The TFA9812 is a high-efficiency Bridge Tied Load (BTL) stereo Class-D audio amplifier
with a digital I2S audio input. It is available in a HVQFN48 package with exposed die
paddle. The exposed die paddle technology enhances the thermal and electrical
performances of the device.
The TFA9812 features digital sound processing and audio poweramplification.Itsupports
I2C control mode and Legacy mode. In Legacy mode I2C involvement is not needed
because the key features are controlled by hardware pin connections.
2.Features
2.1 General features
A continuous time output power of 2 × 12W(RL=8Ω,V
an external heat sink. Due to the implementation of a programmable thermal foldback
even for high supply voltages, higher ambient temperatures, and/or lower load
impedances, the device operates without sound interrupting behavior.
TFA9812 is designed in such a way that it starts up easily (no special power-up sequence
required). It features various soft and hard impact protection mechanisms to ensure an
application that is both user friendly and robust.
A modulation technique is applied for the TFA9812, which supports common mode choke
approach (1 common mode choke only per BTL amplifier stage). This minimizes the
number of external components.
n 3.3 V and 8 V to 20 V external power supply
n High efficiency and low power dissipation
n Speaker outputs fully short circuit proof across load, to supply lines and ground
n Pop noise free at power-up/power-down and sample rate switching
n Low power Sleep mode
n Overvoltage and undervoltage protection on the 8 V to 20 V power supply
n Undervoltage protection on the 3.3 V power supply
n Overcurrent protection (no audible interruptions)
n Overdissipation protection
n Thermally protected and programmable thermal foldback
n Clock error protection
n I2C mode control or Legacy mode (i.e. no I2C) control
n Four different I2C addresses supported
n Internal Phase-Locked Loop (PLL) without using external components
= 15 V) is supported without
DDP
NXP Semiconductors
n No high system clock required (PLL is able to lock on BCK)
n No external heat sink required
n 5 V tolerant digital inputs
n Supports dual coil inductor application
n Easy application and limited external components required
2.2 DSP features
n Digital parametric 10-band equalizer
n Digital volume control per channel
n Selectable +24 dB gain boost
n Analog interface to digital volume control in Legacy mode
n Digital clip level control
n Soft and hard mute
n Thermal foldback threshold temperature control
n De-emphasis
n Output power limiting control
n Polarity switch
n Four Pulse Width Modulation (PWM) switching frequency settings
TFA9812
BTL stereo Class-D audio amplifier with I2S input
2.3 Audio data input interface format support
n Master or slave Master Clock (MCLK), Bit Clock (BCK) and Word Select (WS) signals
n Philips I2S, standard I2S
n Japanese I2S, Most Significant Bit (MSB) justified
n Sony I2S, Least Significant Bit (LSB) justified
n Sample rates from 8 kHz to 192 kHz
3.Applications
n Digital-in Class-D audio amplifier applications
n CRT and flat-panel television sets
n Flat-panel monitors
n Multimedia systems
n Wireless speakers
n Docking stations for MP3 players
Preliminary data sheetRev. 02 — 22 January 20094 of 66
xxxxxxxxxxxxxxxxxxxxx xxxxxxxxxxxxxxxxxxxxxxxxxx xxxxxxx x x x xxxxxxxxxxxxxxxxxxxxxxxxxxxxxx xxxxxxxxxxxxxxxxxxx xx xx
xxxxx xxxxxxxxxxxxxxxxxxxxxxxxxxx xxxxxxxxxxxxxxxxxxx xxxxxx xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx xxxxxxxxxxxx x x
xxxxxxxxxxxxxxxxxxxxx xxxxxxxxxxxxxxxxxxxxxxxxxxxxxx xxxxx xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx xxxxxxxx
PNegative power supply voltage for channel 1 and channel 2
PNegative power supply voltage for channel 1 and channel 2
28STAB1ODecoupling of internal 11 V regulator for channel 1 drivers
