CONTENTS

Diode is the basics power semiconductor

Structure, symbol, and basic nature of diode

Rectifier diode for general use, and fast recovery diode

Experiments to ascertain basic nature of diode Forward and Reverse characteristics

Forward power loss and reverse power loss

As for PN diode, we can ignore reverse-power loss, but .

Diode as a Temperature Sensor

Packages of Power Semiconductors and Nominal Current

Inside the Package

Chip Size and Nominal Current

Reverse Power Loss of large chip SBD is notably high.

Resistance of Current Path and of Bonding Wire

What reverse voltage is your best choice?

Rectification Circuits and Diode VRRM

Rated Reverse Voltage

10

Thermal resistance and Allowable current

13

Transient thermal resistance

13

How power semiconductors break down

15

Surge current ratings of diode

15

I2t and I2t

16

I2t is now obtained by Spice simulation.

16

Switching time of diodes -- reverse recovery time

19

Differentiating SBD and FRD

19

Why are fast recovery diodes needed?

20

Commentary: Rectification noise

22

Types of SBDs and FRDs

24

Thermal runaway of SBD

25

Conclusions

26

Characteristics and Applications

of Power Diode
Diode is the basics power semiconductor
Handling current of power semiconductor is
generally more than 1 A. Among power semiconductors, power diode is the simplest one,
and here we call it simply diode. Upper operating frequency is below MHz, and major usage is for rectification.

Structure, symbol, and basic nature of

diode
At present, almost all of diodes are silicon-based
semiconductors. One of diode structures is PN
junction type, where P layer is formed in N type
silicon. The other is schottky type, where metal
(barrier metal) is piled onto N type silicon. In
addition, mesa and planar both belong to the
family of PN junction type..

Current

Anode

Cathode

Power Diode

1A

Cathode

Anode
Small signal Diode

Hz

kHz

Current Flow

MHz

Symbol and Nature of Diode

Diode has 2 external terminals; P side (barrier
metal side for schottky diode) is anode, and N
side is cathode. The circuit symbol comes from
the nature that current flows solely from anode to
cathode.
The direction which current flows is the forward
direction of diode. On the other hand, if positive
voltage is applied to cathode, current hardly flows
before the voltage reaches at a certain value. This
voltage limit is called rated reverse voltage. For
example, as for 200V diode, a reverse voltage

GHz

Frequency
In many cases, diodes are used with MOSFET,
bipolar-transistor and IGBT, and we call these
transistors simply transistors.
P layer
N-type silicon

Planar

P layer
N-type silicon

Anode

Mesa

Barrier Metal (e.g. Mo)

N-type silicon

Cathode

Current doesnt flow below 200V.

Schottky
Current flows over 200V.
(Nature of diode is lost.)

Structure of Diode

Diode of VRRM = 200V

April, 2004/June, 2005 S. Hashizume

higher than 200V forces to lose the diode nature

that current doesnt flow in reverse direction.
Here, loss of nature means that eternal destruction may possibly occur. We say that reverse
applied voltage must not exceed the rated reverse voltage even for a shortest possible period.
The following figure is section of MOSFET. As
shown in this figure, it is the PN junction that
owes drain-source voltage of the MOSFET.
Unless a gate bias is supplied, the drain refers to
anode, and source to cathode, so this composition is exactly same as diode. (N-channel
MOSFET) Moreover, PN junction is responsible for off-voltage blocking capability also in
bipolar-transistor and IGBT. That is to say, to
know reverse characteristic of diode is linking to
understand off-characteristics of these transistors.
PN Junction

Source

General Use
AC to DC
AC100/
200V

Fast Recovery
DC12V
DC to DC
DC to DC
DC to DC

50/60Hz

DC5V
DC3.3V

More than tens of kHz

ATX Power Supply

popular than before. For example, checking
ATX power supply for desktop PC, we found
only one general use bridge rectifier diode, 14
pieces of FRD, and 10 pieces of SBD. Counting
the number of diodes, we understand that this
comes from the fact that operating frequency of
AC to DC converter is low (50/60Hz), but that of
other circuits is as high as more than tens of kHz.
Moreover, SBDs are especially useful for low
voltage power supplies such as for battery operating equipments.
Compared with transistors, rectifier diode for
general use and FRD (PN diode) correspond to
bipolar-transistor, and SBD corresponds to
MOSFET. Because nature of SBD differs much
in some points from that of PN diode, when
applying it, you should master specific characteristics of SBD. While paying attention on
these points, we will study diode, and as well as
power semiconductors. By the way, we will
later discuss fast recovery diode in detail.

Gate

Drain

Section of MOSFET
Principal capabilities of power semiconductor
are rated current and rated voltage, such as 10 A
and 200 V. In addition to diode, basic idea for
current handling capability is common to all
power semiconductors. So, seeing a standard
package device, such as TO-220, with no identification information, we cannot distinguish in
appearance whether it is diode, or MOSFET, or
others. For semiconductor manufacturers, even
if assembled chip is different, package design
and manufacturing technology are common. We
will mainly discuss diode, but the story is also
valid to all the power semiconductors.

Experiments to ascertain basic nature of

diode Forward and Reverse characteristics
Let's check the most basic fact that diode conducts current in forward direction but not in
reverse direction.
Firstly, we measure forward and reverse characteristics of both 31DF2 and 31DQ04, which are
FRD (3A 200V) and SBD (3A 40V), respectively.

Rectifier diode for general use, and fast

recovery diode
In general, diode corresponds to rectifier diode
for general use which is mostly applied to 50/60
Hz AC to DC conversion. However, at present,
many fast recovery diodes including SBD
(Schottky Barrier Diode sometimes called
simply schottky), are becoming more and more

31DF2 3A 200V FRD

31DQ04 3A 40V SBD
(Same package

31DQ04 (SBD) Forward Characteristics at 25

Oven

Measuring Forward and Reverse Characteristics

Forward Current

Reverse Voltage

IF

1mA

10mA

0.1A

3A

10A

VF

0.182V

0.242V

0.305V

0.448V

0.659V

182

24.2

0.149

66m

31DQ04 (SBD) Forward Characteristics at 125

IF

1mA

10mA

0.1A

3A

10A

VF

0.0175

0.0699

0.150

.0.379

0.611

17.5

6.99

1.5

0.126

61m

31DQ04 (SBD) Reverse Characteristics at 25

To begin with, we measure forward-current IF vs.

forward voltage VF of 31DF2 at 25 C and at 150
C.

VR

1V

5V

10V

20V

40V

IR

1.3A

2A

2.8A

4.4A

12A

R(M)

0.76

2.5

3.6

4.5

3.3

31DF2 (FRD) Forward Characteristics at 25

31DQ04 (SBD) Reverse Characteristics at 125

IF

1mA

10mA

0.1A

3A

10A

VR

1V

5V

10V

20V

40V

VF

0.466V

0.535V

0.630V

0.842V

0.985V

IR

2mA

2.7mA

3.3mA

4.6mA

9.1mA

466

53.5

6.3

0.28

98.5m

R (k)

0.5

1.8

2.9

4.3

4.3

31DF2 (FRD) Forward Characteristics at 150

IF

1mA

10mA

0.1A

3A

10A

VF

0.217

0.308

0.414

0.654

0.821

210

30

4.1

0.22

82.1m

An example of Forward Resistance

FRD (PN Diode) Several to less than 1
SBD
Half of FRD

Using the same diodes, we also measured reverse

voltage VR vs. reverse current IR at 25 C and 150
C.

