SWITCHING POWER SUPPLY DESIGN:

DISCONTINUOUS MODE FLYBACK CONVERTER

Written by Michele Sclocchi
[email protected]
Application Engineer
National Semiconductor

Typical Flyback 3 output power supply:

*Output filter
L3

D3

V input
*Needed if Vin > 40V
Q2

C i1

C i2

N sp3

R1

Co 3a

D z1

N sp2

Cs
Isen

Co 2a

N sp1 Co
1a

Vin

Co 1b

Rc
comp
Cc
C f1

fb

fa/sync/sd

LM 3488

R fa
dr

R f1
agnd

pgnd

R sense

R f2

Notes:
Write down the power supply requirements on :
Xxx :=
Get the results on:
Rsults xx :=
This Mathcad file helps the calculation of the external components of a typical
discontinuous mode switching power supply.

Input voltage:
-Minimum input voltage:

Vi min := 30 volt

Maximum input voltage:

Vi max := 50 volt

Output:

o2

o1

Co 2b

L1

D1

C1

o3

Co 3b

L2

D2

Nominal output voltage, maximum output ripple, minimum output current, maximum output
current
Io1min := 0.250 amp
Io1max := 5 amp
Vo1 := 5 volt
Vrp1 := 10mV
Vo2 := 12volt

Vrp2 := 20mV

Io2min := 0.250 amp

Io2max := 1 amp

Vo3 := 12volt
(negative)

Vrp3 := 20mV

Io3min := 0.250 amp

Io3max := 1 amp

Vdfw := 0.6 volt ( diode's forward drop voltage)

Pomin
: Vo1
=
+ Vdfw Io1min + Vo2 + Vdfw Io2min +

Pomin = 7.7watt

Pomax :

Pom ax = 53.2watt

=
+
( Vo1

Vdfw Io1max +

( Vo2 +

( Vo3 + Vdfw) Io3min

Vdfw) Io2max + ( Vo3 + Vdfw) Io3max

Switching Frequency:
fsw := 150kHz
1
T: =
T=
6.667
sec
fsw
Transformer's Efficiency:

:= 0.95

1) Maximum Stress on the switching mosfet :

- Define the flyback voltage across the mutual inductance: Vfm
(Kfb is a value between 1 to 0.5)
Kfb :=
Vfm
: 0.8
Kfb
= Vi min
Vfm =
24volt
Nps1 : =

Vfm
Vo1 + Vdfw

Nps1 =
4.286

-Maximum
F
= Switching voltage on the switching-mosfet:
spike : 0.4
Vdsmax :

( F= spike + 1) ( Vimax +

Vfm

Vds max = 103.6volt

(assume spikes of 30% of Vdc )
Klk :
=
0.95
-The leakage inductance coefficient:

Safe factor

(Klk =0.95 means that the leakage inductance is 5% of the primary inductance)
-The total energy stored in the transformer: Wfly
(Energy delivered
to the outputs plus the Energy lost due to the leakage inductance)
1
Wlptot = 1.053
Wl ptot : =
Klk
Wl ptot Pomax
Wfly : =
Wfly = 3.73333 10 4 joule
fsw

2) Maximum and minimum duty cycle : Dmax and Dmin

To maintain the discontinuous mode of operation the maximum on-time has to be < 0.5 :
Ton+Tr+Tdt = T (On time+ reset time+ dead time = 1/ switching frequency )

-Choose minimum dead time duty cycle:

Ddt := 0.1

-Maximum drop voltage across the switching mosfet during the on time:
On resistance of the Mosfet:
Rdson := 0.06 ohm
Pomax
Vds on = 0.112volt
Vds on : =
Rdson
Vi min

Vfb :

Nps1
=
Vo1 + Vdfw

Tonmax : =

Tonmin : =

( Vimin

Vfb 1 Ddt T

Vds on Klk + Vfb

Vfb 1 Ddt T

( Vimax

Vds on Klk + Vfb

Tonmax = 2.748 sec

Tonmin = 2.017 sec

-Maximum
duty cycle
Tonmax
Dmax : =
T
Dmax = 0.412
-Minimum
cycle
Tonduty
min
Dmin : =
T