29DIAGOFault mode indication output (open-drain pin)
30CDELAYITiming reference
31POWERUPIPower-up pin to switch betweenSleep and other operational
modes
32AVOLIAnalog volume control (Legacy mode)
33ENABLEIEnable input to switch between 3-state and other
operational modes
34GAINIGain selection input to select between 0 dB and +24 dB
gain (Legacy mode)
35CSELIControl selection input to select between Legacy mode
2
C bus control) and I2C bus control
(no I
2
36ADSEL2/PLIM2 IAddress selection in I
C mode input 2, power limiter
selection input 2 in Legacy mode
2
37ADSEL1/PLIM1 IAddress selection in I
C mode input 1, power limiter
selection input 1 in Legacy mode
2
38SCL/SFORII
C bus clock input in I2C mode, I2S serial data format
selection input in Legacy mode
2
39SDA/MSI/OI
C bus data input and output in I2C mode, master/slave
selection input in Legacy mode
40V
DDD(3V3)
PDigital supply voltage (3.3 V)
41STABDO1.8 V digital stabilizer output
42REFDPDigital reference voltage
43TEST2ITest signal input 2; for test purposes only (connect to V
2
44DATAII
45WSI/OI
S bus data input
2
S bus word select input (I2S slave mode) or output (I2S
SS
master mode)
2
46BCKI/OI
S bus bit clock input (I2S slave mode) or output (I2S
Preliminary data sheetRev. 02 — 22 January 20097 of 66
NXP Semiconductors
TFA9812
BTL stereo Class-D audio amplifier with I2S input
Table 3.Pinning description TFA9812
PinSymbolType Description
47MCLKI/OMaster clock input (I2S slave mode) or output (I2S master
48V
Exposed
die-paddle
SS2
-PPCB ground reference
8.Functional description
8.1 General
The TFA9812 is a high-efficiency stereo BTL Class-D amplifier with a digital I2S audio
input. It supports all commonly used I2S formats.
Figure 1 shows the functional block diagram, which includes the key function blocks of the
TFA9812. In the digital domain the audio signal is processed and converted to a pulse
width modulated signal using BD modulation. A BTL configured power comparator carries
out power amplification.
The audio signal processing path is as follows:
1. The Digital Audio Input (DAI) block translates the I2S (-like) input signal into a
standard internal stereo audio stream.
2. The 10-band parametric equalizer can optionally equalize the stereo audio stream.
Both channels have separate equalization streams. It can be used forspeaker transfer
curve compensation to optimize the audio performance of applied speakers.
3. Volume control in the TFA9812 is done by attenuation. The attenuation depends on
the volume control settings and the thermal foldbackvalue. Soft mute is also arranged
at this part. In Legacy mode the volume control is done by an on-board
Analog-to-Digital Converter (ADC) which measures the analog voltage on pin 32.
4. The interpolation filter interpolates from 1 fs to the PWM controller sample rate
(2048 fs at 44.1 kHz) by cascading FIR filters.
5. The gain block can boost the signal with 0 dB or +24 dB. Four specific gain settings
are also provided in this block. These specific gain settings are related to maximum
clip levels of < 0.5 %, 10 %, 20 % or 30 % THD at the TFA9812 output. These
maximum clip levels are only valid with the gain boost set to 0 dB and a 0 dBFS input
signal.
6. The power limiter limits the maximum output signal of the TFA9812. The power limiter
settings are 0 dB, −1.5 dB, −3 dB, and −4.5 dB. This function can be used to reduce
the maximum output power delivered to the speakers at a fixed supply voltage and
speaker impedance.
7. The PWM controller block transforms the audio signal into a BD-modulated PWM
signal. The BD-modulation provides a high signal-to-noise performance and
eliminates clock jitter noise.
8. Via four differential comparators the PWM signals are amplified by two BTL power
output stages. By default the left audio signal is connected to channel 1 and the right
audio signal to channel 2.
