AND8199

http://onsemi.com

2

125°

Figure 3. Fairchild HUF75631SK8

50

40

30

20

10

0

234

5

6

Pulse Duration = 80 ms

Duty Cycle = 0.5% Maximum

1

2

3

4

1

234

9

7

6

5

8

10

6

57

8

9

10

Figure 4. Typical Transconductance Curve

25°

125°

−40°

Point of Inflection

−40°

All three devices shown have one thing in common:

a point of inflection at which the temperature coefficient is

zero. At greater gate−to−source voltages, the coefficient is

positive, and, at lower gate−to−source voltages it is negative.

Figure 4 illustrates the change from negative to positive.

At a gate−to−source voltage greater than that of the

exists. At this gate voltage, the drain conducts more than

9.0 A of current. However, at 125

°C the drain current

reduces to less than 7.0 A. The arrow at the left of Figure 4,

which shows the current decreasing due to an increase in

temperature, indicates this drop.

At a gate−to−source voltage below the inflection point

the drain current is close to zero. However, at 125

°C, the

drain current is more than 1.0 A. A second arrow at the left

of Figure 4 indicates this relationship, and the current rises

due to an increase in temperature.

The implication is that when you are controlling the FET

with a gate−to−source voltage below the inflection point,

thermal runaway can occur. When one cell or a small group

of cells becomes hotter than the surrounding cells, they tend

to conduct more current. This situation, in turn, creates more

heat, which allows more current to flow. These cells can pull

a large amount of current and, if not limited in time, can

cause the device to fail.

This situation is similar to the well−known phenomenon

of secondary breakdown that occurs in bipolar transistors

except that a bipolar junction transistor is a single device,

and you can take steps to avoid its destruction. A power

MOSFET contains thousands of parallel devices that are

internal to the die, and you cannot individually protect them.

If hot spots occur, the SOA characteristics of the heavily

conducting cells differ greatly from those of the marginally

conducting cells.

The thermal−runaway situation occurs when you use large

devices at low current−limit settings. Even though it would

appear to be desirable to use a very large MOSFET for an

application such as a hot swap and limit it to a low current,

it may be an inappropriate approach. Use of a

very−low−on−resistance device offers low losses for

steady−state operation but may cause the device to fail

during a short circuit or an overload.

One way to overcome this problem is to directly sense the

die temperature of the MOSFET by integrating the

MOSFET with the controller using a monolithic approach.

ON Semiconductor takes this approach with its new line of

hot−swap ICs. In this case, the temperature can be sensed

directly on the FET die. The location of the sense element on

the die is critical for ensured protection of the device. If a hot

spot occurs too far from the sense location, the device may

be unable to protect itself.

Discrete hot−swap controllers employ a number of

protection schemes. Thermal instability is an issue only if

the controller can go into a constant−current mode of

operation. Some protection circuits simply shut off the

MOSFET switch if a number of conditions indicate a

dangerous area of operation. Controllers that use a

constant−current method of protection can use timers or

other schemes along with the current−limit circuit to reduce

the risk of failure.

Because system efficiency is an important parameter, it is

tempting to use the largest MOSFET possible to reduce

losses. It is important to keep in mind, however, that this

approach may require you to make a trade−off with the

system reliability if you are not mindful of the possible

thermal instability. You can reliably use a large power device

at a low current limit level if you handle it properly.