Dmin = 0.303

3) Primary current:
- Primary
peak
current:
2 Wfly
fsw
Ippk : =
Vimin Dmax
Ippk = 9.056amp
- Primary RMS current:
Ippk
Tonmax
Iprms :=

Iprms = 3.357amp
T
3
- Primary
current:
PoDC
max
Ipdc : =
Vimin

Ipdc = 1.867amp

- Primary AC current:
Ipac : =

Iprms Ipdc

Ipac = 2.79amp

4) Primary inductance:

Po T
2
Vomin Tonmax
Lp Ip
The energy stored is:E :=
=
and Ip :=
2
Lp
Edt : Vi
= min Tonmax
5
Edt = 8.245 10 voltsec

Primary inductance:

Lp : =

2 Wfly
2

Ippk

Lp = 9.105H

5) Secondary currents and turns ratios (secondary/primary) : Nsp1 & Nsp2

Nsp1 : =
Nsp2 : =
Nsp3 : =

1
Nsp1

Vo1 + Vdfw
Vfb

Nsp1 = 0.233

Vo2 + Vdfw
Vfb

Nsp2 = 0.525

Vo3 + Vdfw
Vfb

Nsp3 = 0.525
1

= 4.286

Nsp2

= 1.905

1
Nsp2

= 1.905

-Master output:
- Secondary
current:
Io1peak
max 2
Is1pk : =
1 Dmax Ddt

Is1pk = 20.503amp

- Secondary
Is1pkRMS current:
Is1rms : =
1 Dmax Ddt
3

Is1rms = 8.267amp

- Secondary AC current:
Is1ac : =

Is1rms Io1max

- Secondary2 inductance :
Ls1 : = Nsp1 Lp

Is1ac = 6.584amp

Ls1 = 0.496H

-First slave output:

- Secondary
current:
Io2peak
max 2
Is2pk : =
1 Dmax Ddt

Is2pk = 4.101amp

- Secondary
Is2pkRMS current:
Is2rms : =
1 Dmax Ddt
3

Is2rms = 1.653amp

- Secondary AC current:
Is2ac : =

Is2rms Io2max

Is2ac = 1.317amp

- Secondary2 inductance :
Ls2 : = Nsp2 Lp

Ls2 = 2.51H

-Second slave output:

- Secondary
current:
Io3peak
max 2
Is3pk : =
1 Dmax Ddt

Is3pk = 4.101amp

- Secondary
Is3pkRMS current:
Is3rms : =
1 Dmax Ddt
3

Is3rms = 1.653amp

- Secondary AC current:
Is3ac : =

Is3rms Io3max

- Secondary2 inductance:
Ls3 : = Nsp3 Lp

Is3ac = 1.317amp

Ls3 = 2.51H

6) Maximum Stress across the output diodes: Vdiode

Maximum voltage
present
the cathode of diodes
Vdiode1
= max
Nsp1 +onVo1
max : Vi
Vdiode1max = 16.667volt
Vdiode2max :

Vi
= max Nsp2 + Vo2

Vdiode3max :

Vi
= max Nsp3 + Vo3

Vdiode2max = 38.25volt
Vdiode3max = 38.25volt

Select a diode with Va-c>> Vdiode.max, and ultra-fast switching diode

7-a) Output ripple Specifications : Output Capacitors

- Secondary2 inductance :
Ls2 : = Nsp2 Lp

Ls2 = 2.51H

To meet the output ripple specifications without using an external LC filter, the output capacitors
have to meet two criteria:
- satisfy the standard capacitance definition: I=C*dV/dt where t is the Ton time, V is 25% of the
allowable output ripple.
- The Equivalent Series Resistance (ESR) of the capacitor has to provide less than 75% of the
maximum output ripple. (Vripple=Ipeak*ESR)
-Maximum
output
ripple:
Vrp1
=
10mV
Vrp2
=
20mV
Vrp3 =
20mV

