
 Page 38

Iin L
Vin Vout
RM Cout

R1

C5
R2 R4
C6
V'in
R5
PWM
R3
V2
multiplier
input current
control C7
R9
R7 R8

R10
output voltage Vref
control

Figure 6.4: The control loops of the PFC

The RMS value of the input current is controled by the output voltage control loop
while the input current control loop drives the input current to be sinusoidal.
 Page 39

Radio Interference Suppression of Switch Mode Power Supplies

Switch mode power supplies generate radio interference due to the high frequency switching.
This interference propagates through space by means of the electromagnetic fields or via the
mains supply in the form of currents and voltages.

The legislation created limits for the levels of interference. These limits are published in the
European Standards. Table 7.1 gives some of the most important limits for mobile high
frequency equipment (interference class B). High frequency equipment is that which operates
at a frequency in excess of 9kHz.

measurand frequency range limits standard

electromagnetic interference 30 to 230 MHz 30 dB(µV/m) EN55022
at 10m distance 230 to 1000 MHz 37 dB(µV/m) class B
current harmonics 0 to 2 kHz see table 6.1 EN61000
in the mains (PFC)
conducted-mode interference 0.15 to 0.5 MHz** 66 to 56 dB(µV) Q* EN55022
voltages at the mains wires in 56 to 46 dB(µV) M* class B
repect to earth potential 0.5 to 5 MHz 56 dB(µV) Q*
46 dB(µV) M*
5 to 30 MHz
60 dB(µV) Q*
50 dB(µV) M*
* Q: Measured by quasi-peak detector
M: Measured by average-detector
** Linear decrease to the logarithm of the frequency

Table 7.1: Limits for mobile high frequency equipment class B

Radio interference radiation:

High frequency equipment emission radio interference is measured as radio noise field
strength (µV/m) . The amount of radio interference radiation depends on the rise time of the
switched currents and voltages and significantly on the layout of the printed circuit board. To
keep the radio interference radiation low, three principles should be adhered to:

Meshes, in which a switched current flows, should be as small as possible in their

surrounded area to keep their electromagnetic field low.
Nodes whose potential are in respect to earth step up and down with switching, should
be as small as possilble in their volumetric space, to keep their parasitic capacitance to
earth low.
The switch mode power supply should have a metal housing.

HINT:
In addition to the reduction of the interference radiation, the first two principles are
also good for keeping the conducted interferences low, which leave the power supply
via the mains. It should also be noted that a high interference level results in
inaccurate switching of the transistors and problems with the closed loop control
circuit. This often causes audible noise.
 Page 40

Mains input conducted-mode interference:

Switch mode power supplies take high frequency currents out of the mains. These currents
cause a voltage drop at the source impedance of the mains which can be measured at the
mains terminals. According to the European Standards the interference voltages have to be
measured between the mains terminals and earth. For this measurement specific radio
interference test equipment is needed which includes a radio interference meter and an
artificial mains network. This equipment is required to define a specific mains impedance for
comparable measurments.

We distinguish between three different radio interference voltages (see Fig.7.1):

Unsymmetric radio interference voltage: This is the high frequency voltage between
earth and each mains terminal. Only this voltage is measured corresponding to the
standards. The limits in table 7.1 are valid or this voltage only.
Common-mode radio interference voltage (asymmetric radio interference voltage):
This is the sum of all unsymmetric interference voltages in respect to earth.
Differential-mode radio interference voltage (symmetric radio interference
voltage): This is the high frequency voltage between the mains terminals.

L1
radio interference
source
diff. (switch mode
Mains radio int. com.
voltage radio int. power supply )

N
voltage ~
unsym.
radio int.
voltage
PE

Figure 7.1: radio interference voltages at the single phase mains

Although the legislation requires only the measurement of the unsymmetric radio
interference voltages, the common-mode and differential-mode interferences are decisive for
the radio interference suppression. The respective suppression of common-mode and
differential-mode interference needs different designs and components.

