Application Information
(Continued)
The LM4849’s power dissipation is twice that given by Equa-
tion (2) or Equation (3) when operating in the single-ended
mode or bridge mode, respectively. Twice the maximum
power dissipation point given by Equation (3) must not ex-
ceed the power dissipation given by Equation (4):
P
DMAX
' = (T
JMAX
T
A
)/
θ
JA
(4)
The LM4849’s T
= 150C. In the MH package soldered
to a DAP pad that expands to a copper area of 2in
2
on a
PCB, the LM4849’s
θ
is 41C/W. At any given ambient
temperature T
, use Equation (4) to find the maximum inter-
nal power dissipation supported by the IC packaging. Rear-
ranging Equation (4) and substituting P
for P
' re-
sults in Equation (5). This equation gives the maximum
ambient temperature that still allows maximum stereo power
dissipation without violating the LM4849’s maximum junction
temperature.
T
A
= T
JMAX
– 2*P
DMAX
θ
JA
(5)
For a typical application with a 5V power supply and an 4
load, the maximum ambient temperature that allows maxi-
mum stereo power dissipation without exceeding the maxi-
mum junction temperature is approximately 45C for the MH
package.
T
JMAX
= P
DMAX
θ
JA
+ T
A
(6)
Equation (6) gives the maximum junction temperature
T
. If the result violates the LM4849’s T
of 150C,
reduce the maximum junction temperature by reducing the
power supply voltage or increasing the load resistance. Fur-
ther allowance should be made for increased ambient tem-
peratures.
The above examples assume that a device is a surface
mount part operating around the maximum power dissipation
point. Since internal power dissipation is a function of output
power, higher ambient temperatures are allowed as output
power or duty cycle decreases.
If twice the value given by Equation (3) is greater than that of
Equation (4), then decrease the supply voltage, increase the
load impedance, or reduce the ambient temperature. If these
measures are insufficient, a heat sink can be added to
reduce
θ
. The heat sink can be created using additional
copper area around the package, with connections to the
ground pin(s), supply pin and amplifier output pins. External,
solder attached SMT heatsinks such as the Thermalloy
7106D can also improve power dissipation. When adding a
heat sink, the
θ
JA
is the sum of
θ
JC
,
θ
CS
, and
θ
SA
. (
θ
JC
is the
junction-to-case thermal impedance,
θ
CS
is the case-to-sink
thermal impedance, and
θ
is the sink-to-ambient thermal
impedance.) Refer to the
Typical Performance Character-
istics
curves for power dissipation information at lower out-
put power levels.
POWER SUPPLY BYPASSING
As with any power amplifier, proper supply bypassing is
critical for low noise performance and high power supply
rejection. Applications that employ a 5V regulator typically
use a 10 μF in parallel with a 0.1 μF filter capacitor to
stabilize the regulator’s output, reduce noise on the supply
line, and improve the supply’s transient response. However,
their presence does not eliminate the need for a local 1.0 μF
tantalum bypass capacitance connected between the
LM4849’s supply pins and ground. Do not substitute a ce-
ramic capacitor for the tantalum. Doing so may cause oscil-
lation. Keep the length of leads and traces that connect
capacitors between the LM4849’s power supply pin and
ground as short as possible. Connecting a 1μF capacitor,
C
B
, between the BYPASS pin and ground improves the
internal bias voltage’s stability and improves the amplifier’s
PSRR. The PSRR improvements increase as the bypass pin
capacitor value increases. Too large a capacitor, however,
increases turn-on time and can compromise the amplifier’s
click and pop performance. The selection of bypass capaci-
tor values, especially C
B
, depends on desired PSRR require-
ments, click and pop performance (as explained in the sec-
tion,
Proper Selection of External Components
), system
cost, and size constraints.
PROPER SELECTION OF EXTERNAL COMPONENTS
Optimizing the LM4849’s performance requires properly se-
lecting external components. Though the LM4849 operates
well when using external components with wide tolerances,
best performance is achieved by optimizing component val-
ues.
The LM4849 is unity-gain stable, giving a designer maximum
design flexibility. The gain should be set to no more than a
given application requires. This allows the amplifier to
achieve minimum THD+N and maximum signal-to-noise ra-
tio. These parameters are compromised as the closed-loop
gain increases. However, low gain circuits demand input
signals with greater voltage swings to achieve maximum
output power. Fortunately, many signal sources such as
audio CODECs have outputs of 1V
(2.83V
P-P
). Please
refer to the
Audio Power Amplifier Design
section for more
information on selecting the proper gain.
Input Capacitor Value Selection
Amplifying the lowest audio frequencies requires a high
value input coupling capacitor (0.33μF in
Figure 1
). A high
value capacitor can be expensive and may compromise
space efficiency in portable designs. In many cases, how-
ever, the speakers used in portable systems, whether inter-
nal or external, have little ability to reproduce signals below
150Hz. Applications using speakers with this limited fre-
quency response reap little improvement by using a large
input capacitor.
Besides affecting system cost and size, the input coupling
capacitor has an effect on the LM4849’s click and pop per-
formance. When the supply voltage is first applied, a tran-
sient (pop) is created as the charge on the input capacitor
changes from zero to a quiescent state. The magnitude of
the pop is directly proportional to the input capacitor’s size.
Higher value capacitors need more time to reach a quiescent
DC voltage (usually V
/2) when charged with a fixed cur-
rent. The amplifier’s output charges the input capacitor
through the feedback resistors, (R1, 2, 7, 8). Thus, pops can
be minimized by selecting an input capacitor value that is no
higher than necessary to meet the desired 3dB frequency.
As shown in
Figure 1
, the input resistors (R3, 4, 5, 6) and the
input capacitors (CIN = 0.33μF) produce a 6dB high pass
filter cutoff frequency that is found using Equation (7).
L
www.national.com
14