-Minimum output
capacitance:
Tonmax
Co1 : = Is1pk
Vrp1 0.25

Co1 = 2.254 104 F

-Maximum
ESR
value:
Vrp1
0.75
ESR1 : =
Is1pk

ESR1 = 3.658 10 4 ohm

-Minimum output
capacitance:
Ton
max
Co2 : = Io2max
Vrp2 0.25

Co2 = 549.685F

-Maximum
ESR
value:
0.75
Vrp2
ESR2 : =
Io2max

ESR2 = 0.015ohm

-Minimum output
capacitance:
Ton
max
Co3 : = Io3max
Vrp3 0.25

Co3 = 549.685F

-Maximum
ESR
value:
0.75
Vrp3
ESR3 : =
Io3max

ESR2 = 0.015ohm

7-b) Output ripple Specifications with external LC filter :

In some cases to meet the output ripple specifications, it is more convenient to add a second
order
LC dB
filter and therefore reduce the size and cost of the output capacitors:
Att
:
=20

LC filter attenuation:
Tonmax
Co1a : = Is1pk
Vrp1 0.25 Att
ESR1a : =

Co1a = 1.127 103 F

1 Vrp1 Att 0.75

Is1pk

ESR1a = 7.316 10 3 ohm

-Second order filter:

Filter order loss per octave*
1
6
2
12
3
18
4
24
k:
=1000
Attenuation required (in dB):
Loss per Octave for filter:

Db :
=10
:=12

Frequency
at needed att. : (switching frequency)
fx
:
=fsw
Required cutoff frequency:

fx

fc : =

fc = 84.185kHz

Db

One of the critical factor

of a filter design is the attenuating character at the corner frequency.
The damping factor ( zeta) describes the gain at the corner frequency and the time response of
the filter. As the damping factor becomes smaller, the gain at the corner frequency becomes
larger.
For many filters, a damping factor of 1 and a cutoff frequency within about an octave of the
calculated ideal should provide suitable filtering.
Damping factor much greater than 1 may cause unacceptably high attenuation of lower
frequencies and a damping factor much less than .707 may cause undesired ringing and the filter
may itself produce noise.
(* octave= interval that has the frequency ratio 2:1 )
-For each output:
-Output load
Vo1 resistance:
Ro1 : =
Io1max

-Desired damping factor:

c:

:= .7

c = 5.289 105

2= fc

rad
sec

-Inductance
2 Ro1calculated:
L1 : =
c
L1 = 2.647H
-Capacitance
1 calculated:
Co1b : =
2
L1 c
Co1b = 1.35F
-Inductance used:

L1used := 10H

-Capacitance used:

Co1bused := 50F

i:
=
1 .5000

cu : =

1
2

L1used Co1bused

( i 500)

f i : = 100

250

wi : = f i 2
n : =

rad
sec
1

L1used Co1bused

si :

j= wi

nc : =

1
L1 Co1b

1 : =

L1used
2 Ro1

1c : =

1=
0.224

L1used Co1bused
L1

2 Ro1

As1 i : =

1c =
0.7

L1 Co1b

wi
1 + i 2 1

cu n

wi

Mags1i := 20 log As1i

As1c i : =

1
2
wi
1 + i 2 1c

c nc

wi

Magsc1i := 20 log As1c i

)
20

10

Magnitude, dB

10

20

30

40
10

-Capacitor Selection:
The performance of a filter critically depends on the capacitor used. Besides the basic voltage
and capacity requirements, select capacitors with low ESL, for high frequency attenuation, and
low ESR, for mid band attenuation and/or high ripple current capability.
-Choke Design:
Filter chokes should be designed to reduce parasitic capacity as much as possible:
-the input and output should be as far as possible.
-singular layer or Banked windings.
-use thicker than usual inter-layer insulation.
69
358
1247
(Banked windings)

8) Input capacitor:
The input capacitor has to meet the maximum ripple current rating Ip(rms) and the maximum
input voltage ripple ESR value.

