Φ Abstract—In this paper, a high power single switch flyback

power supply is presented. Performance of this power supply in continuous conduction mode is investigated in implementation. An appropriate efficiency is achieved by designing a proper snubber circuit. With regard to the importance of reliability in switch mode power supplies, reliability consideration is discussed for this high power flyback in details. This paper demonstrates the flyback topology has suitable failure rate in high power applications because it has only one power switch.

Index Terms—Flyback, reliability, snubber circuit, switch mode power supply.

I. INTRODUCTION HE flyback topology has long been attractive because of its relative simplicity when compared with other

topologies used in industrial applications [1]-[3]. Dislike other power supply topologies, flyback topology has no secondary output inductors that leads to saving in cost and volume of output inductors [4]. In addition, in comparison with forward topology, the out put freewheeling diode is not needed in flyback power supply that results in increase of reliability as well as saving in cost. The flyback ‘‘transformer” serves the dual purpose of providing energy storage and converter isolation, theoretically minimizing the magnetic component count when compared with the forward converter [5]. On the other hand, this transformer can be considered as a main drawback of this converter because of the existence of the transformer leakage inductance. As a result, design of an appropriate snubber circuit is necessary in order to achieve less voltage tension, especially in high power range.

As power supplies are the heart of every electronic equipment, special attention must be paid to their reliability. The overall reliability of a system is strongly dependent on the reliability of its power supply. Usually a system can incorporate some degree of redundancy to enhance reliability of but is more difficult to incorporate redundant power supply. Reliability is a measure of the ability of a component or piece of equipment to perform its function, in an adequate manner, for a stated period of time. It is a necessary attribute for a produce to be fit for purpose for which it is designed and every bit as important as the actual functionality of the product.

According to improvement in semiconductor technology and widespread application of IGBTs in recent years, the flyback

A. Rahnamaee is a MS.c. Student. Department of Electrical Engineering, Amirkabir University of Technology, Tehran, Iran (e-mail:rahnamaee@ieee.org)

J. Milimonfared is with the Department of Electrical Engineering, Amirkabir University of Technology, Tehran, Iran (e-mail: monfared@aut.ac.ir)

K. Malekian is a MS.c. Student. Department of Electrical Engineering, Amirkabir University of Technology, Tehran, Iran (e-mail: malekian@ieee.org)

power supplies can be used in the high power application to achieve its above-mentioned advantages. In the present paper, a 1kW power supply is constructed using flyback topology. Also, reliability computations are done on all components of the constructed flyback power supply.

II. DESIGN OF 1KW FLYBACK POWER SUPPLY Design relations of power stage of flyback topology are

thoroughly discussed in [6] and [7] as well as design procedure. Basic structure of flyback topology is shown Fig. 1. Also, the current and voltage waveforms of main switch under continuous conduction mode (CCM) are depicted in Figs. 1(a) and (b), respectively.

Diode

Capacitor Load

N1 N2

V

Drive&

Control Circuit

Transformer

+

-

Vo

IdealTransformer

Lm

Llk

N1 N2

magnetizinginductance

Leakageinductance

Transformer Model

dc

isw

Fig. 1. Typical flyback topology.

i

i

Switching Period

Iswitch

Time

2

1

= dc

m

VSlopeL

(a)

Voltagearise

due leakageinductance

Switching Period

VSwitch

Time

input voltage+

reflectedoutputvoltage

(b)

Fig. 2. Typical waveforms of main switch in continuous mode operation of flyback power supply: (a) current waveform and (b) voltage waveform.

Reliability Consideration for a High Power Single Switch Flyback Power Supply

Arash Rahnamaee, Student Member, IEEE, Jafar Milimonfared, and Kaveh Malekian, Student Member, IEEE

T

527978-1-4244-1633-2/08/.00 ©2008 IEEE

The switch applied in flyback topology endures less voltage and current tensions in CCM rather than discontinues conduction mode (DCM) [8]. Moreover, ohmic losses and EMI are less in CCM because of having lower current pick. Hence, the flyback power supply is designed to operate in CCM.

Characteristics of the constructed flyback power supply are given in Table I.