Preliminary data sheetRev. 02 — 22 January 20098 of 66
NXP Semiconductors
The block control defines the operational control settings of the TFA9812 in line with the
actual I2C settings and the pin-controlled settings.
The PLL block creates the system clock and can take the I2S BCK, the MCLK or an
external crystal as reference source.
The following protections are built into the TFA9812:
• Thermal Foldback (TF)
• OverTemperature Protection (OTP)
• OverCurrent Protection (OCP)
• OverVoltage Protection (OVP)
• UnderVoltage Protection (UVP)
• Window Protection (WP)
• Lock Protection (LP)
• UnderFrequency Protection (UFP)
• OverFrequency Protection (OFP)
• Invalid BCK Protection (IBP)
• DC-blocking
• ElectroStatic Discharge (ESD)
TFA9812
BTL stereo Class-D audio amplifier with I2S input
8.2 Functional modes
8.2.1 Control modes
The two control modes of the TFA9812 are I2C and legacy.
2
• In I
C mode the I2C format control is enabled.
• In Legacy mode a pin-based subset of the control options is available. The control
settings for features which are not available in Legacy mode are set to the default I2C
register settings.
The control mode is selected via pin CSEL as shown in Table 4.
Table 4.Control mode selection
CSEL Pin valueControl mode
0Legacy (no I
2
1I
In the functional descriptions below the control for the various functions will be described
for each control mode. Section 9.6 summarizes the support given by each control mode
for the various TFA9812 functions.
8.2.2 Key operating modes
There are six key operating modes:
C
2
C)
• InSleep mode the voltage supplies are present, but power consumption for the whole
device is reduced to the minimum level. The output stages in Sleep mode are 3-state
and I2C communication is disabled.
Preliminary data sheetRev. 02 — 22 January 20099 of 66
NXP Semiconductors
TFA9812
BTL stereo Class-D audio amplifier with I2S input
• In Soft mute mode the I
2
S input signal is overruled with a soft mute.
– In Legacy control mode the analog input pin AVOL controls Soft mute mode.
– In I2C control mode I2C control can be used to enable an automatic soft mute
function. See also Section 8.5.3.
• In Hard mute mode the PWM controller is overruled with a 50 % duty cycle square
pulse. The Hard mute mode is only available in I2C control mode.
• In Operating mode the TFA9812 amplifies the I
2
S audio input signal in line with the
actual control setting.
• In 3-state mode the output stages are switched off.
• Fault mode is entered when a fault condition is detected by one or more of the
protection mechanisms implemented in the TFA9812. In Fault mode the actual device
configuration depends on the fault detected: see Section 8.7 for more information.
Fault mode is for a subset of the faults flagged on the DIAG output pin. When the
DIAG pin is flagged the output stages will be forced to enter 3-state mode. In Sleep
mode the DIAG pin will not flag fault modes.
Table 5.Operational mode selection
Pin:DIAG OutputOperational mode
selected:
POWERUP ENABLECSELAVOL
0---floatingSleep mode
1---0 / floatingFault mode (enabled by
[1] Clocking faults do not trigger DIAG output.
[2] Under these conditions soft mute still has to be enabled by the appropriate I2C setting.
8.2.3 I2S master/slave modes and MCLK/BCK clock modes
The I2S interface can be set in master or in slave.
2
• In I
S master mode the PLL locks to the output signal of the internal crystal oscillator
circuit which uses an external crystal. The BCK, WS and MCLK signals are generated
by the TFA9812. On the MCLK pin the TFA9812 delivers a master clock running at the
crystal frequency.
2
• In I
S slave mode the PLL can lock to:
– The external MCLK signal on the MCLK pin called MCLK clock mode.
– The I2S input BCK signal on the BCK pin called BCK clock mode.
The I2S master or slave mode can be selected:
2
• In I
C control mode by selecting the right I2C setting.
• In legacy control mode by selecting the right setting on the SDA/MS pin.
