• Extensive Signal Processing Options• Portable Computing
• Stereo Digital Microphone Input–
• Stereo Headphone Outputs
• Low Power Analog Bypass Mode
• Programmable PLL
• Integrated LDO
• 4mm × 4mm QFN and 2.7mm × 2.7mm
WCSPPackage
Chapter 1
SLAU434–May 2012
TLV320DAC3203 Overview
• Chapter 1: Device Overview
• Chapter 2: TLV320DAC3203 Application
• Chapter 3: Device Initialization
• Chapter 4: Example Setups
• Chapter 5: Register Map and Descriptions
The TLV320DAC3203 (sometimes referred to as the DAC3203) is a flexible, low-power, low-voltage
stereo audio codec with programmable outputs, PowerTune capabilities, fixed predefined and
parameterizable signal processing blocks, integrated PLL, integrated LDO and flexible digital interfaces.
Extensive register-based control of power, input/output channel configuration, gains, effects, pinmultiplexing and clocks is included, allowing the device to be precisely targeted to its application.
Combined with the advanced PowerTune technology, the device can cover operations from 8kHz mono
voice playback to stereo 192kHz DAC playback, making it ideal for portable battery-powered audio and
telephony applications.
The record path of the TLV320DAC3203 consists of a stereo digital microphone PDM interface (not
available when using SPI control interface) typically used at 64Fs or 128Fs.
The playback path offers signal processing blocks for filtering and effects, true differential output signal,
flexible mixing of DAC and analog input signals as well as programmable volume controls. The
TLV320DAC3203 contains two high-power output drivers which can be configured in multiple ways,
including stereo, and mono BTL. The integrated PowerTune technology allows the device to be tuned to
just the right power-performance trade-off. Mobile applications frequently have multiple use cases
requiring very low-power operation while being used in a mobile environment. When used in a docked
environment power consumption typically is less of a concern while lowest possible noise is important.
With PowerTune the TLV320DAC3203 can address both cases.
The voltage supply range for the TLV320DAC3203 for analog is 1.5V–1.95V, and for digital it is
1.26V–1.95V. To ease system-level design, a low-dropout regulator (LDO) is integrated to generate the
appropriate analog supply from input voltages ranging from 1.8V to 3.6V. Digital I/O voltages are
supported in the range of 1.1V–3.6V.
The required internal clock of the TLV320DAC3203 can be derived from multiple sources, including the
MCLK, BCLK or GPIO pins or the output of the internal PLL, where the input to the PLL again can be
derived from the MCLK, BCLK or GPIO pins. Although using the internal, fractional PLL ensures the
availability of a suitable clock signal, it is not recommended for the lowest power settings. The PLL is
highly programmable and can accept available input clocks in the range of 512kHz to 50MHz.
The device is available in the 4mm × 4mm QFN and 2.7mm × 2.7mm WCSPpackage.
Only a small number of digital pins are dedicated to a single function; whenever possible, the digital pins
have a default function, and also can be reprogrammed to cover alternative functions for various
applications.
The fixed-function pins are Reset and the SPI_Select pin, which are HW control pins. Depending on the
state of SPI_Select, the two control-bus pins SCL/SS and SDA/MOSI are configured for either I2C or SPI
protocol.
Other digital IO pins can be configured for various functions via register control. An overview of available
functionality is given in Section 2.1.3.
2.1.2 Analog Pins
Analog functions can also be configured to a large degree. For minimum power consumption, analog
blocks are powered down by default. The blocks can be powered up with fine granularity according to the
application needs.
Chapter 2
SLAU434–May 2012
TLV320DAC3203 Application
2.1.3 Multifunction Pins
Table 2-1 shows the possible allocation of pins for specific functions. The PLL input, for example, can be
programmed to be any of 4 pins (MCLK, BCLK, DIN, GPIO).
: The MCLK pin can be used to drive the PLL and Codec Clock inputs simultaneously
(2)S(2)
: The BCLK pin can be used to drive the PLL and Codec Clock and audio interface bit clock inputs simultaneously
(3)S(3)
: The GPIO/MFP5 pin can be used to drive the PLL and Codec Clock inputs simultaneously
(4)
D: Default Function
(5)
E: The pin is exclusively used for this function, no other function can be implemented with the same pin (e.g. if GPIO/MFP5 has
been allocated for General Purpose Output, it cannot be used as the INT1 output at the same time)
To configure the settings seen in Table 2-1, please see the letter/number combination in for the
appropriate registers to modify. In , the letter/number combination represents the row and the column
number from Table 2-1 in bold type.
Please be aware that more settings may be necessary to obtain a full interface definition matching the
application requirement (e.g. register Page 0 / Register 25 to 33).
The analog I/O path of the TLV320DAC3203 offers a variety of options for signal conditioning and routing:
•2 headphone amplifier outputs
•Analog gain setting
•Single ended and differential modes
2.2.1 Analog Low Power Bypass
The TLV320DAC3203 offers an analog-bypass mode. An analog signal can be routed from the analog
input pin to the output amplifier. Neither the digital-input processing blocks nor the DAC resources are
required for such operation; this supports low-power operation during analog-bypass mode.
In analog low-power bypass mode, line-level signals can be routed directly from the analog inputs INL to
the left headphone amplifier (HPL) and INR to HPR.