Because primary current of transformer is the magnetizing current while switch is turned on, an air gap is utilized to avoid magnetic saturation in transformer. Indeed, this transformer operates as two coupled inductors. This issue results in an increase in leakage inductance. Also, there is large stored leakage energy that appears as a large spike across switch when switch turns off. This phenomenon intensifies in high power applications. As a result, in order to provide a suitable voltage tension across switch, the snubber circuit should be designed carefully. With regard to design procedure mentioned in [9], power transformer is designed. Considering operating frequency of 50kHz, a 5mm air gap is needed to avoid magnetic saturation in ETD59 ferrite core. In order to decrease leakage inductance and ohmic losses, interleaved winding method is used in the power transformer.

According to the above discussion, the dissipative RCD snubber [10] is selected that its corresponding topology is shown in Fig. 3.

Using the dissipative RCD snubber, the switch voltage is clamped to the voltage which is defined as follows,

CE(max) dc cv v v> + (1)

where CE(max)v is the switch breakdown voltage; dcv is the dc supply voltage; and cv is the voltage across the snubber capacitor sC . It should be noted that cv is equal to on v , where

ov is the output voltage and n is the transformer ratio. The snubber capacitor should be chosen so that bellow condition is satisfied.

TABLE I CHARACTERISTICS OF THE CONSTRUCTED FLYBACK POWER SUPPLY

Input voltage 180-250 Vac Output voltage 48 Vdc

Output Current 21A Switching Frequency 50kHz

Diode

Capacitor Load

N1 N2

Vdc

Drive&

Control Circuit

Transformer

+-

RS CS

Snubber Diode

+

-

Vo

RCDSnubber

Fig. 3. Dissipative RCD snubber in flyback power supply.

Fig. 4. Constructed flyback power supply.

200 300 400 500 600 700 800 900 1000 11000.7

0.72

0.74

0.76

0.78

0.8

0.82

0.84

0.86

Output Power

Effi

cien

cy

Experimental MeasurmentFitted Diagram

Fig. 5. Efficiency curve with respect to the output power.

sss fR

C.1>> (2)

where sf is the switching frequency; sR is snubber resistor.

The voltage across snubber resistance will rise unit energy stored in the leakage inductance of power transformer is dissipated.

To satisfy all above conditions, snubber parameters are chosen as follows,

Ω==

kRFC

s

s

102547.0 μ

The constructed flyback power supply is shown in Fig. 4. Since using a dissipative RCD snubber circuit, the voltage across the snubber capacitor is constant and equal to nVo under all output powers. Consequently, snubber losses are always approximately constant. As a result, efficiency is less in lower output power in the manner shown in Fig. 5.

III. RELIABILITY COMPUTATIONS FOR THE CONSTRUCTED FLYBACK

A. Concept of Reliability

In order to be able to predict the reliability of a component, it is necessary to know its statistical failure rate. This knowledge is based on data gathered from either past operational experience of the component or from bath testing. For most components, the failure rate varies with time according to the well known "bath-tub" curve in Fig. 6.

528

Infantmortality

Falling

Failurerate

constant rising

Operatinglife

λ λ λ

λ

Wearout

Time Fig. 6. The bath-tub curve.

During the early stages of life (region 1) the failure rate is

high and decreasing. This is often referred to as the infant mortality period where components with inherent manufacturing weakness will fail prematurely. During the middle life period (region 2), the failure rate is constant and caused by random failures. This period is the useful life region. Finally, the wear out region (region 3) is reached where failures occur because of aging effects, and the failure rate starts to slowly increase.

Various measures are used to quantify reliability, the commonest being failure rate, λ , and the meant time to failure, MTTF. These are related thus,

λ1=MTTF (3)

Typical values of failure rate for a high reliability part are around 810− failures per hour. Another measure often used is the mean time between failures, MTBF. Strictly, this is the mean time between failures as device goes through successive cycles of failure and repair, and related to the MTTF thus:

timerepairMTTFMTBF += (4)

For semiconductor parts, repair is not normally possible and the MTTF and MTBF are often used interchangeably.