Preliminary data sheetRev. 02 — 22 January 200910 of 66
NXP Semiconductors
Table 6.I2S master/slave mode selection
Pin valueClock modeI2S mode
CSELSDA/MS
00legacyslave
01legacymaster
1- I
[1] Under these conditions the mode is enabled by the appropriate I2C setting.
In I2S slave mode selection between BCK and MCLK clock modes is automatic.
MCLK clock mode is given higher priority than BCK. If the MCLK clock is judged valid by
the protection circuit then MCLK clock mode is enabled. BCK clock mode is enabled when
the MCLK clock is invalid (e.g. not available) and the BCK clock is judged valid by the
protection circuit (see Section 8.7.11).
Table 7 shows the supported crystal frequencies in I2S master mode.
Table 8 shows the supported MCLK frequencies in MCLK mode (I2S slave mode).
TFA9812
BTL stereo Class-D audio amplifier with I2S input
2
Cslave or master
[1]
Table 9 shows the supported BCK frequencies in BCK mode (I2S slave mode).
Table 7.Valid crystal frequencies in I2S master mode
Preliminary data sheetRev. 02 — 22 January 200913 of 66
NXP Semiconductors
In I2C control mode communication is enabled after 4 ms. The preferred I2C settings can
be made within 66 ms before the PLL starts running. Finally, the output stages are
enabled and the audio level is increased via a demute sequence if mute has previously
been disabled.
Remark: In I2C mode soft mute is enabled by default. It can be disabled at any time while
I2C communication is valid. In order to prevent audio clicks volume control (default setting
is 0 dB) should be set before soft mute is disabled.
Remark: For a proper start-up in I2S master mode and I2C mode the following sequence
should be followed:
1. The I2S master setting should be set and keep the default sample rate setting active.
2. Next, another sample rate setting than the default one should be selected.
3. Finally, when the default sample rate is used the default sample rate setting should be
8.3.2 Power-down
Figure 3 includes the power-down timing while Table 11 shows the pin control for enabling
power-down.
TFA9812
BTL stereo Class-D audio amplifier with I2S input
selected again.
Table 11. Power-up/power-down selection
Power-up pin
value
0Power-down (Sleep mode)
1Power-up
Description
Putting the TFA9812 into power-down is equivalent to enabling Sleep mode
(see Section 8.2.2). This mode is entered immediately and no additional clock cycles are
required.
In order to prevent audible clicks, soft mute should be enabled at least T
d(soft_mute)
seconds before enabling Sleep mode.
The specified low current and power conditions in Table 1 are valid within 10 µs after
enabling Sleep mode.
8.4 Digital audio data input
8.4.1 Digital audio data format support
The TFA9812 supports a commonly used range of I2S and I2S-like digital audio data input
formats. These are listed in Table 12.
Table 12. Supported digital audio data formats
BCK frequency Interface format (MSB first)Supported in I2C
Preliminary data sheetRev. 02 — 22 January 200914 of 66
I2S up to 16-bit datayesyes
MSB-justified 16-bit datayesyes
LSB-justified 16-bit datayesyes
I2S up to 24-bit datayesyes
MSB-justified up to 24-bit datayesyes
Supported in Legacy
control mode
NXP Semiconductors
Table 12. Supported digital audio data formats
BCK frequency Interface format (MSB first)Supported in I2C
48 f
48 f
48 f
48 f
64 f
64 f
64 f
64 f
64 f
64 f
Remark: Only MSB-first formats are supported.
