English
Language : 

AB-039 Datasheet, PDF (2/6 Pages) Burr-Brown (TI) – POWER AMPLIFIER STRESS AND POWER HANDLING LIMITATIONS
As VCE is further increased, beyond the thermally limited
region, the safe output current decreases more rapidly. This
so-called second breakdown region is a characteristic of
bipolar output transistors. It is caused by the tendency of
bipolar transistors to produce “hot spots”—points on the
transistor where current flow concentrates at high VCE.
Exceeding the safe output current in the second breakdown
region can produce a localized thermal runaway, destroying
the transistor.
The final limit is the breakdown voltage of the transistor.
This maximum power supply voltage cannot be exceeded.
Often, an SOA curve provides information showing how the
safe output current varies with case temperature. This ac-
counts for the affect of case temperature on junction tem-
perature. Additional lines may show the maximum safe
current for pulses of various durations according to the
thermal time constants of a device.
The SOA curve should be interpreted as an absolute maxi-
mum rating. Operation at any point on the thermal limit
portion of the curve produces the maximum allowable junc-
tion temperature—a condition not advised for long-term
operation. Although operation on the second-breakdown
portion of the curve produces lower temperature, this line is
still an absolute maximum. Operation below this limit will
provide better reliability (i.e.—better MTTF).
HEAT SINKING
In addition to assuring that an application does not exceed
the safe operating area of the power amplifier, you must also
assure that the amplifier does not overheat. To provide an
adequate heat sink, you must determine the maximum power
dissipation. The following discussions detail methods and
considerations that affect SOA requirements and power
dissipation and heat sink requirements.
SHORT-CIRCUITS
Some amplifier applications must be designed to survive a
short-circuit to ground. This forces the full power supply
voltage (either V+ or V–) across the conducting output
transistor. The amplifier will immediately go into current
limit. To survive this condition a power op amp with
adjustable current limit must be set to limit at a safe level.
Example 1
What is the maximum current limit value which would
protect against short-circuit to ground when OPA502
(Figure 2) power supplies are ±40V?
Answer—
If the case temperature could be held to 25°C, the
current limit could be set to 3A, maximum. This would
be unlikely, however, since the amplifier would dissi-
pate 120W during short-circuit. It would require an
“infinite” or ideal heat sink to maintain the case tem-
perature at 25°C in normal room ambient conditions.
If the case temperature were held to 85°C, a 2A current
limit would be safe. Power dissipation would be 80W,
requiring a heat sink of 0.75°C/W—a large heat sink.
(See Application Bulletin AB-038 for heat sink calcula-
tions.)
If the op amp must survive a short-circuit to one of the power
supplies, for instance, the maximum VCE would be the total
of both supplies—a very demanding case.
Not all applications must (or can be) designed for short-
circuit protection. It is a severe condition for a power
amplifier. Additional measures such as fuses or circuitry to
sense a fault condition can limit the time the amplifier must
endure a short-circuit. This can greatly reduce the heat sink
requirement.
An additional feature of the OPA502 and OPA512 power
amplifiers, the optional fold-over circuit, can be connected
on the current limit circuit. This can be set to reduce the
current limit value when VCE is large—exactly the condition
that exists with a short-circuit. While useful in some appli-
cations, the foldover limiter can produce unusual behavior—
especially with reactive loads. See the OPA502 data sheet
for details.
RESISTIVE LOADS—DC OPERATION
Consider a power amplifier driving a resistive load. It is
tempting to check for safe operation only at maximum
output voltage and current. But this condition is not usually
the most stressful.
At maximum output voltage, the voltage across the conduct-
ing transistor, VCE, is at a minimum and the power dissipa-
tion is low. In fact, if the amplifier output could swing all the
way to the power supply rail, the current output would be
high, but the amplifier power dissipation would be zero
because VCE would be zero.
Figure 3 plots power from the power supply, load power,
and amplifier power dissipation as a function of output
voltage delivered to a resistive load. The power delivered to
the load increases with the square of the output voltage
(P = I2R), while the power from the power supply increases
linearly. The amplifier dissipation (equal to the difference of
the first two curves) follows a parabola. If the amplifier
output could swing all the way to the power supply rail
(dotted portion of lines), all the power from the supply
would be delivered to the load and the amplifier dissipation
would be zero.
Peak amplifier dissipation occurs at an output voltage of
(V+)/2, or 50% output. At this point, VCE is (V+)/2 and IO is
(V+)/(2RL). The amplifier dissipation at this worst-case
point is the product of VCE and IO, or (V+)2/(4RL). Check this
condition to assure that it is within the SOA of the amplifier.
Also be sure that you have sufficient heat sinking for the
calculated power dissipation to prevent overheating.
2