The reliability, ( )R t , of a component is related to its failure rate and time, t, thus:

( )( ) tR t e λ−= (5) This assumes that the failure rate is constant and random.

This assumption can be justified because of the many diverse forces leading to failure for complex parts; these conspire to produce, in effect, random failures.

To calculate the reliability of a system comprising many components, the individual failure rates are summed and used to calculate the overall reliability, therefore:

1 2total nλ λ λ λ= + + +… (6) ( )( ) total t

totalR t e λ−= (7)

B. Reliability Prediction

Methods of reliability calculation are outlined in HDBK-217F [11]. This contains failure rate data on all types of components that have been used commonly. These data have been built up from experience of component failures and also from the results of accelerated testing at elevated stress.

As a design progress, more detailed design information is available and it is possible to estimate overall reliability by calculating the individual component failure rates. This

method is known as part stress analysis as it is based on the operating stress of each component.

A variety of different models are used depending on the component type; take, for example, semiconductors:

p b E A Q R S Cλ λ π π π π π π= × × × × × × (8) where pλ is the part failure rate; bλ is the base failure rate;

Eπ is the environment factor; Aπ is the application factor (linear or switched); Qπ is the quality factor (e.g. JAN,

JANTX etc.); Rπ is the power rating; Sπ is the voltage stress factor; and Cπ is the complexity factor.

1) Main switch: Measuring dynamic and static losses is necessary to calculate semiconductors reliability. The voltage and current waveforms of main switch are illustrated in Fig. 7(a). As shown in this figure, there are remarkable dynamic losses while the switch is turning off. Fig. 7(b), properly, shows the voltage and current overlap during turn-off. A BUP314D IGBT is used as a main switch in the constructed power supply. Considering Fig. 7(b) and characteristic of this IGBT, the main switch losses can be calculated as follows,

Loss Static DynamicP P P= + (9)

( ) 3 13.81CE sat Staticv V P W= ⇒ =

1 s overlap

s

DT t

Dynamic switch switchDTs

P v i dtT

+= ∫

(10)

55 68.81Dynamic LossP W P W= ⇒ = where D is duty cycle; Ts is switching period; and toverlap is current and voltage overlap duration for main switch.

The base failure rate of IGBT is considered 0.01 6/10failures Hours , and other factors used in IGBT reliability

computation are considered like ones related to BJT [12]. Reliability factors are derived using measured values and

MIL-HDBK-217F as follows,

P b T A R S Q Eλ λ π π π π π π= × × × × × × (11)

1 12114273 298

0.42 /36 68.81 0.42 64.9

2.30.78

J

J C JC Loss

JC

J

TT

A

R

T T PC W

T C C

e

θθ

πππ

⎛ ⎞⎛ ⎞− −⎜ ⎟⎜ ⎟⎜ ⎟+⎝ ⎠⎝ ⎠

= += °

= ° + × = °

= ===

(3.1 )

710 0.5911200

0.045 0.28s

CEs

CEV

S

Applied VVRated V

eπ ⋅

= = =

= ≈

where VCE is voltage across collector-emitter of IGBT; TJ is junction temperature; TC is case temperature; and θJC is junction-to-case thermal resistance.

529

(a)

(b)

Fig. 7. Voltage and current waveforms of main switch (a) in two switching period (b) while switch is turning off.

6

5.560.01 2.3 0.7 8 0.28 5.5 6 1.19

/10

Q

E

p

failures Hours

ππλ

=

== × × × × × × =

2) Output diode: In order to calculate output diode failure rate, the same procedure like what applied for IGBT has been done. Voltage and current waveforms are shown in Fig. 8. As shown in this figure, the output diode dynamic losses are negligible. Consequently, in this work, the static losses are only considered. For calculating diode static losses, the relations given in datasheet of applied diode, BYW99P-200, is used [13].