TFA9812
BTL stereo Class-D audio amplifier with I2S input
Supported in Legacy
control mode
s
s
s
s
s
s
s
s
s
s
LSB-justified 16-bit datayesno
LSB-justified 18-bit datayesno
LSB-justified 20-bit datayesno
LSB-justified 24-bit datayesyes
I2S up to 24-bit datayesyes
MSB-justified up to 24-bit datayesyes
LSB-justified 16-bit datayesno
LSB-justified 18-bit datayesno
LSB-justified 20-bit datayesno
LSB-justified 24-bit datayesno
Preliminary data sheetRev. 02 — 22 January 200915 of 66
NXP Semiconductors
In I2C control mode the following sample frequency fs can be used: 8 kHz, 11.025 kHz,
12 kHz, 16 kHz, 22.05 kHz, 24 kHz, 32 kHz, 44.1 kHz, 48 kHz, 64 kHz, 88.2 kHz, 96 kHz,
128 kHz, 176.4 kHz or 192 kHz. The I2C control for fs selection can be found in
Section 9.5.7.
In Legacy control mode the following sample frequencies (fs) can be used: 32 kHz,
44.1 kHz or 48 kHz.
8.4.2 Digital audio data format control
The BCK-to-WS and MCLK-to-WS frequency ratios are automatically detected, so no
control settings need to be configured for these.
In I2C control mode all the formats listed in Table 12 are supported. The appropriate I2C
controls for selecting the supported formats can be found in Section 9. In the Legacy
control mode only a subset of the supported formats can be used. These are shown in
Table 12 and the required pin control is given in Table 13.
See Section 8.2.1 for details of how to enable Legacy control mode.
Table 13. Digital audio data format selection in Legacy control mode
SCL/SFOR pin valueInterface formats (MSB-first)
0I
1MSB-justified
TFA9812
BTL stereo Class-D audio amplifier with I2S input
2
S
8.5 Digital signal-processing features
8.5.1 Equalizer
8.5.1.1 Equalizer options
The equalizer function can be bypassed and the equalizer can be configured to either a
5-band or 10-band function. These settings are for both audio channels simultaneously.
There are 20 bands in the equalizer. These are distributed as follows:
• Bands A1 to A5 are bands 1 to 5 of output 1 (used in 5-band and 10-band
configuration).
• Bands B1 to B5 are bands 1 to 5 of output 2 (used in 5-band and 10-band
configuration).
• Bands C1 to C5 are bands 6 to 10 of output 1 (used in 10-band configuration only).
• Bands D1 to D5 are bands 6 to 10 of output 2 (used in 10-band configuration only).
In I2C control mode each band can be configured separately using I2C register settings.
In Legacy control mode the equalizer is bypassed.
8.5.1.2 Equalizer band function
The shape of each parametric equalizer band is determined by the three filter parameters:
Preliminary data sheetRev. 02 — 22 January 200916 of 66
NXP Semiconductors
In the above equation fcis the center frequency and fs is the sample frequency.
The definition of the quality factor is the center frequency divided by the 3 dB bandwidth,
see Equation 1. In parametric equalizers this is only valid when the gain is set very small
(−30 dB).
Q
Each band filter can be programmed to perform a band-suppression (G < 1) or a
band-amplification (G > 1) function around the center frequency.
Each band of the TFA9812 equalizer has a second-order Regalia-Mitra all-pass filter
structure. The structure is shown in Figure 5.
f
c
-----------------
f2f1–
;=
f1:20
f2:20
BTL stereo Class-D audio amplifier with I2S input
A
10
10
f
1
log3dB fcf1>=
--------
A
f
c
A
f
2
log3dB f2fc>,=
--------
A
f
c
TFA9812
(1)
s
X(z)
A(z)
Fig 5.Regalia filter flow-diagram
+
+
−
½
+
K0/2
Y(z)
010aaa406
The transfer function of this all-pass filter is shown in Equation 2:
Hz() 12⁄1Az()+()K02⁄1Az()–()⋅+⋅=
A(z) is the second-order filter structure. The transfer function of A(z) is shown in
Preliminary data sheetRev. 02 — 22 January 200917 of 66
NXP Semiconductors
TFA9812
BTL stereo Class-D audio amplifier with I2S input
K0G=
K
1
K
2Qωsin–()2Qωsin+()⁄=
2
ωcos–=
G1≥
The ranges of the TFA9812 parametric equalizer settings for each band are:
• The Gain, G is from −30 dB to +12 dB.