9) Snubber Circuit:
The basic intent of a snubber is to absorb energy from the leakage inductance in the circuit.
The leakage inductance is part of the primary inductance that is not mutually coupled with the
secondary inductance. It is important to keep the leakage inductance as low as possible because
it reduces the efficiency of the transformer and it causes spikes on the drain of the switching
device.
A capacitor placed in parallel with other circuit elements will control the voltage across those
elements.
Usually :the0.03
2-5% of
the inductance
of the primary winding:
L
= leakage
Lp inductance isLleakage
= 0.273
H
leakage
Total energy:
1 1/2LI^2
2
Esnubber : = Lleakage Ippk
2
Esnubber fsw = 1.68watt

E snubber = 1.12 10 5 joule

There are different ways to dissipate this energy and reduce the spikes on the drain of the
switching mosfet.
A typical snubber circuit is a resistance and a capacitor connected in series between the input
voltage and the drain of the mosfet.
(approximately half of this energy has to be dissipated on the snubber circuit)
A good starting pointEsnubber
for the snubber capacitance could be:
Csnubber : =
Csnubber = 2.557 10 3 F
2 Fspike ( Vimax + Vfm)2
The RC timeTon
has
to be larger than the on time switching period: 1/RC<Ton
min
Rsnubber = 197.233
Rsnubber : =
4Csnubber
2

Prsnubber : = Vimax Csnubber fsw

Prsnubber = 0.959watt

For low output power applications, a clamp zener or a transient suppressor can be used as
shown on the flyback application of the LM3488 datasheet.

9) Switching Mosfet: Power Dissipation

The Mosfet is chosen based on maximum Stress voltage (section1), maximum peak input current
(section 3), total power losses, maximum allowed operating temperature, and driver capability of
the LM3488.
-Themax
drain
to source
Vds
= 103.6
volt Breakdown of the mosfet (Vdss) has to be greater than:
-Continuous
Drain current of the mosfet (Id) has to be greater than:
Ip
pk = 9.056amp
- Maximum drive voltage:
The voltage on the drive pin of the LM3488, Vdr is equal to the input voltage when input voltage
is less
than
Vdr
:
=
7.2

volt7.2V, and Vdr is equal to 7.2V when the input voltage is greater than 7.2V
Rdron :

7= ohm

-Total Mosfet losses and maximum junction temperature:

The goal in selecting a Mosfet is to minimize junction temperature rise by minimizing the power
loss while being cost effective. Besides maximum voltage rating, and maximum current rating, the

other three important parameters of a Mosfet are Rds(on), gate threshold voltage, and gate
capacitance.
The switching Mosfet has three types of losses, conduction loss, switching loss, and gate charge
loss:
-Conduction losses are equal to: I^2*R losses, therefore the total resistance between the source
and drain during the on state, Rds(on) has to be as low as possible.
-Switching losses are equal to: Switching-time*Vds*I*frequecy. The switching time, rise time and
fall time is a function of the gate to drain Miller-charge of the Mosfet, Qgd, the internal resistance
of the driver and the Threshold Voltage, Vgs(th), the minimum gate voltage which enables the
current through drain source of the Mosfet.
-Gate charge losses are caused by charging up the gate capacitance and then dumping the
charge to ground every cycle. The gate charge losses are equal to: frequency Qg(tot) Vdr
Unfortunately, the lowest on resistance devices tend to have higher gate capacitance.
Because this loss is frequency dependent, in very high current supplies with very large FETs, with
large gate capacitance, a more optimal design may result from reducing the operating frequency.
Switching losses are also effected by gate capacitance. If the gate driver has to charge a larger
capacitance, then the time the Mosfet spends in the linear region increases and the losses
increase. The faster the rise time, the lower the switching loss. Unfortunately this causes high
frequency noise.
n : = 10 9
Mosfet: Fairchild FQB10N20L- D2PAK
Rdson := 0.3 ohm
(Total resistance between the source and drain during the on state)
Coss := 95pF
(Output capacitance)
Qgtot := 13 n coul
(Total gate charge)
Qgdmiller := 6.1 n coul
(Gate drain Miller charge)
Vgs th := 2 volt
(Threshold voltage)
-Conduction losses: 2Pcond
Pcond : = Rdson Iprms Dmax
Pcond =
1.394watt
-Switching losses: Psw(max)
Turn On time:
Rdron
tsw : = Qgdmiller
Vdr Vgs th

t sw = 8.212 10 9 sec

Pswmax : = t sw Vds max Ippk fsw +

Coss Vds max fsw

Psw max = 1.232watt

- Gate charge losses: Pgate

to drive the gate
capacitor of the Mosfet:
Igateawg The
: fsw
= average
Qgtot current required
Igate
= 1.95 10 3 amp
awg

Pgate :

Igate
=
awg Vdr

Pgate =
0.014watt

-Total losses:
Ptot
= Ptot(max)
+ Psw m ax + Pgate
m ax : Pcond

Ptot max = 2.64watt

-Maximum junction temperature and heat sink requirement:

-Maximum junction
Celsiustemperature desired:
Tjmax := 130
-Maximum ambient temperature:

Tamax := 50

Celsius

-Thermal
resistance
Tjmax
Tamax junction to ambient temperature:
1
ja : =
ja = 30.305
Ptot max
watt Celsius
If the thermal resistance calculated is lower than that one specified on the Mosfet's data sheet a
heat sink or higher copper area is needed.
For Example for a T0-263 (D2pak) package the Tja of the Mosfet versus copper plane area is:

10) Current limit:

The LM3488 uses a current mode control scheme. The main advantages of current mode control
are inherent cycle-by-cycle current limit for the switch, and simple control loop characteristics.
Since the LM3488 has a maximum duty cycle of 100%, and the power supply is designed to work
in discontinuous mode with a 50% maximum duty cycle, the current limit should be designed so
that the peak
short circuit current limit is reached just before the 50% boundary is reached.
160mV
Rsense : =
Ippk 1.1
Rsense = 0.016

11) Transformer Design:

The inductor- transformer should be designed to minimize the leakage inductance, ac winding
losses, and core losses.
When the transformer is designed to operate in discontinuous mode the total inductance is lower
than in continuous mode, and the size of the transformer may be smaller. But the peak currents
will be at least twice the average current, therefore ac winding losses and core losses are the
predominant factors rather than the dc losses and core saturation. The total losses are minimized
when core losses and winding losses are approximately the same value.
-Core selection:
To reduce the core losses, ferrite-P material is usually the preferred material for discontinuous

flyback transformers with operating switching frequencies higher than 100Khz. (TDK PC40,
Philips 3C85)
The window shape of the core should be as wide as possible to minimize the number of layers
and therefore minimize the the ac winding losses and the leakage inductance.
E-type cores with an internal air-gap are the best choice for low cost and lower leakage
inductance.
- Winding techniques to minimize leakage inductance, ac losses and EMI noise:
To minimize the ac losses, leakage inductance and the EMI noise, particular attention has to be
paid to the design of the primary and secondary windings of the transformer.
The primary winding should be designed for less than three layers, thus minimizing the winding
capacitance and the leakage inductance of the transformer. In high switching frequency
applications an additional insulating layer between windings is usually used.
If the transformer has multiple secondary windings, the highest power secondary should be
closest to the primary of the transformer.
For high power applications, a split primary construction is typically used to reduce the leakage
inductance.
To avoid high ac winding losses due to the skin effect (at high frequency currents tend to flow
close to the surface of the conductor), Litz wire or Foil windings are typically used.
Litz wire for power applications is usually made with a few small diameter wires twisted together
in a strand, and few of these strands twisted into bigger strands.
Shielding tape or an additional winding between primary and secondaries is typically used to
reduce the capacitive coupling of common mode noise between primary and secondary. The end
of this additional winding has to be connected to ground or to the high input voltage of the
transformer.
To assist in the design of the transformer any application note from the transformer core
manufacturers can be used, or the "Transformer and Inductor Design Hadbook" written by
Colonel Wm. T. McLyman" or the "Magnetics Designer Software" from Intusoft may be used.
The required parameters for the flyback inductor-transformer design are:
-Core type and material: (TDK E core ferrite with internal air-gap)
-Switching frequency:
fsw = 150kHz

-Edt: volt-seconds
Edt = 8.245 10 5 sec volt

-Primary and secondaries currents:

Ippk = 9.056amp

Ipdc = 1.867amp

Ipac = 2.79amp

Is1pk = 20.503amp

Io1max = 5 amp

Is1ac = 6.584amp

Is2pk = 4.101amp

Io2max = 1 amp

Is2ac = 1.317amp

Is3pk = 4.101amp

Io3max = 1 amp

Is3ac = 1.317amp

-Primary and secondarys inductance:

Lp = 9.105H

Ls1 = 0.496H

Ls2 = 2.51H

Ls3 = 2.51H