Latest Progress in Power Modules for Appliance Inverter
Applications
E. Motto*, J. Donlon*, Shinya Shirakawa**, Toru Iwagami**, Hisashi Kawafuji**,
Mamoru Seo**, Katsumi Satou**
* Powerex Incorporated, Youngwood, Pennsylvania, USA
** Power Device Works, Mitsubishi Electric Corporation, Fukuoka, Japan
VUFS
Level Shift
Gate Drive
UV Prot.
Level Shift
Input
Condition
U
Gate Drive
UV Prot.
VVFS
+VCC
HVIC
UP
V
VCC
VP1
VN
FO
CFO
CIN
VNC
Input Signal
Conditioning
Fault
Logic
UV
Prot.
VN1
+
+VCC
L
15V
W
Gate Drive
WN
Overcurrent
Protection
The transfer molded DIP-IPM was first
introduced by Mitsubishi Electric in 1998 to
address the rapidly growing demand for cost
effective motor control in consumer appliance
applications. These devices soon became widely
accepted due to their performance, reliability and
cost advantages in small motor drives. In the
years that followed continuous improvements in
package thermal performance, power chip design,
and HVIC (High Voltage Integrated Circuit)
technology has enabled the development of a
complete line of modules for motors rated from
about 100W to more than 10KW at line voltages of
100VAC to 480VAC. This paper will describe
some of the key technologies utilized in the latest
generation of these devices.
Level Shift
+VCC
Gate Drive
UV Prot.
Motor
HVIC
VP1
WP
HVIC
VWFS
VWFB
+VCC
Input
Condition
Typi
Input
Condition
VVFB
I. INTRODUCTION
To meet the demanding cost and size
requirements of consumer appliance inverters,
Mitsubishi developed a unique completely transfermolded intelligent power device. The transfer
molded DIP-IPM is less expensive to produce than
conventional hybrid modules because it does not
+
+
VP1
NC
UN
II. T HE DIP-IPM CONCEPT
P
VUFB
5V Logic Interface to MCU
Abstract - This paper presents a new version of
the Dual In-line Package Intelligent Power
Module
(DIP-IPM
Ver.4)
developed
by
Mitsubishi Electric for home appliance motor
control. The DIP-IPM Ver.4, features a
completely lead free process with both chip
bonding solders and lead plating compliant
with international initiatives for the reduction of
hazardous materials. Package miniaturization
has been achieved by utilizing a new insulating
resin sheet with high thermal conductivity,
direct wire bonding technology, and an
optimized lead frame design.
RSHUNT
N
RSF
CSF
Fig.1 DIP-IPM Functional Diagram
require an IMS or ceramic substrate and plastic
shell housing. The transfer molding process is
also well suited for high volume, automated mass
production, thus substantially reducing cost. The
DIP-IPM provides the low cost of a discrete
component
design while maintaining the
advantages of an intelligent power module.
Compared to a discrete approach these devices
offer high reliability, small size, and reduced
manufacturing costs by integrating optimally
matched power devices and HVIC drivers in a
single module.
Fig. 1 presents a basic block diagram of
the DIP-IPM integrated features, which include the
power devices and custom control ICs for gate
1-4244-0365-0/06/$20.00 (c) 2006 IEEE
drive and protection. The key to the DIP-IPM is the
integration of HVICs to provide level shifting and
gate drive for the high side IGBTs. This results in
significant cost savings by enabling direct
connection of all six IGBT control signals to the
controller. The HVIC also provides undervoltage
lockout
protection
to
allow
simplified
implementation of the required floating power
supplies using bootstrap techniques. With just a
few external components the entire three-phase
power stage can operate from a single 15V control
power supply. The DIP-IPM also utilizes a custom
LVIC (Low Voltage Integrated Circuit) to provide
gate
drive,
overcurrent
protection
and
undervoltage lockout for the low side IGBTs.
Incorporating the level shifting into the
DIP-IPM
reduces
high
voltage
spacing
requirements on the control PCB allowing a
significant savings in circuit board space. The
factory verified coordination of ICs and power chips
assures that it is highly reliable. All of these
features are combined in a compact low cost
transfer
molded
package
that
allows
miniaturization of inverter designs.
Al wire
nd
2
Mold resin
Au wire
IC
Cu frame
st
Al heat spreader
1
Mold
A) Original DIP-IPM
Cu Frame
Al Wire
IGBT
FWDi
Au Wire
B) Mini DIP-IPM
Al wire
IC
Mold resin
Cu frame
FWDi ,IGBT
IC
Cu heat spreader
III. DIP-IPM PACKAGE DESIGN REVIEW
In order to cover a large power range cost
effectively four different transfer molded package
structures have been developed. The cross
sections of these package structures are shown in
fig.2. All DIP-IPMs are fabricated using a transfer
molding process like a very large integrated circuit.
First, bare power chips and the custom HVIC and
LVIC die are assembled on a lead frame.
Ultrasonic bonding of large diameter aluminum
wires makes electrical connections between the
power chips and lead frame. Small diameter gold
wires are bonded to make the signal level
connections between the IC die and lead frame.
This part of the process is basically the same for all
devices. Next, they are encapsulated. This is
where the packages differ.
The Original DIP-IPM shown in fig.2A
was the first transfer molded design and it is still
used for higher current 600V and 1200V devices.
This package is fabricated using a two-step
injection molding process. In the first step, a thin
layer of thermally conductive epoxy is formed
between the lead frame and an aluminum block.
The thin layer of epoxy and the aluminum heat
spreader allow good heat transfer and provide
electrical isolation between the power chips and
heat sink. The Original DIP’s integrated aluminum
block provides the thermal characteristics needed
FWDi, IGBT
Au wire
Mold resin
C) DIP-IPM Generation 3
Cu Frame
Al Wire FWDi
Mold
IGBT
Au Wire
IC
Insulated thermal radiating sheet
(Cu foil + insulated resin)
D) New Gen. 4 Super Mini DIP-IPM
Fig.2 DIP IPM Package Cross Section
for the higher power devices. A second injectionmolding step then encapsulates the entire lead
frame assembly to achieve the final form. This
structure has been used effectively for modules
with nominal ratings up to 30A at elevated case
temperature.
A cross-section drawing of the original Mini
DIP-IPM is shown in Fig. 2B. In this device the
lead frame is formed to produce a thin, flat layer of
thermally conductive epoxy between the power
chips and heat sink mounting surface of the
device. This thin layer of epoxy and bent lead
frame allow good heat transfer and provide
electrical isolation. A single transfer molding step
encapsulates the entire lead frame assembly to
achieve the final form. The single step molding
Fig.3 Generation 4 Super Mini-DIP
process has been utilized to fabricate modules with
IGBT ratings of up to 15A at elevated case
temperatures.
Fig. 2C shows the cross section of the
generation 3 DIP IPM.
This device has a
mechanical form that is similar to the original DIPIPM. The main difference is that the heat spreader
is made of copper rather than aluminum and the
insulation layer is at the mounting surface rather
than between the lead frame and the heat
spreader.
The result is superior thermal
performance compared to the original DIP-IPM
while using a simplified single step molding
process. This structure has been used to fabricate
devices with ratings of up to 50A at elevated case
temperatures.
IV. T HE GENERATION 4 DIP-IPM PACKAGE
The new generation 4 Super Mini-DIP
package cross section is shown in fig. 2D. In all
previous DIP-IPMs the insulating layer between
lead frame and heat-sink was composed of the
injection molded epoxy resin. One way to reduce
the thermal resistance of this interface is to add
ceramic powder filler with high thermal conductivity
to the epoxy resin. However, the effectiveness of
this is approach limited because it is difficult to
maintain the required fluidity and insulation
strength of the epoxy resin when significant
amounts of ceramic are added.
To get around this problem a new low
thermal impedance structure using an insulating
resin sheet has been adopted as shown in fig. 2D.
In this novel structure a partially cured resin sheet
is adhered to the rear surface of lead frame after
chip bonding. The lead frame with the resin sheet
attached is then transfer-molded using epoxy resin.
The transfer molding process causes the resin
sheet to cure simultaneously with the epoxy resin.
The result is a stable high reliability joint with low
thermal impedance between the resin sheet, epoxy
and lead frame. The thin insulating resin sheet
stays in a fixed form during the process so it does
not need to have the fluidity of the epoxy resin over
mold and thus it is possible to increase the amount
of ceramic fill to improve the thermal conductivity.
In addition, it is possible to achieve a thinner
insulating layer because it is not constrained by the
limitations of the molding process. The extremely
thin layer of high thermal conductivity resin yields a
substantial reduction in thermal impedance
compared to previous DIP-IPM designs.
Another feature of the new generation 4
package is that it is completely lead free and meets
the requirements of the RoHS (Restriction of the
use of Hazardous Substances in Electrical and
Electronic Equipment) directive which takes effect
on 1 July 2006 and requires that certain equipment
must not contain 6 chemical substances including
Lead (Pb). In the generation 4 super Mini DIP-IPM
lead free lead plating (external) and chip bonding
solder (internal) have been implemented. After
careful consideration of the thermal and
mechanical properties a Sn-Cu derivative solder
was selected for terminal plating. For stable power
chip bonding solder wettability and reliability are
the key points. For this critical application a SnAg-Cu based solder was selected. Pb-free solders
also have a tendency to oxidize rapidly compared
with Sn-Pb solder. To control this characteristic
the atmosphere and temperature used in the chip
bonding process is controlled appropriately.
V. PACKAGE MINIATURIZATION
The new package structure is remarkably
effective for miniaturization of the generation 4
Super Mini-DIP. Fig. 3 is a photograph of the new
generation 4 super Mini DIP-IPM in its final form.
The compact 38mm x 24mm x 3.5mm package is
available with nominal current ratings at elevated
case temperature of 3A to 30A. The keys to
reducing the size of the device are its improved
thermal performance discussed above and a new
direct wire bonding process.
The
significantly
improved
thermal
performance of the generation 4 package allows
smaller power chips to be used for a given rating.
In addition to saving space this also helps to
reduce the cost of the device.
In previous DIP-IPMs the power chips
were connected to the lead frame using aluminium
wires and the lead frame was connected to the ICs
using gold wire. In these older designs the lead
frame served as an intermediate connection point.
In order to do this it was necessary to provide
islands that connect to the lead frame via “dummypins”. These extra pins and islands waste space
inside the package. The new Ver.4 DIP-IPM uses
direct wire bonding technology. In this process the
power chips are directly connected to the ICs using
gold wires. As a result, the dummy-pins and
associated “islands” are eliminated allowing a
significant reduction in package size. In general,
the ratio of power module mounting area to printed
circuit PCB area is relatively large in home
appliance applications. Therefore, miniaturizing the
DIP-IPM also helps to reduce the size of the PCB.
M
VI. POWER CHIP DESIGN
The input voltage for most consumer
appliance and low-end industrial applications is
between 100VAC and 240VAC. To cover this
range, IGBTs and free-wheel diodes with a 600V
breakdown rating were selected for the generation
4 DIP-IPM. The IGBT chips are fabricated using
the most cost efficient process after considering
the performance requirements of the application.
th
The 5A, 8A, 10A, 15A devices use a 5 generation
sub 1µm planar chip design while the 20A and 30A
devices use an advanced trench gate CSTBT chip.
All free-wheel diodes used in the DIP-IPMs
are super fast/soft recovery shallow diffused types.
These diodes have been carefully optimized to
have soft recovery characteristics over a wide
range of currents and temperatures in order to
minimize EMI/RFI noise.
The 3A super Mini-DIP device utilizes a
sate-of-the-art RCIGBT (Reverse Conducting
IGBT) chip.
The RCIGBT combines a fast
recovery free wheeling diode and IGBT in a single
silicon chip as shown in fig. 4. The unit cell
structure of the RCIGBT is shown in fig. 5. By
combining the IGBT and free wheeling diode into a
single silicon chip the number of power chips in the
DIP-IPM is reduced from 12 to 6. This helps to
simplify assembly and reduce cost.
VII. FEATURES OF THE HVIC AND LVIC
All devices in the DIP-IPM family contain
HVIC and LVIC chips to provide gate drive and
protection for the power devices. These features
are described in this section.
A. High Voltage Level Shift
RC-IGBT
(One Chip)
IGBT+ FWDi
(Two Chips)
Fig.4 RCIGBT Chip
n+ emitter
p base
Carrier stored
n layer
n-body
N Cathode
P Collector
Collector electrode
Fig.5 RCIGBT Chip Structure
The main feature of the DIP-IPMs is the
high voltage level shifting provided by the
integrated HVIC. The built-in level shift eliminates
the need for relatively expensive opto-couplers or
pulse transformers and allows direct connection of
all six control inputs to the CPU/DSP.
B. Undervoltage Lockout
The DIP-IPM is protected from failure of
the 15V control power supply by a built in
undervoltage lock out circuit. If the voltage of the
control supply falls below the UV level specified on
the data sheet, the low side IGBTs are turned off
and a fault signal is asserted. In addition, the pside HVIC gate drive circuits have independent
undervoltage lock out circuits that turn off the IGBT
to protect against failure if the voltage of the
floating power supply becomes too low. If the high
side undervoltage lockout protection is activated
OT trip temp.
Reset temp
LVIC temp.
OT hysteresis
be asserted. The fault condition will automatically
clear once the device has cooled below the over
temperature reset level. Approximately 10C of
hysteresis is included to prevent oscillations of the
over temperature protection.
E. Interface Circuit
Low-side
intput
Low-side
gate output
Fo output
Item
min
typ
max
OT trip temp.*
100˚C
120˚C
140˚C
OT hysteresis
-
10˚C
-
*LVIC temperature
Fig.6 Over Temperature Protection
the respective IGBT will be turned off but a fault
signal is not supplied.
C. Short-Circuit Protection
The DIP-IPMs have an integrated shortcircuit protection function. The LVIC monitors the
voltage across an external shunt resistor (RSHUNT)
to detect excessive current in the DC link. An RC
filter (RSF, CSF) with a time constant of 1.5 to 2µs is
normally inserted as shown in Fig. 1 to prevent
erroneous fault detection due to di/dt induced
noise on the shunt resistor and free-wheel diode
recovery currents. When the voltage at the CIN pin
exceeds the VSC reference level specified on the
device data sheet the lower arm IGBTs are turned
off and a fault signal is asserted at the FO output.
When an overcurrent condition is detected the
IGBTs remain off until the fault time (tFO) has
expired and the input signal has cycled to its off
state. The duration of tFO is set by an external
timing capacitor CFO.
The DIP-IPM has seven microprocessor
compatible input and output signals. The built in
HVIC level shifters allow all signals to be
referenced to the common ground of the 15V
control power supply. The signals are compatible
with 3.3V to 15V TTL/CMOS logic in order to
permit direct connection to a PWM controller. Fig.
7 shows the equivalent internal circuit of the DIPIPMs control signals and a simplified schematic of
a typical external interface circuit.
The
components shown in dashed blue lines are
optional noise filtering that may be required
depending on the circuit layout and its proximity to
noise sources. On and off operations for all six of
the DIP-IPM’s IGBTs are controlled by the active
high control inputs UP, VP, W P, UN, VN, W N. These
inputs are pulled low internally by a 3.3kΩ resistor.
The controller commands the respective IGBT to
turn on by pulling the input high. Approximately
1.8V of hysteresis is provided on all control inputs
to help prevent oscillations and enhance noise
immunity.
The fault signal output (FO) is in an open
collector configuration. Normally, the fault signal
line is pulled high to the 5V logic supply with a
10kΩ resistor as shown in Fig.7. When an overcurrent, over temperature condition or improper
control power supply voltage is detected the DIPIPM turns on the internal open collector device and
pulls the fault line low.
3.3V +
15V
10KΩ
D. Over Temperature Protection
R
The generation 4 DIP IPM is available with
optional over temperature protection.
The
operation of this circuit is shown in fig.6. A
temperature sensor is fabricated on the LVIC chip.
If the temperature of the LVIC, which is essentially
the same as the device case temperature, exceeds
the specified over temperature trip point the three
lower IGBTs will be turned off and a fault signal will
Controller
C
R
+
VD
DIP-IPM
UP, VP,
W P, UN,
VN, W N
RPD
(3.3k Typ.)
FO
Gate
Drive
Vth(off)=0.8V Min.
Vth(on)=2.6V Max.
Fault Logic
C
GND
Fig.7 Gen. 4 DIP IPM Interface Circuit
VIII. DIP-IPM SYSTEM ADVANTAGES
Inverters for small AC motors used in
appliance applications are required to meet
stringent efficiency, reliability, size and cost
constraints. Historically, many of these small
inverters have utilized discrete IGBTs (Insulated
Gate Bipolar Transistors) and free-wheel diodes in
TO-220, TO-247 or similar packages along with
separately packaged HVICs (High Voltage
Integrated Circuits). There are, however, several
problems with this approach. One drawback is the
high manufacturing cost associated with mounting
and isolating multiple high voltage discrete
components. Each of the discrete devices must be
individually mounted using special hardware and
insulating materials which typically results in a
complex assembly and significant manufacturing
time. In addition, relatively large and complex
printed circuit designs are required to meet all of
the spacing and layout requirements for the HVIC
and discrete power device combination. Another
equally perplexing problem is maintaining
consistent performance and reliability when the
characteristics of the HVIC drivers and IGBTs are
not properly matched.
A much better approach, realized in the
DIP-IPM described in this paper, is to assemble
bare power chips and HVICs using a transfer
molded lead frame design to maintain low cost and
consistent, reliable performance. Clearly, there are
significant manufacturing advantages to the DIPIPM approach. With the fully isolated DIP-IPM
mounting is accomplished with only two screws
and no additional isolation material is required.
The reduced manufacturing time and simplified
assembly provided by the DIP-IPM will allow
improvements in both cost and reliability of the
finished system. Another advantage of the DIPIPM is that the integrated HVIC and LVIC gate
drive and protection functions are factory tested
with the IGBTs as a subsystem. This eliminates
uncertainty about the critical coordination of the
electrical characteristics of these components.
The end result is more consistent system
performance and reliability.
IX. PRODUCT LINE-UP
The gen. 4 super Mini-DIP line-up is
shown in Table 1. All modules have a blocking
voltage rating of 600V which is appropriate for
100VAC to 240VAC applications. Devices with
nominal current ratings of 3A to 30A are all
available in the same compact package outline.
The table also shows the usable sinusoidal RMS
motor current per phase for some typical
application conditions.
These values are
calculated using the loss simulation software
available from the Powerex website. The table
also list some of the available options which
include several different lead forms, over
temperature protection, and open low side emitter
configurations. The open emitter configuration
allows the use of separate shunt resistors in each
of the three legs.
In addition to the devices listed here
Powerex offers devices in larger transfer molded
packages with nominal current ratings up to 50A
and also blocking voltages of 1200V.
Table 1: Gen. 4 Super Mini-DIP Line-Up
Nominal / Peak
Current Rating
IGBT and Free
Wheeling Diode
Continuous Sinusoidal
Inverter
Output Current (ARMS)*
Tsink≤ 80C, Tj≤ 125C, IPEAK≤ 1.7*IC
PF=0.8, VCC=300V
Part Number
IC/ICP
fsw=5KHz
fsw=15KHz
3A / 6A
3.6
3.6
PS21961
5A / 10A
6.0
6.0
PS21962
8A / 16A
9.6
7.4
PS21963-E
10A / 20A
11.2
8.1
PS21963
15A / 30A
14.0
9.6
PS21964
20A / 40A
16.2
11.0
PS21965
30A / 60A
TBD
TBD
PS21967
Options
(-Part number suffix)
-A Long (16mm) pins
-S Open emitters
-C ZigZag leadform
-W Double ZigZag leadform
-T Over Temp. Protection
(New Option Available
Summer 2006)
* Tj≤ 125C and IPEAK≤ 1.7*iC are selected according to recommended design margins. The actual device limit is: Tj≤ 150C, IPEAK≤ ICP
X. CONCLUSION
The new gen. 4 super Mini DIP-IPM has
been presented. This device features a compact
transfer molded package with significantly
improved thermal performance compared to older
devices.
Cost is minimized by selecting an
optimized combination of power chips and ICs for
each rating.
REFERENCES
[1]
G. Majumdar, et al. "A New Generation
High Performance Intelligent Module" PCIM
Europe May 1992
[2]
E. Motto, et al. "A New Generation of
Intelligent Power Devices for Motor Drive
Applications" IEEE IAS Conference October 1993
[3]
E. Motto "Protecting High Current IGBT
Modules From Over Current and Short Circuits"
HFPC Conference, May 1995
[4]
John
Donlon,
et
al.
"A
New
Converter/Inverter
System
for
Windpower
Generation Utilizing a New 600 Amp, 1200 Volt
Intelligent IGBT Power Module" IEEE IAS
Conference October 1994
[5]
E. Motto, et al. “A New Intelligent Power
Module With Microprocessor Compatible Analog
Current Feedback, Control Input, and Status
Output Signals”, 1996 IEEE IAS Conference
Proceedings
[6]
Eric R. Motto “A New Ultracompact ASIPM
with integrated HVASIC” 1997 Powersystems
World conference proceedings
[7]
G. Majumdar et al. “Novel Intelligent Power
Modules for Low-Power Inverters” 1998 IEEE
PESC Proceedings
[8]
S. Noda et al. “A Novel Super Compact
Intelligent Power Module” 1997 PCIM Europe
conference proceedings
[9]
Eric R. Motto “Application Specific
Intelligent Power Modules – A Novel Approach to
System Integration in Low Power Drives”
Proceedings of the 1998 Powersystems World
Conference
[10]
H. Iwamoto, E. Motto, J. Achhammer, M.
Iwasaki, M. Seo, T. Iwagami, "New Intelligent
Power Module for Appliance Motor Control", 2000
PCIM Europe,
[11]
M. Iwasaki, T. Iwagami, M. Fukunaga, X.
Kong, H. kawafuji, G. Majumdar, "A New Version
Intelligent Power Module for High Performance
Motor Control", 2004 PCIM China,
[12]
T. Sasaki, H. Takao, T. Shikano, S. Fujita,
D. Nakajima, T. Shinohara, "Development of High
Current Transfer-mold type Power Module with
High Heat-cycle Durability", 2004 ISPSD
Kitakyushu,
[13]
J. Donlon, et al. “Appliance Motor Control
Using New Intelligent Power Modules in Injection
Molded Packages” 2001, IATC Conference