• The center frequency, f
is from 0.0004 * fs to 0.49 * fs.
c
• The quality factor Q is from 0.001 to 8.
Using I2C control, filter coefficients need to be entered for each filter stage to configure it
as desired.
Figure 6, Figure 7 and Figure 8 show some of the possible transfer functions of the
equalizer bands. The relations are symmetrical for the suppression and amplification
functions. A skewing effect can be observed for the higher frequencies.
Different configurations are available for the same filter transfer function, thus allowing
optimum numerical noise performance. The binary filter configuration parameters t1and t
control the actual configuration and should be chosen according to Equation 6.
0ω<=π 2⁄
t
=
1ω>π 2⁄
1
0k2>=0
1k2<0
=
t
2
(5)
2
(6)
A maximum of 12 dB amplification per equalizer stage can be achievedwith respect to the
input signal. Each band of the equalizer is provided with a −6 dB amplification, so in order
to prevent numerical clipping for some filter settings with over 6 dB of amplification, band
filters can be scaled by 0 dB or −6 dB. For optimum numerical noise performance steps of
−6 dB amplification should be applied to the highest possible sections that are still within
scale signal processing safeguards. Band filters can be scaled with the binary parameters
listed in Table 14.
Table 14. Equalizer scale factor coding
sscale factor (dB)
00
1−6
8.5.1.3 Equalizer band control
For compact representation with positive signed parameters, parameters k1’ and k2’ are
introduced in Equation 7.
The parameters k0, k1', k2', t1, t2 and s must be combined in two 16-bit control words,
word1 and word2, and must fit within the representation given in Table 15. Parameters k1'
and k2' are unsigned floating-point representations in Equation 8.
Preliminary data sheetRev. 02 — 22 January 200918 of 66
NXP Semiconductors
k1′
k2′
1k
–t11=
=
1k
+t
1k
–t20=
=
1k
+t
TFA9812
BTL stereo Class-D audio amplifier with I2S input
1
1
2
2
0=
1
(7)
1=
2
kxM2E–⋅
=
M1<
In Equation 8, M is the unsigned mantissa and E the negative signed exponent. For
example, in word2 bits [14:8] = [0111 010] represent k2' = (7/24) × 2−2 = 1.09375 10−1.
Table 15. Equalizer control word construction
WordSectionData
word115t
1
word1[14:4]11 mantissa bits of k1’
word1[3:0]Four exponent bits of k
word215t
2
’
1
word2[14:11]Four mantissa bits of k2’
word2[10:8]Three exponents bits of k
word2[7:1]k
0
’
2
word20s
Section 9.5.4 shows the I2C address locations of the controls for various bands of the
equalizer.
010aaa222
Gain
(dB)
12
8
Q1 = 0.27
Q2 = 0.61
Q3 = 1.65
(8)
4
0
1
10
2
10
3
10
4
10
Frequency (Hz)
5
10
Fig 6.Transfer functions for several quality factors Q
Preliminary data sheetRev. 02 — 22 January 200919 of 66
NXP Semiconductors
TFA9812
BTL stereo Class-D audio amplifier with I2S input
12
Gain
(dB)
8
4
0
1
10
2
10
3
10
Fig 7.Transfer functions for several center frequencies f
12
Gain
(dB)
6
10
c
4
Frequency (Hz)
010aaa223
010aaa224
5
10
0
-6
-12
1
10
Fig 8.Transfer functions for several gain factors G
8.5.2 Digital volume control
In I2C control mode both audio channels have separate digital volume control. In Legacy
control mode the volume control of both channels is common and the volume control
setting depends on the supply voltage on the pin AVOL (32).
8-bit volume control is availableper channel. This is dB-linear down to −124 dB in steps of
0.5 dB. The last step of the volume control is mute.
Table 16 shows the various settings and their related channel suppression: