MOSFET: Planar vs. New Trench
technology comparison in Linear Mode Applications
By Giuseppe Longo and Filippo
Scrimizzi, ST Microelectronics
INTRODUCTION
In most applications a power MOSFET works in switching
mode. In order to reduce its power dissipation the device
switches on to have the minimum voltage drop Vds.
Instead, in specific applications power MOSFET devices are
used in operating mode called “Linear zone”. The
linear zone is that output characteristic area, where the
Ids is almost controlled by the Vgs
value. In linear mode operation the device is similar to
a variable resistor, in which the value is selected by
Vgs and by an external voltage power supply and load
condition. In this report the technical results performed
will be shown at bench in automotive applications which
work in linear mode, comparing new trench technology devices
with planar ones.
LINEAR MODE OPERATION DESCRIPTION
A power MOSFET device is designed to achieve the best performance
in terms of efficiency and reliability in customer applications.
The main parameters can be summarized as follows: on state
resistance Rdson,
threshold voltage Vth, breakdown voltage
BVdss, gate charge Qg, reverse
recovery parameters and low voltage drop during the operation
of body drain diode. Furthermore, the device has to be reliable
in terms of:
- Temperature management
- Life time
When the power MOSFET works in linear mode, the main features
are limited only to managing temperature and robustness
in what is known as thermal instability or thermal runway.
The operation in linear mode is characterized in choosing
Ids current and Vds voltage
drop by a Vgs voltage selected by
feedback control driver (almost always the control signal
is set to join a specific current value). In this condition,
the device power dissipation becomes high and the mean junction
temperature increases its value. During this phase the device
working stability depends on the intrinsic structure features
and exchange heating system.
The temperature is an important factor, since if the max
rating is exceeded, the device fails. When the device is
at thermal steady state, the current is limited by the temperature
because the Rdson increases with it
and consequently the current is forced to an auto limitation.
However, in linear mode other factors can contribute to
bringing the device to exceed the maximum temperature. These
factors are:
- Non uniform power dissipation
- Zero tempco point
- Technology features
The power dissipation achieves high values when the device
works in linear mode, because the device works with high
voltage drop and high current, so that the intrinsic temperature
increases rapidly. This means that total junction-ambient
Rthj-a
thermal resistance (device+heatsink) has to be chosen in
a proper way to dissipate the total heat amount in external
ambient. Moreover, in the source surface, the areas next
to the source wire bonding become hot spots due to the high
power management, therefore the local temperature can exceed
the max rating.
Zero tempco defines a specific feature of the power MOSFET
device at a certain temperature. If we trace the curve Ids
vs Vgs at various mean junction temperatures,
we obtain the following graph:
The curve is traced at fixed Vds
values, so that for each Vds value
we have a different curve Ids vs Vgs
. This curve shows that if the “Vgs>Vgs
zero tempco” the current decreases its value if the
junction temperature increases. Instead if the Vgs>Vgs
zero tempco the current increases if the temperature increases.
“Zero tempco” is the temperature matching point
of curves Vgs-Ids
at a fixed Vds value. If the device
is working exactly at “zero tempco” point, the
current Ids remains constant at a
changing temperature. In our case, the device is working
in linear mode at Vds fixed values,
the matching point Ids-Vgs
can be lower zero tempco and so it works in unstable condition
because each temperature increase causes an increase of
current. This phenomenon could increase when during the
normal working conditions in the application; the Vgs is
reduced by the control driver, for a fast reduction of the
load. The increasing current for an increasing temperature
can determine the thermal runway. The thermal runway is
an unstable condition that occurs when the temperature increases
without control until the device fails. The phenomenon is
emphasized since the power dissipation is high and non uniform
power distribution makes hot spots where the temperatures
can exceed maximum rating. Even if the device has a well
designed heat-sink, the hot spots are difficult to control
because the heat-sink helps to reduce its total mean junction
temperature not the hot spots that are localized in small
areas.
Technology has referred to the procedure and skills to create
a device with specific parameters. Designers receive inputs
by customer to define a device with the right matching parameters
but sometimes to improve the ones they are forced to get
others worse. In general, new devices are characterized
by two main principal features:
- Low Rdson
- Low Gate Charge, Qg
The two features are joined by specific work processes that
allow the achievement of high currents in areas that are
always smaller, in order to have high max current, low voltage
drop Vds, fast switching. But these
features on the other hand, make the device not properly
work in linear mode because the high current capability
in a small area creates a high current density. So during
linear mode operation, hot spots become hotter and consequently
the device can probably fall in thermal runway.
Linear mode working is preferred by some automotive customers,
and some power devices are designed to operate in this mode.
The typical application is VBC (variable blower control)
that is the air ventilation system for automotive air conditioning.
The advantages can be summarized as follows:
- The electrical load variation is fluent and linear when
device switches from a working point to another one
- The driver circuit is very simple to realize
- Electromagnetic interference are null since the device
doesn’t switch
Instead, the drawbacks are:
- High efficient heat-sink need
- High power dissipation device at medium load not recoverable
- Potential thermal runway trigger.
TEST BENCH
In this chapter we'll illustrate the test results performed
on VBC, where the main switch, that is a power MOSFET, works
in linear mode.
The picture above shows a typical electrical scheme. The
power MOSFET working in linear mode, can be assumed like
a variable resistor, limiting the motor current and consequently
the power dissipation.