( )( )

( ) ( )

( ) ( )( ) ( )

( )

2

21

28

0.65 0.016

1 1

2

1.8 /

0.2 /

AV

RMS

F AV F RMS

JC

jc c

jc

c

I Output Diode A

I Output Diode A

P I I

T Diode P Diode

Per diode P Diode

Per diode C W

C W

θ θ

θθ

=

=

= × + ×

=

× + ×

= °

= °

( )2

34 6 1.8 6 0.2 46

0.55 21 0.016 28 24

J C JC

J

T T TT C C

P W

= += ° + × + × = °

= × + × =

Failure rate of the output diode can be predicted using above calculations.

P b T S C Q Eλ λ π π π π π= × × × × × (12)

6

1 13091273 298

2.43

0.025 /10

1.9120 0.6200

0.318

6.0

J

b

TT

S

S S

C

Q

E

Failures Hours

eApplied VoltageVRated Voltage

V

PlasticCase

λ

π

ππππ

⎛ ⎞⎛ ⎞− −⎜ ⎟⎜ ⎟⎜ ⎟+⎝ ⎠⎝ ⎠

=

= =

= = =

= ===

=

For each diode, it can be written, 60.025 1.9 0.3 1 8 6 0.68 /10P Failures Hoursλ = × × × × × =

Fig. 8. Voltage and current waveforms of output diode.

3) Snubber diode: The procedure used to predict the reliability of snubber diode is similar to what done for output diode.

60.025 /10

45 /

36 4 5 562.60.58 0.291

b

Loss

JC

J C JC loss

T

S S

C

Failures HoursP W

C WT T P C

V

λ

θθ

ππ

π

==

= °= + = + × = °== ⇒ ==

530

6

5.56

0.025 2.6 0.291 5.5 6 0.62 /10

Q

E

p Failures Hours

ππλ

=

=

= × × × × =

4) Bridge diode: In the same way, the bridge diode reliability can be predicted as follows,

( )

6

Loss

6

0.0038 /104.6626A

P 3.262.2 /

35 2.2 3.26 42.171.8310 0.31

10000.0615.5

6

0.038 1.8 0.06 1 5.5 6 0.14 /10

b

AV input stage

JC

J C JC loss

T

S

S

C

Q

E

p

Failures HoursI

WC W

T T P C

V

Failures Hours

λ

θθ

π

ππππλ

==

== °

= + = + × = °=

= =

===

=

= × × × × × =

5) Power Transformer: The failure rate for a transformer is given as,

p b T Q Eλ λ π π π= × × × (13)

The hot spot temperature is required to calculate Tπ . So, reliability prediction procedure can done as follows,

( ) ( )5

6

.11 1 1273 2988.617 10

6

0.49 /101.1 41 27 1.1 27 42

2.63

6

0.049 1.25 3 6 1.1 /10

HS

b

HS A

TT

Q

E

p

Failures HoursT T T C

e

Failures Hours

λ

πππλ

−

⎛ ⎞⎛ ⎞− −⎜ ⎟⎜ ⎟⎜ ⎟+× ⎝ ⎠⎝ ⎠

== + Δ = − × + = °

= ==

=

= × × × =

6) Input and output capacitors: To predict reliability of capacitors, the factors mentioned in following relation should be derived,

p b T SR Q E cap Vλ λ π π π π π π= × × × × × × (14)

where capπ is capacitance factor; Vπ is voltage stress factor;

and SRπ is series resistance factor which is considered only for Tantalum capacitors. Thus,

0.0001227

b

T Cλ =

= °

50.35 1 1

273 2988.617 10

1.11310

TT

T

SR

Q

E

eπππππ

−⎛ ⎞− ⎛ ⎞−⎜ ⎟⎜ ⎟+× ⎝ ⎠⎝ ⎠=

==

=

=

For the output capacitors, it can be written, 0.23 0.23

5

6

1000 4.9

48 0.860

1 5.210.6

0.00012 1.1 1 3 10 4.9 5.2 0.1

/10

cap

V

p

C

Actual Power DissipationSRated Power

S

Failures Hours

π

π

λ

= = =

= = =

⎛ ⎞= + =⎜ ⎟⎝ ⎠

= × × × × × × =

And in the same way for the input capacitor, the failure rate

can be predicted as follows, 0.23

6

560 4.3

310 0.7754004.660.00012 1.1 1 3 10 4.3 4.66 0.08

/10

cap

V

p

S

Failures Hours

π

πλ

= =

= =

== × × × × × × =

7) Snubber capacitor: In this work, an MKT capacitor is

applied as a snubber capacitor. This kind of capacitor has a different base failure from electrolytic capacitors. To obtain voltage stress factor ( Vπ ), the voltage across snubber capacitor shown in Fig. 9 is required. Thus,