An example of Reverse Resistance

FRD (PN Diode)

31DF2 Reverse Characteristics at 25

VR

1V

10V

50V

100V

200V

IR

1nA

1nA

1nA

2nA

3nA

1,000

10,000

5,000

5,000

6,700

R (M)

SBD

31DF2 Reverse Characteristics at 150

VR

1V

10V

50V

100V

200V

IR

5.4A

6.6A

8.2A

9.7A

14.4A

R (M)

0.186

1.51

6.04

10.2

13.9

5,000M 25
10M 150
5M 25
5k125

1mA at VR=5V
Compared with FRD, we have following information from these measurements.
Here, rated reverse voltage of SBD is 40V,
which is much smaller than that of FRD (200 V),
but forward resistance of SBD is about 1/2 of
FRD.
At high temperature, reverse resistance of SBD
is so small as in 1 digit k. And the resistance is
not enough large for reverse current to be ignored
like PN diode.
As temperature rises from 25 C to 125 C, VF
at 3A becomes by 0.85 times smaller and IR at 40V
becomes by 760 times larger.

To summarize these measurement;

In usable current area, forward resistance is
several at small currents, and less than 1 at
larger currents.
The reverse resistance is as high as several M
even at the elevated temperature at usable voltages.
In the region where forward current or reverse
voltage is extremely small, both and are not
true.
When temperature rises from 25 C to 150 C,
VF at 3A becomes by 0.78 times lower and IR at
200 V becomes by 4,800 times larger.

Low forward-voltage is valuable feature of SBD.

On other hand, it has a weak point that reverse
current is large. When applying SBD, It is important for circuit designers that this unfavorable
characteristic is not to come to light.

Next, let's see the result of similar measurements

on SBD of 31DQ04.

Forward power loss and reverse power loss

As for PN diode, we can ignore reversepower loss, but .
The loss due to forward current is forward power
loss, and the loss due to reverse voltage is reversepower loss. The smaller the total losses of the two
is, the lower the temperature rise is.
Based on the result of measurement, we attempt to
calculate losses of 31DF2 and 31DQ04. We suppose that forward-current is 3A with 50 % duty
and reverse voltage is 20 V for 31DQ04, or 100 V
for 31DF2 during remainder of each cycle.

Forward
Current

However, for SBD like 31DQ04, because the ratio

of reverse power loss is high, your thermal design
should include this loss. The reverse power loss
influences temperature rise, and it could possibly
lead to thermal runaway.

3A

100V (31DF2)
20V (31DQ04)

Reverse
Voltage

Forward
Loss

Reverse
Loss

Ratio of Reverse
Loss to Total Loss

31DF2

0.981W

0.00049W

0.05%

31DQ04

0.568W

0.046W

7.5%

Note : Forward losses and reverse losses are calculated using actual values at 150 C for 31DF2
and at 125 C for 31DQ04.
For PN diode like 31DF2, even if duty of reverse
voltage would be as high as 100 %, the loss is so
small as about 1mW, so we may ignore this loss.

Diode as a Temperature Sensor

Forward voltage of diode linearly varies with
temperature. Utilizing this nature, we can use
diode (or B-E of bipolar-transistor) as a temperature sensor. Below is the forward voltage characteristic of 10EDB20, 1A 200V diode.

Forward Current (A)

coefficients are;
At 10A
-2.22mV/
At 100A
-2.03mV/
As a result, the coefficient is about -2mV/ at
1/10,000 of rated current. Likewise, we can
know chip temperature of power semiconductors
by flowing one small current.
In temperature monitor ICs available from
Maxim, National semiconductor and others, two
currents method is used.

Small current
region is good
for a sensor.

IFW

Forward Voltage (V)

VF

Forward Characteristics if a Diode

Because heat generation by current prevents correct temperature sensing, a small current is employed. Generally, temperature rise of 1C results in decrease in forward voltage by 2 mV.
Let's exucute actual measurement on 10EDB20.
IF(A)

27

50

75

80

Measure Forward Voltage at 2 currents

Supposing VF is the difference in forward voltage for two currents at a specific temperature;
T= q VF/[n k ln(N)]273.15
Using this equation, we can calculate T().
Constants q, n, and k are;
q : 1.60210-19coulomb
n : Ideality Factor
k : Boltzman constant 1.3810-23
Referring the table in the left, difference in VF
between at 10A and at 100A is 72mV at
75. We have n of 10EDB20 separately, and
it is 1.0293. Calculated VF is 71 mV using
the above relationship. So, two VF obtained
by measurement and by calculation are well
agreed.
You may understand that seeing "the difference"
is a cleverest ways.

85

345.2

292.9

236.0

224.7

213.2

10

361.6

310.3

254.2

243.1

232.4

25

388.3

337.5

282.6

271.5

259.9

50

403.5

356.2

304.0

293.4

283.1

100

423.5

376.5

325.8

315.6

305.7

250

449.3

403.9

354.2

344.1

334.2

300

454.1

408.9

359.8

350.0

340.5

NIFW

Forward Voltage as a function of Forward

Current and Temperature (Compensated
In the table above, forward voltage is compensated by linear regression. Obtained temperature

Packages of Power Semiconductors and

Nominal Current
Typical packages of power semiconductors and
their names are shown below. It's true that package names are a little confusing; some are specified by JEDEC or EIAJ, unique by manufacturer,
or popular intermingle. In countries except for
Japan, naming tends to be common among manufactures, but, in Japan, we find a lot of manufactures original. Even if a package is standard one,
dimension has a some band. So, when you apply
an identical package of another manufacturer's in
place of your original, you will sometimes have
difficulties such that lead terminal cannot be inserted into hole of PCB, or ferrite bead cannot be
equipped. Check outside drawing of the product
very carefully.
Here, we discuss mainly power semiconductors of
less than 100A, but those of thousands amperes
are also available, which are used in highest power
electronics and electric railroad applications.
The ancestor of plastic package power semiconductor is axial diode. (For example, 10DDA; its
original is 10D).

When assembling axial, TO-220, TO-247 and

other through-hole devices on PCB, their lead
terminals are inserted into through-holes and soldered. In contrast, lead terminals of surface
mounting devices (SMD) are soldered onto soldering pad on PCB. For 50/60Hz rectification,
bridge type or axial type diodes are still major
players. However, surface mounting devices are
continuously increasing their application areas.
Inside the Package
The photo is an example of internal structure of
TO-220 full-pack package, where two diode chips
are loaded on copper frame. Popularly, chips and
two external leads on both sides are connected
with aluminum wires. After soldering chips on
frame and wire bonding, they are covered with
resin mold.

Chip

Aluminum Wire

Copper Frame

TO-263 package MOSFET on PC mother board

TO-220 Full Pack

TO-247

TO-220
Bridge Diode
Axial

TO-262
TO-251

TO-252

SOD-123

SMA

Power Module
TO-263

Wire bonding machine is one of general manufacturing facilities, and, inside of a variety of
packages, they are applied to connect between
chip and external terminal. However, if a lower
wiring resistance, a better heat transfer, or a thinner dimension is required, copper connecting bars,
instead of wire, are used to achieve such requirements.

There is an important point to evaluate loss. It is

not at 25 C, but at high temperature that loss
calculation should be based on. For diodes, generally VF at high temperature is specified by
graph. Unless such direct information is supplied,
considering temperature dependency, we should
estimate the loss based on value at 25 C.
These two, 31DF2 and 31DQ04, are 3 A diodes.
Pay attention also to VF at 1 A and 10 A, together
with at 3 A. At 1 A, current densities is 1/3 compared with 3 A. Also, at 10 A, current densities is
3.3 times larger compared with 3 A. These values
are indication of predicted forward voltage when
we apply 0.9A or 9A diode in place of 3A diode.