On the application system we tested two kinds of devices
on bench:
- Planar technology characterized by higher
Rdson per die size
- New trench technology with lower Rdson
per area.
The application system has the main power device fixed by
a screw to a big heat-sink with Rth = 7°C/W, and the
heat-sink is inserted in the plenum and ventilated by air
flow of motor, which means the equivalent thermal resistance
can have lower value that depends on the air flow velocity.
The air flow velocity is high at high duty cycle and low
at low duty cycle. A thermo resistor control assures that,
if the system temperature exceeds a certain value, the control
feedback increases the duty cycle at max value to increase
the air flow ventilation and so reduces the system temperature.
In the application system the following steps were performed:
- device power monitoring and plastic body temperature measurement
at various duty cycles
- device power monitoring and plastic body temperature measurement
at various duty cycles with heat-sink not ventilated by
air flow, with disconnected control thermo-resistor
- Dynamic switching variation of the duty cycle from 0%
to 100%
- Load short circuit test for 30ms
- Load dump by pulse generator
The selected device for this testing was the STP141NF55.
During the test at various duty cycles the max plastic body
temperature was around 129°C with a power dissipation
of 75W in continuous mode. The working operation at various
duty cycles was stable, and when the test was finished,
the main device parameters were tested, and all measured
values were found aligned to the datasheet ones.
The above picture shows the system efficiency vs. duty
cycle. The variation is almost linear and the max efficiency
is achieved when the duty cycle d is at maximum value.
The picture above shows the estimated Tjmean vs duty cycle
d. The Tjmean was calculated at thermal steady state at
each d, by measuring plastic body and heat-sink - copper
frame interface temperature and by the device thermal impedance
Rthj-c . The graph gives us a clear idea that the max temperature
is joined at around 50%-60% d, that are in general the common
working values. At lower and higher d than the above specified
value, the temperatures have low values, for the following
reasons:
- at low d the power dissipation
is low so the mean device temperature is low
- at high d the power dissipation is low since the device
works at low Rdson and at the same time the air flow joins
the max velocity
During the test, we also monitored the power dissipation
at start up for each duty cycle d, (see the above picture).
The power dissipation joins a high value due to the matching
of high Ids - Vds. During this start up, we measured the
device temperatures and calculated the estimated junction
mean temperatures vs. duty cycle d (see the picture below):
The graph shows that the max temperature achieved is around
85°C at 80% d when the device current is almost at max
value.
To have a clear estimation of the max power stress allowed
on the device involved, we performed a test with the heat-sink
out of the plenum and disabling the thermo resistor. The
test was performed starting from low duty cycles continuously
monitoring the plastic body temperatures. The device failed
at 30% duty cycle after almost 25 minutes continuous working,
joining in the plastic body case a temperature of 330°C.
The result gives us an indication that the device has a
high thermal stability before failing. In fact, the thermal
runways happened after the plastic body exceeded 330°C.
The high temperature managing, in linear mode, is sustainable
for devices having low current density per area, to avoid
the creation of hot spots that exceed suddenly the max failure
temperature of the silicon die.
The picture shows a device thermo picture with not ventilated
heat-sink/board. In this case the working condition, even
if the temperature exceeds the absolute max rating, is still
stable. This working condition at this high temperature
guarantees the thermal stability of the device.
The picture above shows the source surface area of a failed
STP141NF55 after chemical opening. It can be noticed that
the burnt area includes the area around the source wire
bonding pad.
The area that is not uniform is the one where the hot spot
is starting, next to the source wire. Since the power density
is higher than the boundary limits (the current density
increases its value when is next to the source wires), but
was a consequence of difficult heat transferring by heat
sink. The following tests were also performed on the board:
- Load short circuit test
- Load dump by pulse generator
The load short circuit test was performed by a temporized
contactor for 30ms
Load dump at 150V 350ms ; the board was working at d =
50%
The load dump was also performed by a temporized contactor
connected to a power supply. The test was set at 50V-50A
for 100ms. This load dump (very stressful) destroyed the
device (short all pins). In general, this specific load
dump never happens in systems during its working life, but
it gives us an estimation of absolute maximum rating (see
typical waveform below):
During this load dump the Ids and Vds quickly join high
values with high power dissipation peak.
In the application STP141NF55 was replaced by a new trench
device and the test results were vice versa negative. The
device failed almost instantaneously when the duty cycle
was switched on. This confirms that high current density
technology makes a device weak in linear mode operation.
In fact, linear mode working inside the device structure
creates hot spots that probably exceed the 330°C, pushing
the device itself in thermal runways.
The test was repeated many times and after that, one sample
was chemically opened and analyzed by microscope. The picture
shows the details of the failures:
The black area next to the right source wire is the area
where the hot spot begins exceeding the max silicon temperature.
The failure point probably is due to the bonding source
wires, which are the areas where there is max current density.
The boundary of the black area is well defined, meaning
that the hot spot starts very rapidly fusing the silicon
without spreading to other areas.
The hot spot are smaller areas that are characterized by
high temperature because the current density is high. So
statistically the thermal runway has more probability to
start in this area than other ones.
Conclusions
The SP ST MOSFET technology is suitable for linear mode
applications where not only the power dissipation capability
is very important, but also the thermal instability robustness.
Vice versa the new Trench power MOSFETs have an even higher
equivalent cell density per area and then a lower Rdson.
These Power MOSFETs do not work properly in this kind of
application because of their intrinsic weakness linked to
the thermal runaway.