Suppression of common-mode radio interference:

Common-mode radio interference voltages at the mains terminals L 1 and N (for three phase
mains L 1 , L 2 , L 3 and N ) are common mode voltages in respect to earth potential PE , which
means they are equal in magnitude and phase. The interference currents I ∼≈ , which are driven
by this common-mode voltage, are also common-mode currents. These flow via earth (earth
conductor) and back through the parasitic capacitance C earth . C earth is very low. Due to this,
the common-mode interference voltage has a very high impedance, which means that this
interference source acts like a current source. A low-pass filter to suppress the interference
voltages at the mains terminals must therefore be arranged as in figure 7.2. Looking from the
switch mode power supply the required low pass filter must have a shunt capacitor (Cy) and a
 Page 41

current compensated inductor. Current compensated chokes are wound so that no magnetic
field is generated by the operating current (50- or 60Hz), see figure 7.3. Due to this the choke
only acts against the common-mode interference current and does not effect the operating
current.

I50Hz I~~
L1 ~ radio interference
current- source
Mains compensated (switch mode
I50Hz choke I~~ power supply)
N ~ ~
~
Cy Cy 2I~~ C earth
PE earth leakage current ~

Figure 7.2: Suppression of the asymmetric (common-mode) radio interference voltages

The capacitors are called y-capacitors. Y-capacitors have to fulfil special safety requirements,
because they would connect the mains phase to ground in the case of a fault. Y-capacitors
may not exceed a certain capacity to ensure that the permitted maximum earth leakage
current is not exceeded. The earth leakage current is a 50Hz-current (or 60Hz in certain
countries). The maximum earth leakage current is 3.5mA (in medical equipment it is a
maximum of 0.5mA). According to the standards for the measurement of earth leakage
current, terminals L 1 and N have to be connected and the maximum mains voltage has to be
applied between L 1 & N and PE . This means that the y-capacitors are in parallel. For the
European 230V/50Hz-mains it follows that for the maximum y-capacitor:

Cy ≤ 1 × 230 V + 10% ≈ 22 nF
2 2 π 50 Hz × 3.5 mA

I50Hz I~~ I50Hz I~~

~ ~

I50Hz I~~ I~~

~
~ I50Hz

current compensated choke ring core double choke with powder core
Figure 7.3: left: current compensated choke for common-mode interfernces, right: not current compensated
choke (in this case a ring core double choke with powder core) for differential-mode interferences

Suppression of the differential-mode radio interference:

Differential-mode radio interference voltages are high frequency voltages between the mains
terminals L 1 and N . To reduce the interference level, a LC-low-pass filter has to be inserted
between the mains conductors L 1 , N (figure 7.4). The differential-mode interference voltage
 Page 42

results mainly from the pulsed current, which is taken from the switch mode power supply
from the mains rectifier smoothing capacitor. Due to the impedance of the smoothing
capacitor a high frequency voltage is generated between L 1 und N . This is a low impedance
which means that the interference source acts as a voltage source. Looking from the switch
mode power supply the interference filter must be arranged using a series choke followed by a
shunt capacitor (see figure 7.4). The choke must not be a compensated choke, because
differential-mode interference current and 50Hz-operating current (which is also a differential
type) cause a mgnetic field within the core (see figure 7.3). To avoid saturation these chokes
require an air gap. With a ring core the air gap is not visible, because the air gap is achieved
due to the amount of glue used in the iron powder. Open cores are also used. With this type
the magnetic field loop closes through space. Ring cores are prefered because they have a low
magnetic field outside the core.

I50Hz I~~
L1 ~ radio interference
source
mains (switch mode
Cx power supply)
I50Hz I~~
N ~ ~
~
choke

Figure 7.4: Suppression of the differential-mode interferences

The capacitors for this purpose are called x-capacitors. They have a lower test-voltage than
y-capacitors and are not limited in their value. Foil type capacitors up to 1µF are normally
used.

HINT:
Sometimes the impedance of the differential-mode interference source is
approximately equal to the mains impedance. In that case a π-low-pass filter using
two x-capacitors are appropriate (in figure 7.4 dotted lined).

Complete radio interference filter:

I50Hz
L1 radio interference
source
mains (switch mode
I50Hz Cx Cx power supply)
N ~
~
powder- current-
core comp.
choke choke Cy Cy
PE housing

Figure 7.5: radio interference filter for common-mode and differential-mode filtering

Figure 7.5 shows a complete radio interference filter. The component values can be found
iteratively and with the help of experience. With the radio interference meter only the
unsymmetric interference voltages can be measured. Due to this it is not possible to
differentiate between common-mode and differential-mode interference. In practice the
operating frequency and several harmonics are differential-mode interference and all high
frequencies, say above 5MHz, are common-mode. Often a powder core choke is not required.
Page 43