Chip
External Lead

31DF2 (FRD) Forward Voltage at 150

Inner Structure of an Axial Diode

This figure shows inner structure of an axial lead
diode. Chip is soldered between two external
leads, and sealed with resin.

10

1A

3A

10A

VF

0.560

0.654

0.821

0.85

1.25

31DQ04 (SBD) Forward Voltage at 125

Chip Size and Nominal Current

1

IF

IF

1A

3A

10A

VF

0.269

.0.379

0.611

0.71

9A Diode

Reference

1.61

0.9A Diode

Prediction of Forward Voltages

1 Amp.
chip

3 Amp.
chip

As such, even if you replace a diode to one which

has current capacity of 3 times larger, forward loss
is not decreased by 1/3. Similarly, it doesn't increase by 3 times by a diode of 1/3 current capacity. As regard to SBD, chip size is more influential compared with PN diode.
This information will be useful when you choose
appropriate one among a series of diodes in a same
package.

10Amp. chip

For example, diodes of from 3 A to 30 A are

available in TO-220 package. As shown in the
image above, current densities is designed to be
identical, chip of same effective area which is
proportional to nominal current is assembled in
these products. The effective area, which is the
area for current flow, roughly regards as whole
area of chip.
Well, how does forward power loss change according to chip size? If temperature rise of 3A
diode would be too high, you will choose larger
diode such as 5A or 10A in place of the original.
In this case, this information would be useful to
estimate which degree the loss can be reduced.
Let's review the result of measurement at the end.

Reverse Power Loss of large chip SBD is

notably high.
Reverse current is proportional to chip area, and
that of FRD is so small that its reverse loss can be
ignored. However, reverse current of SBD is
high, and it increases with increasing chip area, so
the loss must be incorporated into loss calculation.
Measured reverse current of 3A 40V SBD at 125
C was 9.1mA. It means that expected reverse
current of 9A SBD, which has larger area of 3
times, is 27.3mA. If 100 % duty of 40 V is applied, the loss is so large as 0.7 W. As we knew

31DF2 (FRD) Forward Voltage at 150

IF

10mA

0.1A

1A

3A

10A

VF

0.308

0.414

0.560

0.654

0.821

R ()

30

4.1

0.56

0.22

0.08

Expected reverse current of 31DQ04 (SBD) at 125

31DQ04 (SBD) Forward Voltage at 125

IF

10mA

0.1A

1A

3A

10A

VF

0.0699

0.150

0.269

.0.379

0.611

R ()

6.9

1.5

0.26

0.126

0.061

Nominal Forward Current

9A

3A

0.9A

IR (mA) at 40V 125

27.3

9.1

2.73

Reverse Loss (W) at 40V

1.092

0.364

0.109

Duty : 100%

before, forward voltage would decrease by 0.71

times, total loss may be equal or possibly increased, depending on operating condition. Such
study would sometimes be important when using
SBD.
Resistance of Current Path and of Bonding Wire
The resistance of copper pattern with 35 m
thickness, 1.4 mm width, and 10 mm length is 5
m at 125 C. This is a resistance which doesn't
appear in circuit diagram.
Let's check resistances of lead wire and bonding
wire in power semiconductors.
MOSFET as large as 180A, and diode of tens of
amperes, are introduced in TO-220 package,
Even not in such an extreme one, resistance of
lead wire and bonding wire is one of principal
elements to determine current handling capability
of power semiconductors.
The physical constants of copper (Cu) and aluminum (Al) are as follows.
Resistivity
(cm) 20

Temperature coefficient
Referred to 20

Cu

1.72

0.00393

Al

2.82

0.0039

2.2m

Resistance of aluminum wire of 400m,

10mm 1.41 times at 125
0.55m

Chip

Frame

Resistance of Current Path in TO-220 Package

RR0 1+ (T 20)
At first, let's calculate resistance of Cu lead of
axial type using dimensional information.

60A into TO-220

Package

60A

2.36m

0.5mm
L=27mm

0.6mm
L=27mm

1.64m

0.8mm
L=27mm

0.92m

0.52m

2.2m

0.55m (per one lead

33mV 1.98W

voltage drop is 33 mV. What's more, the loss here

is as large as 1.98 W! If forward voltage of diode
is 400 mV, 33 mV is equivalent to 8.25 % of it.
These values are not so small as to be neglected.
Measuring point of lead terminal is specified at the
point nearest to resinous body, so you will have
incorrect value if measuring point is end of terminal lead. Also, we have understood that long lead
is not preferable for high current applications. In
addition, an example of lead resistance of a bigger
TO-247 package is 0.26 m, and it is approximately half of TO-220.
Today, highest current MOSFETs are available,
and their on-resistance, for example, is less than
4m. And diode of 200 A is introduced. What
connecting methods are employed between chip
and external terminal? An important design criteria is to maintain the ratio of wiring resistance to
chip resistance to less than a certain rate. One
practical solution to get lowest possible resistance

0.23m

1.4mm
L=21mm

Resistance of Cu Lead of Axial Diodes

1.41 times at 125

Calculated value above is resistance of whole

lead in one side at 20 C. At 125 C, it increases
by 1.41 times.
Next item is aluminum wire of 400m in diameter. Supposed length is 10mm. Shown is an example which calculates resistance of current path
in TO-220 package.
You may feel that the resistance of m is small
enough. However, how much is the voltage drop
when current of 60A flows into 0.55 m? The
8

is to use maximum diameter aluminum wires,

such as 500 m phi, simultaneously, to use maximum possible pieces, e.g., three pieces. As a
result, wire resistance can be reduced as small as
1/5 compared with the previous example. Another
choice for manufacturer would use copper bar
instead of wire. Furthermore, for power semiconductors, like multiple-pin ICs, solder bump
technology is becoming popular these days. To
obtain minimum area and minimum height of
package, CSP (Chip Scale Package) may be a final
goal.
Thus we conclude that it's true that even a smallest
resistance such as 1 m can never be neglected for
power semiconductors.

22 e

e 22 e
Current
flows
alternately.

What reverse voltage is your best choice?

At low temperature, VRRM decreases.
Rated reverse voltage VRRM of diode, and off state
voltage of transistor becomes lower with lowering temperature. Be careful whether prescribed
temperature for reverse voltage rating is at whole
of the operating temperature range (for example,
-40 C to 150 C), or only at 25 C. Recently, we
sometimes find the latter. If provided only at 25
C, you should consider what is your lowest temperature. Generally, temperature coefficient is
+0.1 % per C. This means that rated reverse
voltage declines by 6.5 % at -40 C compared with
at 25 C, so rated reverse voltage of 200V decreases to 187V. (200-0.065*200)

e
22 e

Current flows alternatively.

2 e

Temperature
Low

High

Reverse Current
Reverse Voltage
Temperature Dependency of Reverse Voltage

At a lower temperature, switching speed of transistor generally becomes faster. Corresponding to

faster switching, accompanied surge voltage becomes larger.
Consequently, we have understood the implication of cold operating test, and the checkpoint
(observing transient voltage).

e
e

Rectification Circuits and Diode VRRM

Rated reverse voltage for diode in rectification
circuit of commercial frequency (50 / 60Hz) is

22 e

Filter
AC line filter is effective to suppress noise both from outside
and from inside the power supply. In switching power supply,
the latter is more significant.
There is no doubt that such a
strict filter is extremely effective
to foreign surge voltage and
current.

Rectification Circuit of ATX Power Supply

discussed here. Shown on the previous page are
four typical rectification circuits, and, referred to
input voltage, how many times of voltage is
needed for diode VRRM (rated reverse voltage). It
is the actual diode VRRM that includes voltage
fluctuation, surge voltage, and some margin. For
example, supposing AC 200V input for bridge
rectification circuit, 2 times 200V is 282 V.
And, if 1.2 times for voltage fluctuation is assumed, then peak voltage becomes 382V. Additional margin should be entrusted to user, but
diode of 600V or 800V is the general choice in
this case, so, margin is equivalent to 1.5 through
2. An 800V diode for AC200 V input, and a
400V diode for AC 100V input, are common
choices through these 4 circuits.
Remember that these choices may have not
enough margin unless there is transformer or filter
between power supply line and rectification circuit. It is because we cannot guess how a lager
foreign surge voltage comes in. Actually, every
rectification circuit has some filter or transformer
in most cases, this may not be practical discussion.
However, in extreme cases, this is not always very
safe, so, you should care how large is reverse
voltage applied to diode.

slightly larger stress than rated value because

diode has some margin.
As shown, surge voltage must not be larger than
the rating even for a shortest period such as 0.1s.

Applied
Voltage

Rated VRRM
of Diode

Rated and Applied Reverse Voltage

By the way, some diodes are guaranteed the rating
for 500s width under a specific condition. Confirm with the data sheet individually.
Based on supply voltage, we cannot know how
many time the VRRM of diodes is needed. In rectification circuit at the end, we added with some
changing parts to supply voltage. In switching
circuits, we should pay attention to inductive
transient voltage which appears when current is
shut down.
The voltage which occurs at the inductance is
function of inductance and switching speed;
Ldi/dt. For example, 10H times 10A/s equals
to 100V. A summation between this voltage and
supply voltage is the applied voltage to diode.
But, this voltage is influenced by stray capacitance, so calculated voltage may have some error.

Rated Reverse Voltage

Other than rectification circuit, to what we should
pay attention, and, as well as, how to choose VRRM
of diode? Most significant point for us is that
maximum voltage applied to diode during operation must be less than rated repetitive reverse
voltage VRRM.
This figure shows relationship between peak inverse voltage applied to diode and rated reverse
voltage specified in data sheet. Specifically
speaking, 40V SBD can be applicable when applied peak voltage is 38V. In another case, 40V
SBD is not applicable because the voltage exceeds
40V, so our choice may be 60V SBD.
The word, rated voltage, means that "excessive
stress over this value causes failure of the device".
Actually, diode may not immediately fail due to

When
current is
switched off

Surge
voltage is
generated

Inductive Surge Voltage

10

Higher frequency operation wins better voltage

regulation, but, for this purpose, inductance and
capacitance must be reduced. Whereas, higher
switching speed leads to higher di/dt. As a result,
transient surge voltage applied to diode becomes
larger. Keeping this point in mind, we should
check operating voltage carefully.
To suppress inductive voltage, our actions include
slowing down switching speed, and minimizing
stray inductance as small as possible.
The inductance does not always appear in circuit
diagram. Let's study an example. Below is DC to
DC step-down converter of 15V input and 5V output. At first, let's simulate diode current and voltage in theoretically ideal circuit. Because there is
no inductance related to transient voltage, supply
voltage, 15 V, is applied to diode as reverse voltage
just as it is.
Now, how about in real circuit? Here, we add stray
inductance of 0.1H in series with transistor, and

IRL3103 2

we carry out simulation once again. In this case,

we see transient voltage added to 15V like actual
circuit. Compare with the basic diagram how
transient voltage is generated, we find that this
stray inductance is the major source of transient
voltage. Therefore, in order to decrease the element L out of L di/dt, we try to make circuit pattern as shorter and thicker as possible. By the
way, 100H in circuit diagram is not the "L" here.
The circuit on the next page is DC to DC boost
converter of 3.6V input and 12V output. Red-dot
is put on the inductance which is source of inductive voltage. Close observation on peak transient
voltage during operation gives you the idea which
rated reverse voltage is enough for your application.

100H 4

Diode Current
5V

15V

47F

EC10QS03L

10F

10

15V

40

Diode Voltage

0.1

Diode Current and Voltage by Simulation

15V to 5V Step-down DC to DC Converter

IRL3103 2

Diode Current

100H 4

0.1H

15V

47F

5V
3

EC10QS03L

Diode Voltage

10

10F

40
0.1
Diode Current and Voltage by Simulation
(Stray Inductance of 0.1H is added)

Stray Inductance is added

11

EC21QS04
30H
0.1H

3.6V
47F

12V
10F
24
0.1

Step-up DC to DC Converter and the L

12

Thermal resistance and Allowable current

The temperature of semiconductors while they are
operating is intimately tied in with the trouble that
the semiconductors may or may not experience as
well as their service life and reliability. The following questions therefore arise: in what way should the
semiconductors be operated under what cooling
conditions and how many degrees Celsius represent
the maximum temperature rise? It is the thermal
resistance that is used to calculate the temperature
rise. It is therefore necessary to implement the kind
of thermal design which will hold the temperature
down below the prescribed level under 150C, for
instance under the most challenging conditions.
After ensuring that the circuitry will operate reliably,
introduce thermal design that is oriented toward
ensuring reliability.
The higher the current flow, the greater the diode
loss and, accompanying this, the higher the temperature inside the device itself. Lets look at two
examples in which the temperature rise has been
calculated from power loss and thermal resistance.
Junction temperature: 136C
(calculated value)
Lead temperature: 90 C
(actually measured value)

second example, lets now rewrite the example of the

thermal resistance and temperature rise.

10W
Tjmax
Tj

Tc

Tf

Example of junction temperature calculation (1 of 2)

(The junction temperature is calculated from the lead temperature
using an axial lead diode.)

Fin temperature
90C (actually measured value)

135

Rthj-c

3/W

Rthc-f

1.5/W

105

90

The temperature difference can be calculated in this

way using the thermal resistance. If there is a figure
for the thermal resistance between junction and ambient, the thermal resistance for the junction can be
calculated using the ambient temperature as the
reference. The objective of the approach used here is
to calculate the temperature increase using either
measured temperature or design temperature as the
basis, and thereby ascertain the junction temperature.
The case temperature is not equivalent to the package surface temperature. An example of case temperature for T0-220 and T0-247 is shown below. If
the position of the reference point is unclear, verify it
by contacting the manufacturer.

Diode loss: 2W, continuous

Thermal resistance between junction
and lead: 23C/W

Case temperature
103C (calculated value)

150

Correlation between thermal resistance

and temperature difference

Difference
in
temperature:: 46C

Junction temperature
135C (calculated value)

Temperature specified on data sheet

Thermocouple

Difference
In temperature
:: 32C

Chip

Package

Frame

Difference
In temperature
:: 13C

Fin (Heatsink)

Examples of case temperature

(cross-section diagram)

Diode loss: 10W, continuous

Thermal resistance between junction and case: 3C/W
Thermal resistance between case and case: 1.5C/W

Transient thermal resistance

The thermal resistance described above is the resistance under conditions where its temperature is
completely saturated or where it is in respect of the
direct current, and this is referred to as the steadystate thermal resistance. In contrast, the transient
thermal resistance rth is the thermal resistance vis-vis the pulse time (t0 in the figure). The junction
temperature rise Tj can be calculated from power
loss P and rth using the following equation.

Example of junction temperature calculation (2 of 2)

(The junction temperature is calculated from the fin temperature
using a TO-220 diode.)

Depending on what should serve as the reference for

the thermal resistance, a multiple number of thermal
resistances may be specified for a single device. In
the first calculation example, the thermal resistance
between junction and lead is used; in the second
example, the thermal resistances between junction
and case and between case and fin are used. With the

Tj = rthP

13

The calculation examples given above applied to one

-shot pulses. With continuous pulses, the pulses
preceding the last two pulses (or last pulse) are handled as the average, and when the temperature is
calculated, this average is combined with the last two
pulses to obtain the peak temperature rise.
In the following example, since the duty ratio is 1/10
at a peak power of 10W, the average power is 1W.
The transient thermal resistance is applied to three
power levels the average power of 1W, the 10W
power of the second to last pulse of 1 ms and the
10W power of the last pulse of 1ms to calculate
the temperature rise.
The temperature rise Tj at the point where temperature is highest can be calculated using the following equation.

Power

Change in
temperature

Time

Definition of transient thermal resistance

Calculating the transient temperature rise using the
transient thermal resistance is valid for square wave
power pulses. In almost all cases, however, the
actual current waveforms are far from these pulses.
How this difference is to be handled will be described below.
Sinusoidal wave and triangular wave currents are
handled as follows. The square wave power for the
identical surface area is substituted for the peak
power.

Tj1(W)Rthr11ms10(W)r11msr10msr1ms)

The meanings of each term are shown in the figure

Look at the equation and figure, and compare them.
Rth is the steady-state thermal resistance while r11ms,
r10ms and r1ms are the transient thermal resistances of

Sinusoidal current
Point at which temperature is calculated

Diode power loss

1W

Equivalent power

1(W)Rth
-1W

Time

-1(W)r11ms

10W

10(W)r11ms

Sinusoidal current handling

Triangular current

-10W

-10(W)r10ms

Diode power loss

10W

10(W)r1ms

Equivalent power

Time

Example of calculating temperature rise

using transient thermal resistance

Triangular current handling

14

11ms, 10ms and 1ms, respectively.

This calculation method is used for all power semiconductors. It is used to determine whether the
junction temperature comes within the rating that
is to say, whether semiconductors can be used from
the perspective of the temperatures involved.
If, for instance, the pulse width is only 100s or less,
it is often quite acceptable to ignore the portion
equivalent to the temperature ripple as calculated
here and to hold down the average junction temperature simply on the basis of steady-state thermal
resistance and average loss. This is because the
temperature ripple is small enough for short pulses.

<3> Short-circuiting occurs.

Short-circuiting accounts for almost all breakdowns.
<4> An explosion occurs
When a level of energy several times higher than the
threshold breakdown level is applied, the plastic
package may explode into pieces which will be sent
flying. This breakdown method is found only in
power semiconductors. If the energy during breakdown become excessively high, it must be limited
Furthermore, plastic packages should never burst
into flames and burn. This is because the resin used
is flame-retardant or it has self-extinguishing properties

How power semiconductors break down

What happens if the current flowing to a power
semiconductor is too high or if an excessively high
reverse voltage is applied to a power semiconductor?
Silicon chips melt at a temperature of about 1400C
so they will definitely short out if their temperature
rises above this figure. When the word breakdown
is generally used in connection with power semiconductors, the event described falls into one of the
following categories.

Surge current ratings of diode

With transistors, the pulse width and duty ratio
change, and when modulation is further added to
these factors, the whole question of surge current
ratings becomes more complex. For this reason, the
surge current rating is specified for some devices
(such as MOSFETs) in such a way that a level of up
to four times, for instance, the continuous current
level is allowed. In this case, there is no limit on the
maximum number of times the device can carry the
current in question so this is referred to as the continuous current rating. In contrast what is given on
the data sheets of diodes is the non-repetitive surge
current rating which generally limits the maximum
number of times the device can carry the current in
question to one time. A once-only limit presupposes
circumstances such as times when failures occur
under which the level will only rarely be exceeded.
Although the device will not break down even when
this current flows, it should be replaced since its
subsequent reliability cannot be guaranteed.

<1> The semiconductor continues to work as

though nothing untoward has happened.
* The silicon chip does not melt, but the solder used
to keep the chip in place melts and then re-solidifies.
* Silicon and the copper frame have different coefficients of thermal expansion, and the rise in temperature causes cracks to form in the silicon. This in
turn causes the reverse voltage to deteriorate.
These symptoms may occur even when the temperature at which silicon melts is not reached. Users
are inconvenienced most when semiconductors
break down in this way. It is likely that the semiconductors will totally cease to work at any moment
so they should be replaced with new devices without
another thought.
This kind of breakdown will not happen if proper
thermal design is implemented and the semiconductors operate within the temperature rating.
You cannot check whether semiconductors will
break down in this way using a multimeter (because
the multimeters supply voltage, which ranges from
1.5V to 3V, is too low).

These are examples of the surge current ratings for

diodes. They apply to either 50Hz or 60Hz sinusoidal half-wave currents. What should be borne in
mind here is whether ratings apply to a cold start or
hot start, whether it is case of purely allowing the
current to flow or whether a reverse voltage is applied. (Surge current ratings are sometimes specified
under full-load conditions, which refers to hot start

<2> Open-circuiting occurs.

Regardless of whether the silicon is alive or
dead, the tendency is to consider that wire burnout
is responsible for open circuit problems. Opencircuit trouble is, generally speaking, the most
convenient of all breakdown scenarios although it
hardly ever happens. The reason is that, as described
above, with the power semiconductors the resistance
of the materials other than silicon is kept low so that
wires conducive to burnout are not used.

Example of surge current ratings for diodes

15

and rated reverse voltage applied.) There is a slight

difference in the tolerance depending on whether a
reverse voltage is applied, but the difference will not
be factored in here because it is taken care of by the
margin provided by the manufacturer. The tolerance
decreases by a maximum of 20% or so for hot starts
(i.e. during high-temperature operation) compared
with cold starts (i.e. under room temperature conditions). In other words, this tolerance drops to 40A
under cold start conditions for a device with a 50A
tolerance. Thoroughly examine that the diodes will
not break down due to this drop under those conditions which can be expected to lead to failures.

when considering the surge current tolerance, I2t

is a more valuable yardstick.
Lets now try to calculate the I2t and I2t of a diode
with a rated surge current of 45A (= peak value) at
one 50Hz sinusoidal half-wave cycle. The pulse
width is 10ms, and since I is the root mean square
value in this case:
I2t(45/220.01
10.125A2s
I2t(45/220.01
101.25A2s1/2
Lets plot the sinusoidal half-wave current peak level
for a pulse width of under 10ms on a graph on the
assumption that these elements are constant.
Sinusoidal peak current (A)

In the AC input unit of the ATX power supply mentioned before, the surge current is minimized by the
positive temperature coefficient (PTC) thermistor.
The resistance of PTC decreases at high temperatures, so the current suppressing effect here is inferior compared at cold start. In this connection, attention must be paid to the occurrence of instantaneous power line failures. In Japan, these power line
failures may or may not happen once a year, but
when one does happen, hot start conditions will
apply. In short, unless some leeway is provided, a
device may break down for no apparent reason when
the power fails an event which will cause a hot
start even when it has been verified that the device will not break down under cold start conditions.
This aspect should be borne in mind when selecting
the diodes for rectification circuits which are directly
coupled to AC power lines.

200A
142A
100A

Constant I2t

80A

50A

45A

Constant I2t

20A
1ms

2ms

5ms

10ms

Sinusoidal half-wave pulse width

Surge current tolerance for pulse widths of 10ms
2
2
and below based on I t and I t

I2t and I2t

In the calculation there is a difference of 50A or so

between I2t and I2t for the pulse width of 1ms.
This ties in with the leeway provided for the guaranteed value vis--vis the breakdown value. If I2t
is revised upward by an amount equivalent to this
leeway, the two curves become similar.
A constant value for I2t helps estimate the extent
of the tolerance when the pulse width is reduced
even with power semiconductors other than diodes

Surge current ratings are provided for commercial

power line frequency rectification circuits, and they
do not support pulse widths of under 10ms at 50Hz
or under 8.3ms at 60Hz. The tolerance from 1ms to
10ms (8.3ms) is specified as I2t and I2t. Anyway,
I2 is very familiar rating. In the high current region,
the properties of semiconductors are lost, and this
tolerance becomes a simple resistance. If r is used
to represent the resistance, the loss will be I2r. In
other words, I2 denotes the loss component. Therefore, I2t is designed to specify the tolerance at a
constant level of energy. I2t, on the other hand,
denotes the approach of keeping the temperature rise
constant. This is because transient thermal resistance at time t can be approximated to at (where
a is the constant inherent to the device).
I2t is also specified for fuses, and if diodes and thermistors have an I2t exceeding the I2t of the fuses,
they will be protected as a result of blowing of fuses.
(The term protection coordination is used in this
context.)
It is believed that the thermal breakdown of semiconductors does not occur when the energy reaches a
specific level but rather that it occurs when the chip
temperature reaches a specific level. In this sense,

I2t is now obtained by Spice (software

programs with integrated circuit evaluation) simulation.
10EBD60

10H

3
370V
C

Capacitor charging current

16

Discharging current

470F
220F
100F
47F

Time

Discharging current of capacitors flowing to diode

2 1/2

470F 326A s

2 1/2

220F 215A s

I2t

2 1/2

100F 138A s
2 1/2

47F 89A s

I2t for discharging current of capacitors

appropriate level.

The number of cases where the actual circuits in

which the current is sinusoidal are rather limited. As
an example of the likely current, we will calculate I2
t of the charging currents of capacitors using Spice
simulation. Although this cannot be directly compared with I2t of diodes, it does help in terms of the
diode selection.
The charging currents for the four capacitors in this
circuit namely, 47F, 100F, 220F and 470F
and the charging current and I2t are obtained.
The figure of 3 ohms is used for the assumed residual
resistance of capacitors and circuit. This resistance
has a significant effect on the result so refer to the
capacitor documentation and set the resistance to the

C Discharge Vswitch
V0 1 0 370volts
S1 1 3 2 3 SW1
.MODEL SW1 VSWITCH(RON=0.01 ROFF=1E6)
Vg 2 3 pwl(0,0 0.1u,1 5000u,1)
Ds 3 4 10EDB60
Rs 4 5 3
Ls 5 6 10uH
Cs 6 0 {KCs} IC=0V
.PARAM KCs=1E-6
.STEP PARAM KCs LIST 47E-6 100E-6 220E-6 470E-6
*
* 10EDB D model
.MODEL 10EDB60 D
+ IS=2.8596E-9

17

+ N=1.1806
+ RS=1.0000E-3
+ CJO=1.0000E-12
+ M=.3333
+ VJ=.75
+ ISR=100.00E-12
+ BV=800
+ IBV=100.00E-6
+ TT=5.0000E-9
*
.TRAN 0.01us 5000us
.PROBE
.END

The charging current (equal diode current) simulated

by this circuit file is shown in the figure on the previous page. When the horizontal axis indicates t
and the s() function is displayed in order to integrate
I2, then the vertical axis will give an exact representation of I2t.
It is clear from the simulation that I2t for the 47F
capacitor is 89A2s1/2. The 10EDB60 is a 1A 600V
diode, and since the surge current at 1 cycle is 45A,
I2t is 101A2s1/2 as per the above calculation. Verification in the actual circuit is performed on the basis
of this result, and at least at a capacitance of 100F
or more breakdown likely appears unless the diode is
larger. The surge current at 1 cycle is 75A for the 2A
20KDA60 and 100A for the 3A 30PDA60, and corresponding I2t is 281A2s1/2 and 500A2s1/2, respectively. When these values are compared with the I2
t values obtained by simulation, it is possible to make
a rough guess as to which one can be subject to actual verification. In this example, the supply voltage
is sufficiently higher than the diode forward voltage
even with different diodes, and so this difference in
the diodes does not pose too much of a problem for
simulation.
Here we have obtained I2t, and if we are to know
I2t, we can use the same method while keeping the
time axis unchanged. The method can also be applied to obtain the average loss when a current subject to pulse width modulation.

18

their switching time cannot be controlled by the

base and gate terminals like transistors. Consequently, fast recovery diodes must be selected for
applications involving high operating frequencies.
SBDs do not, in principle, have a reverse recovery
time. However, they have a high capacitance
component so that they appear to work in the same
way. In terms of time, it is quite justified to think
of these diodes as having the same kind of reverse
recovery time as the fastest kind of fast recovery
diodes.

Switching time of diodes -- reverse recovery time

Refer to the figure below. When a forward biased
diode is suddenly reversed biased, the diode will
move through a state of having virtually zero resistance to the original high-resistance status. The
period of this transient status is called the reverse
recovery time, and it is a kind of switching time.
For ordinary rectifier diodes, this time is measurable in microseconds; for fast recovery diodes
(FRD), it is 1/5 to 1/10 or less of this, and for
diodes with a very fast recovery, it is measurable
in 10 nanosecond increments.

Differentiating SBD and FRD

As has already been explained, SBD and FRD
have different chip structures. This means that the
principle behind their operation is also different.
The guard-ring incorporated as withstand-voltage
structures for SBD is PN diodes and some integrate FRD on the chips. In other words, many
SBDs also have FRD properties and, for this rea-

0
Current

Rated reverse voltage VRRM (V)

FRD

Ordinary rectifier
diode

10

40

60

Time have two terminals and, as such,

Diodes
Reverse recovery waveform of ordinary
rectifier diode and FRD

100

200

400

600

SBD
FRD

Ver. 5A/DIV. Hor. 100ns/DIV.

Samples: Both are 3A diodes.

Differentiation between SBD and FRD

Reverse bias
Forward bias

Resistance

Forward direction
(low resistance)

Zero resistance

Reverse direction
(high resistance)

Time
Reverse recovery time

Reverse recovery time of diode

19

son, no further differentiation between the two

types will be made here.
A voltage of approximately 200V serves as a dividing line between SBDs and FRDs with SBDs
having voltages lower than 200V and FRDs having voltages higher than 200V.

ing the ratio between them in such a way to ensure

that the output is constant. Lets now look at a
detail of the marks provided in the diagram below:
it shows the situation when the transistor is ON.
Originally, the current that flows to a transistor is
passed to the load via the inductor but, during the
reverse recovery period of the diode, it is passed to
the diode instead, causing short-circuiting between power supply rails. At this point in time,
the transistor is transferring to the ON status so the
voltage has not finished falling. In this way, both
the current and voltage are both high so that the
loss, which is the product of the two, is also high.

Why are fast recovery diodes needed?

<1> Failure of the circuits to work
What is the upper limit frequency at which an
ordinary rectifier diode with a reverse recovery
time of 10s will work? Simplistically speaking,
the impedance is zero during the reverse recovery
period. In other words, the diode loses its ability
to pass virtually no reverse current, and starts to
function as a mere conductor rather than as a diode. The sinusoidal wave frequency for a 10s
half wave is 50kHz. The frequency of which 10s
is one-fiftieth or 2% of 50kHz is 1kHz. Based on
this approach, about 1kHz for the upper limit
frequency is a convincing figure for ordinary rectifier diodes. By applying the same standard, the
upper limit frequency for an FRD with a reverse
recovery time of 2ns can be considered to be 500
times higher than this figure ; namely 500kHz.

Controller

The charge that

should be retained
in the capacitor
flows out.

Vcc

Vcc

Loss increases.

Input

Output

If reverse
recovery
time
is
long

Step-down DC to DC converter
Transistor current

VB
HI
Vs

Diode current

LO

Bootstrap circuit
Shown in the above figure is a half-bridge driver
with a bootstrap circuit at the high side. If this
circuit is operated at 20kHz, the capacitor will not
retain the charge if the diode has the abovementioned reverse recovery time of 10us. This is
an example of a circuit which will not work if the
reverse recovery time is too long.

When transistor is ON (detail of time axis)

Voltage

If the reverse recovery

of a diode is slow,

Current
the transistor will not
be fully turned ON,
and this results in a
high loss.

<2> Increased switching transistor loss

The effects of the reverse recovery time manifest
more as increased loss for the transistors, MOSFETs, IGBT and other switching devices than for
the diodes themselves.
Shown next are the currents that flow to the transistor and diode in a step-down DC-to-DC converter. The two currents are controlled by chang-

Transistor and diode currents

20

In this example, the effect which the reverse recovery period of the diode has on the switching
transistor loss is significant, and the loss itself can
be described as serious. This loss is reduced by
either increasing the circuit impedance or reducing the transistor ON speed. However, these actions do not work well for the circuit. As a result,
the sensible solution for reducing the loss is to use
a diode with a fast reverse recovery.

A loss of 0.066W at 100kHz may not be perceived

to be that high. This is because the diode concerned here is the FCU10A30 which has one of
the fastest reverse recovery times. With an FRD
having a slower reverse recovery time, the loss
would be 10 times higher, and with an ordinary
rectifier diode, it would be 100 times higher.
Since the thermal resistance of the T0-220 fullpack with no hestsink is approximately 80W/C,
losses of 0.66W and 6.6W are not low enough to
be ignored.
In any event, if the operating frequency exceeds
10kHz, attention must also be paid to the losses
during reverse recovery.

<3> Increased loss of the diode itself

Loss Prec arising in a diode during its reverse
recovery period can be calculated using the following equation.

<4> The faster and softer the reverse recovery, the lower the EMI in megahertz region.
We compared the levels of noise emitted during
the reverse recovery of two fast recovery diodes
with reverse recovery times of 20ns and 100ns.
We compared these levels using the same circuits
and under the same conditions. Marked by the
arrows are the peaks of 16V for the diode with the
faster reverse recovery and 28V for the diode with
the slower reverse recovery. The energy stored in
the inductance including the stray inductance at
that point in time when the resistance of the diode
starts to recover to the high resistance is the source
of this noise generation. The energy stored in the
inductance stands in proportion to the square of
the current. In the example given, therefore, the
current levels at which the resistance starts to
increase are 0.6A and 3.6A, and the diode with the
longer reverse recovery is 36 times as high in
terms of the energy ratio; this is a significant cause
of the noise. In this way, if the reverse recovery

Prec= 1/ 6tatbVRM di/dtf

f stands for the operating frequency so the loss
increases in proportion to the frequency.

Current
Voltage

Calculating the reverse recovery loss

Lets now actually calculate the loss for the
FCU10A30 (10A 300V T0-220 full pack package)
which has a reverse recovery time of 30ns. If
IFM=5A, di/dt=50A/s and VRM=300V are the
operating conditions, then ta and tb at Tj=100
C will be 22ns and 12ns, respectively, according
to the actual measurements. The correlation between the frequency and reverse recovery loss is
as follows.

trr: Long
Current

Recovery loss
(W)

Voltage

0.1
FCU10A30
2103

trr: Short

0.01

Current
Voltage

0.001

1kHz

10kHz

100kHz 1MHz

Frequency

Difference in diode noise due to difference in reverse recovery

Example of calculating
the reverse recovery loss

Ver. 2A, 10V/DIV. Hor. 20ns/DIV.

Sample: 3A diode

21

characteristics are similar, the shorter the time, the

less noise will be generated.
The reverse recovery time is specified at 25C. Is
the noise generated by two identical diodes the
same? The answer is no, it is not. Actually,
there are two questions here: first, whether the
reverse recovery is soft; and second, what would
be the reverse recovery characteristics at a high
temperature 100C, for instance experienced with actual use? These questions cannot be
answered by referring to the data sheets. As is
clear from the photos, diode switching noise is a
phenomenon in the order of 10ns or 100ns even
for fast recovery diodes. Frequencies from
10MHz to 100MHz are supported. This is intimately bound up with EMC problems so the diode

manufacturers make a special effort to reduce the

noise. Rigorous measures to counter noise must
be taken, and to ascertain which diodes have a low
noise level, make an assessment under conditions of actual operation and after the temperature has become saturated.
The next example compares the extent to which
the diode switching noise differs. This is a very
extreme example, but you will find out straight
away exactly what kind of trouble you will encounter if you use noisy diodes unwittingly. Incidentally, the reverse recovery times on the two
data sheets are identical.

0
Current at 25
Current at 100

Current at 25
Current at 100

Voltage at 25
Voltage at 25

Voltage at 100

Voltage at 100

600V FRD (1 of 2)

600V FRD (2 of 2)

Ver. 5A, 100V/DIV. Hor. 20ns/DIV

Ver. 5A, 100V/DIV. Hor. 20ns/DIV.

Sample: FSF10A60 made by NIEC

Commentary: Rectification noise

It is well-known fact that the noise of
50Hz/60Hz rectification circuits is surprisingly
high. An actual example of this noise can be
heard by visiting NTT EAST website where
noise samples including ones of rectifier have
been collected:
http://www.ntt-east.co.jp/tasc/gijutu/noise

brought the built-in bar antenna of the Sony ICFSW77 AM receiver as close as possible to a
transformer, and fixed it in a position where the
audible noise was loudest.
The following three diodes were compared.
Rectifier diode 1A/600A
200V FRED
FCF/FRF10A20
5A2/200V
200V SBD
FCH/FRH10A20
5A2/200V

We will now talk about rectification noise which

is perceived as radiation noise. The comparison
presented here is for all intents and purposes
merely a relative comparison.
Using a 12V output linear voltage regulator designed to enable its bridge rectifier diodes to be
replaced in the sockets, we set the load to 300,
so the output current was about 40mA. We

We digitized the AM receivers output, and used

the WaveSpectra V1.30 freeware (available from
http://www.ne.jp/asahi/fa/efu/soft/ws/ws.html) to

22

display the spectral characteristics. The diagrams

on the next page show the processing averaged out
over 300 times.

Under the conditions we established for our test, it

is possible to mute the noise by adding a ceramic
capacitor with a capacitance of approximately
1,000pF in parallel with the ordinary rectification
diodes with a high noise level. However, the
temperature and current flow exert an effect so the
muting will not be perfect. If noise is a problem, it
is a good idea in the sense of reducing the noise
level to use fast recovery diodes with a considerably reduced noise energy from the start even in
50/60Hz rectification circuits, for instance.

We used a transformer with a bare winding without any magnetic shielding, emitted the noise
(changes in the magnetic field) to the maximum
extent possible, and picked up the very faint
changes in the magnetic field using the AM receiver. Even from a very slight distance away, we
could not detect this noise.

Rectification noise caught from AM radio

12V
AC100V
50Hz

LM317
Voltage
regulator
Cs Kit
TW-157

300

These can be replaced in the sockets.

Rectifier Diode

ICF-SW77

Voltage regulator

Diode

Detection of diode noise

200V FRD (FCF/FRF10A20)

200V SBD (FCH/FRH10A20)

23

Types of SBDs and FRDs

FRDs, which are PN diodes, break down into two
categories of products depending on their production methods and characteristics: devices with a
long reverse recovery time in 100ns increments,
and devices with a reverse recovery time in 10ns
increments. Manufacturers have given the latter
devices some unique names, but overseas they are
commonly referred to as fast recovery epitaxial
diodes (FREDs). The former devices are used
with operating frequencies up to dozens of kilohertz while the latter devices are used at even
higher ranges.
In the region that borders on SBDs, junction barrier Schottky (JBS) diodes or merged PIN schottky diode (MPS) which have a reverse-breakdown
voltage of 200V or so and which combine the
advantages of both SBDs and FRDs are also being
made available.
The barrier metal used has a major effect on the
characteristics of SBDs. Even for the same reverse-breakdown voltage, manufacturers use different barrier metals.
In terms of the criteria for selecting devices, let us
direct our attention not simply to how low the
forward voltage is but to the reverse current as
well. As has already been pointed out, this is
because the reverse current causes a high loss in
SBDs, and this cannot be overlooked. We will
now introduce some examples where the total loss

30

40

60

100

150

VF (V)
25

IR (mA)
25

EP10LA03

0.39

EP10QY03

0.47

EP10HY03

0.56

0.5

Characteristics of three 1A 30V SBD

devices with same package
A glance at these figures reveals that the
EP10LA03 with the lowest forward voltage is the
best choice, and the EP10HY03 with a forward
voltage which is as much as 40% higher is not an
option.
Lets now estimate the total loss in a typical stepdown DC to DC converter.

Step-down DC-DC converter

Input 15V
Tj=100

Output
current
0.2A

SOD
SOD--123 1A / 30V
EP10LA03
EP10QY03
EP10HY03

Rated reverse voltage (V)

10

P/N

Lets assume that the input voltage is 15V, the

output current is 0.2A and the operating temperature is 100C. The typical forward power loss PF,
reverse current IR and reverse power loss PR for
the three SBDs at a 0.2A forward current IF, a 15V
reverse voltage VR, and Tj of 100C are as follows.
P/N
PF (W)
IR (mA)
PR (W)
@0.2A
@15V
@15V

200

Use of barrier metal which reduces forward voltage

Use of barrier metal which results in standard characteristics

Use of barrier metal which reduces reverse current

Example of SBD device lineup

EP10LA03

0.0315

9.93

0.148

EP10QY03

0.0490

0.554

0.00831

EP10HY03

0.0626

0.0881

0.00132

Also estimated is the total loss under the two output voltage conditions of 12V and 1.5V.
First, in the case of 12V output, a forward current
flows to the diodes for 21% of the time while a
reverse voltage is applied for the remainder of the
time.
Next, in the case of 1.5V output, a forward current
flows to the diodes for 90.4% of the time while a
reverse voltage is applied for the remainder of the
time.

of SBDs with a high forward voltage is low under

actual conditions of use.
Shown is an example of an SBD device lineup.
There are three series offering devices with a 30V
reverse voltage to choose from. For instance,
EP10QY03, EP10LA03 and EP10HY03 are
available as 1A 30V SOD-123 package devices.
The figures on their data sheets are as follows.

24

Input:15V Output :12V / 0.2A

Forward 0.21 Reverse 0.79

Reverse current
Increase

IF
0

Junction temperature

Reverse power loss

Rise

VR

Increase

Thermal runaway

P/N

Total loss (W)

EP10LA03

0.124

EP10QY03

0.0168

EP10HY03

0.0141

a high reverse current and that this high current

may cause thermal runaway as a real problem.
Even with the time constant of just one hour, if the
resulting reverse current increase will not stop,
this will definitely be a case of thermal runaway.
If the temperature rises, the reverse current will
increase exponentially, and when a certain temperature threshold is exceeded, thermal runaway
will occur. This is because the reverse power loss
and junction temperature stand in a positive feedback relationship.
The following equation expresses the balance
between the generation and dissipation of heat.
Thermal runaway occurs when this condition is
satisfied.

Input:15V Output :1.5V / 0.2A

Forward 0.904 Reverse 0.096

IF

dP / dTj 1 / Rth

P : Total loss
Tj : Junction temperature
Rth : Thermal resistance

VR

Total loss (W)

EP10LA03

0.0427

EP10QY03

0.0450

EP10HY03

0.0567

Power loss

P/N

In this way, if the ratio at which the forward current flows is a high one, then the total loss will be
the lowest literally for the device with a low forward voltage VF. However, the data shows that if
the ratio at which the reverse voltage is applied is
a high one, then the device with a low reverse
current IR is a better choice. In actual circuits, the
ON/OFF ratio changes dynamically. Nevertheless, it is worthwhile examining matters in this
light if the circuits work using a step-down transformer ratio in a specific range.

dP/dTj

Junction temperature
Shown on the next page are the results of estimating the temperature at which thermal runaway will
occur for the three S0D-123 package 1A 30V SBD
devices whose total loss was compared. We used
the reverse current as a function for the reverse
voltage and temperature, and calculated the thermal resistance as a parameter. We assumed here
that the reverse voltage is applied with a 100%
duty ratio a condition which is much more
challenging than the great majority of actual operating conditions.

Thermal runaway of SBD

It is probably better to regard the fact that SBD has

25

P/N

Temperature at which thermal runaway occurs

100/W

200/W

270/W

300/W

EP10LA03

91.5

75.7

68.9

66.5

EP10QY03

138.5

127.2

122.4

120.7

EP10HY03

165.8

153.3

147.9

146.0

The thermal resistance of this package when

mounted on circuit boards is about 200 to 270C/
W. The table above shows that, depending on the
operating conditions, thermal runaway is an important issue to be examined. The term thermal
resistance used here refers to the thermal resistance of the circuit board added to the thermal
resistance of the devices. The thermal resistance
of the devices can be reduced by such means as
increasing the size of the lands, distancing the
devices from other sources of heat and improving
the ventilation.
Thermal runaway occurs at different temperatures
in the three devices. A series such as the
EP10LA03 which prioritizes the forward voltage
is suited for 0r diodes which are subject to hardly
any reverse voltage. If these devices are used in
this way, it is impossible for thermal runaway to
occur. A series such as the EP10HA03 with its
reduced level of reverse current has a lower total
loss as seen in the example above. This also ties in
with the fact that it is hard for thermal runaway to
occur, as is shown in the table.
Another important factor in considering thermal
runaway is the thermal resistance of the device
itself. Devices with a poor die attach such as the
ones shown in the photo may not satisfy the
specified thermal resistance level. In this sense, it
is more reassuring to use products which have
been thoroughly controlled by the production

50% duty
107.3

process.
When the temperature rises, the circuit operation
becomes unstable: try reducing the SBD temperature (using a cooling spray, for instance), and if
the stability is restored, try the following and
evaluate the effects:

Reduce the reverse voltage applied to the

diode.
Reduce the duty ratio at which the reverse
voltage is applied.
Improve the cooling.
Replace with an SBD with which thermal
runaway is less likely to occur.

Conclusions
Two questions must be asked with not only diodes
but power semiconductors as well. 1) What levels
of current and voltage does the device allow at the
maximum? 2) What are your operating current
and voltage? No trouble will occur if these two
are well balanced.
Using diodes as a basis, we have attempted to give
you an understanding of power semiconductors.
Not much attention is generally paid to diodes, but
they play a basic role in helping users master the
use of power semiconductors.

Die attach (soldering between chips and frame onto which they are mounted)

No solder
Large voids
Problem-free die attach

Unsatisfactory die attach

26