( )

6

0.090.09

5

6

0.00051 /10113

10

0.47 0.93

250 0.251000

1 1.010.6

0.00051 1 1 3 10 0.93 1.01 0.014

/10

b

T

SR

Q

E

cap

V

p

Failures Hours

C

S

S

Failures Hours

λππππ

π

π

λ

===

=

=

= = =

= =

⎛ ⎞= + =⎜ ⎟⎝ ⎠

= × × × × × × =

531

Fig. 9. Voltage across snubber capacitor.

8) Snubber resistor: In the constructed power supply, eight

paralleled 10W resistors are used as a snubber resistor. The failure rate relation is given as,

P b T P S Q Eλ λ π π π π π= × × × × × (15)

where Pπ is power dissipation factor. So,

( )

5

6

0.08 1 1273 2988.617 10

0.39 0.39

6

0.0024 /1068

1.5

6 2104

0.0024 1.5 2 10 4 0.288 /10

b

TT

T

P

Q

E

p

Failures HoursT C

e

Power Dissipation

Failures Hours

λ

ππ

πππλ

−⎛ ⎞− ⎛ ⎞−⎜ ⎟⎜ ⎟+× ⎝ ⎠⎝ ⎠

== °

==

= = ==

=

= × × × × =

9) Control section: Failure rate of control section is

negligible whit respect to failure rate of power stage components. Failure rate of control section will be discussed in following,

Fuse:

60.01 2 0.02 /10p b E

p Failures Hours

λ λ π

λ

= ×

= × =

IC:

60.0048 1 2 2 0.02 /10p b T Q E

p Failures Hours

λ λ π π π

λ

= × × ×

= × × × =

Resistors:

60.00045 0.58 10 4 0.01 /10p b P Q E

p Failures Hours

λ λ π π π

λ

= × × ×

= × × × =

Ceramic and other types of capacitor:

60.00075 1 10 2 0.015 /10p b CV Q E

p Failures Hours

λ λ π π π

λ

= × × ×

= × × × =

Table II provides summarized results of reliability calculations as well as total failure rate of constructed flyback power supply. As shown in this table, with regard to application of IGBT, which has better failure rate than MOSFETs because of having voltage stress factor (πs), total failure rate of power supply system has been improved properly.

IV. CONCLUSION In this paper, a 1kW flyback power supply has been

designed and constructed. Using a proper snubber circuit, the power supply has been found an appropriate performance. In other words, switch tensions have been kept within switch limits, and also, the power supply has achieved an acceptable efficiency. Moreover, reliability prediction has been done on all components of the constructed flyback power supply. The reliability computations have demonstrated adequate failure rate of main switch and other components. This paper has proved that the flyback topology can be considered as a reliable alternative for the high power applications as well as low power applications.

TABLE II RELIABILITY CALCULATIONS RESULTS AND TOTAL FAILURE RATE OF CONSTRUCTED FLYBACK POWER SUPPLY

Power Stage Control Stage

Main switch

Output diodes

Snubberdiode

Bridge diodes Transformer Input

capacitorsOutput

capacitorsSnubber capacitor

Snubber resistors Resistors Capacitors Fuse IC Others

Number 1 4 1 4 1 4 2 1 8 10 12 1 2 --

Failure rate per part

(failures/106Hours) 1.19 0.68 0.62 0.13 1.1 0.08 0.1 0.14 0.288 0.01 0.015 0.02 0.02 0.04

Total failure rate

(failures/106Hours) 1.19 2.72 0.62 0.52 1.1 0.32 0.2 0.14 2.3 0.1 0.18 0.02 0.04 0.04

Total failure rate of system: 9.49 (failures/106Hours)

532

V. REFERENCES [1] N. P. Papanikolaou and E. C. Tatakis, “Active Voltage Clamp in

Flyback Converters Operating in CCM Mode Under Wide Load Variation,” IEEE Transactions on Industrial Electronics, vol. 51, no. 3, June 2004.

[2] H. Chen, W. Dong, Y. He, and Z. Qian, “Secondary Side Post Regulation Application in Multiple Outputs Flyback Converter,” Power Electronics and Drives Systems, vol. 2, pp. 1273- 1277, 2005.

[3] C.-C. Wen and C.-L. Chen, “Magamp application and limitation for multiwinding flyback converter,” Electric Power Applications, IEE, vol. 152, pp. 517- 525, 2005.

[4] A.I. Pressman, “Switching Power Supply Design,” McGraw-Hill, 2nd edition, 1998.

[5] Watson, R. Lee, F.C. Hua, G.C. Bradley, “Utilization of an active-clamp circuit to achieve soft switching in flyback converters,” IEEE Transactions on Power Electronics, vol. 11, pp. 162-169, 1996.

[6] N. Mohan, "Power electronics: converters, applications, and design," Wiley, 3rd. edition, 2003.

[7] M. Brown, "Practical Switching Power Supply Design," Academic Press, 1990.

[8] S. Howimanporn, C. Bunlaksananusorn, “Performance Comparison of Continuous Conduction Mode (CCM) and Discontinuous Conduction Mode (DCM) Flyback Converters,” Power Electronics and Drive Systems, vol 2, pp. 1434 - 1438, 17-20 Nov. 2003.

[9] W. McLyman, "Transformer and inductor design handbook," Marcel Dekker, 3rd edition, 2004.

[10] A. Hren, J. Korelic, and M. Milanovic, “RC-RCD Clamp Circuit for Ringing Losses Reduction in a Flyback Converter,” IEEE Trans. on Circuits and Systems, vol. 53, no. 5, May 2006.

[11] Department of Defense, “Reliability Prediction Of Electronic Equip-ment,” MIL-HDBK-217F, Washington, DC., Notice 2, February 1995.

[12] G. Chen, R. Burgos, Z. Liang, F. Lacaux, F. Wang, J.D. van Wyk, W.G. Odendaal, and D. Boroyevich, “Reliability-Oriented Design Conside-rations for High-Power Converter Modules,” 35th Annual IEEE Power Electronics Specialists Conference, 2004.

[13] Datasheet of BYW99P/PI/W, “High Efficiency Fast Recovery Rectifier Diodes,” ST Corporation, Oct. 1999.

VI. BIOGRAPHIES

Arash Rahnamaee was born in Tehran, Iran on March 21, 1983. He received the B.Sc degree in electrical engineering from Tabriz University, Tabriz, Iran in 2005 and the M.Sc degree from the Amirkabir University of Technology (Tehran Polytechnic Uni.), Tehran, Iran, in 2008. His research interests include Design and Dynamic behavior of Switch Mode Power Supplies, Electric Motor Drives, and Design of Electrical Machines.

Jafar Milimonfared was born in Tehran, Iran in 1953. He received the B.Sc. degree in electrical engineering from Amirkabir University of Technology (Tehran Polytechnic), Tehran, Iran in 1978 and the M.Sc. and Ph.D. degrees in electrical engineering from Paris VI University, Paris, France in 1981 and 1984 respectively. Dr. Milimonfared joined the Amirkabir University of Technology as an assistant professor in 1984 where he is now a professor of electrical engineering.

His research interests include electrical machines analysis and design, power electronics and variable speed drives. He is with AmirKabir University of Technology as an Professor of Electrical Engineering.

Kaveh Malekian was born in Boroojen, Iran, on September 16, 1983. He received the B.Sc. degree in electrical engineering from Shahied Chamran University (Ahvaz Jondishapour Uni.) Ahvaz, Iran, in 2005 and the M.Sc. degree from the Amirkabir University of Technology (Tehran Polytechnic Uni.), Tehran, Iran, in 2008. His research interests include Power Electronics, Electric Motor Drives, Intelligent Controls, Design and Analysis of Electrical Machines, and Optimization.

533