The stereo headphone drivers on pins HPL and HPR can drive loads with impedances down to 16Ω in
single-ended AC-coupled headphone configurations, or loads down to 32Ω in differential mode, where a
speaker is connected between HPL and HPR. In single-ended drive configuration these drivers can drive
up to 15mW power into each headphone channel while operating from 1.8V analog supplies. While
running from the AVdd supply, the output common-mode of the headphone driver is set by the commonmode setting of analog inputs to allow maximum utilization of the analog supply range while
simultaneously providing a higher output-voltage swing. In cases when higher output-voltage swing is
required, the headphone amplifiers can run directly from the higher supply voltage on LDOIN input (up to
3.6V). To use the higher supply voltage for higher output signal swing, the output common-mode can be
adjusted to either 1.25V, 1.5V or 1.65V. When the common-mode voltage is configured at 1.65V and
LDOIN supply is 3.3V, the headphones can each deliver up to 40mW power into a 16Ω load.
The headphone drivers are capable of driving a mixed combination of DAC signal and bypass from analog
input INL and INR. The analog input signals can be attenuated up to 72dB before routing. The level of the
DAC signal can be controlled using the digital volume control of the DAC. To control the output-voltage
swing of headphone drivers, the digital volume control provides a range of –6.0dB to +29.0dB
of 1dB. These level controls are not meant to be used as dynamic volume control, but more to set output
levels during initial device configuration. Refer to for recommendations for using headphone volume
control for achieving 0dB gain through the DAC channel with various configurations.
Figure 2-1. Low Power Analog Bypass
www.ti.com
(1)
in steps
2.2.2.1Stereo Single Ended Configuration
(1)
16
If the device must be placed into 'mute' from the –6.0dB setting, set the device at a gain of –5.0dB first, then place the device into mute.
The left and right DAC channels are routed to the corresponding left and right headphone amplifier. This
configuration is also used to drive line-level loads.
The TLV320DAC3203 headphone drivers support pop-free operation. Because the HPL and HPR are
high-power drivers, pop can result due to sudden transient changes in the output drivers if care is not
taken. The most critical care is required while using the drivers as stereo single-ended capacitivelycoupled drivers as shown in Figure 2-2. The output drivers achieve pop-free power-up by using slow
power-up modes. Conceptually, the circuit during power-up can be visualized as
Figure 2-3. Conceptual Circuit for Pop-Free Power-up
The value of R
can be chosen by setting register Page 1, Register 20, Bits D1-D0).
pop
Table 2-3. R
Page 1, Register 20, Bits D1-D0)R
002 kΩ
016 kΩ
1025 kΩ
pop
Values
pop
Value
To minimize audible artifacts, two parameters can be adjusted to match application requirements. The
voltage V
the voltage across R
For a typical R
across R
load
of 32Ω, R
load
at the beginning of slow charging should not be more than a few mV. At that time
load
can be determined as:
load
of 6 kΩ or 25 kΩ will deliver good results (see Table 2-3 for register
pop
settings).
According to the conceptual circuit in Figure 2-3, the voltage on PAD will exponentially settle to the output
common-mode voltage based on the value of R
up mode for time T, such that at the end of the slow power-on period, the voltage on V
and Cc. Thus, the output drivers must be in slow power-
pop
is very close to
pad
the common-mode voltage. The TLV320DAC3203 allows the time T to be adjusted to allow for a wide
range of R
and Ccby programming Page 1, Register 20, Bits D5-D2). For the time adjustments, the
load
value of Ccis assumed to be 47μF. N=5 is expected to yield good results.
1010N=6.0
1011N=7.0
1100N=8.0
1101N=16 (Not valid for R
1110N=24 (Not valid for R
1111N=32 (Not valid for R
Again, for example, for R
=25kΩ)
pop
=25kΩ)
pop
=25kΩ)
pop
=32Ω, Cc=47μF and common mode of 0.9V, the number of time constants
load
and 47μF)
pop
required for pop-free operation is 5 or 6. A higher or lower Ccvalue will require higher or lower value for N.
During the slow-charging period, no signal is routed to the output driver. Therefore, choosing a larger than
necessary value of N results in a delay from power-up to signal at output. At the same time, choosing N to
be smaller than the optimal value results in poor pop performance at power-up.
The signals being routed to headphone drivers (e.g. DAC and IN) often have DC offsets due to less-thanideal processing. As a result, when these signals are routed to output drivers, the offset voltage causes a
pop. To improve the pop-performance in such situations, a feature is provided to soft-step the DC-offset.
At the beginning of the signal routing, a high-value attenuation can be applied which can be progressively
reduced in steps until the desired gain in the channel is reached. The time interval between each of these
gain changes can be controlled by programming Page 1, Register 20, Bits D7-D6). This gain soft-stepping
is applied only during the initial routing of the signal to the output driver and not during subsequent gain
changes.
Page 1, Register 20, Bits D7-D6 Soft-stepping Step Time During initial signal routing
000 ms (soft-stepping disabled)
0150ms
10100ms
11200ms
It is recommended to use the following sequence for achieving optimal pop performance at power-up:
1. Choose the value of R
, N (time constants) and soft-stepping step time for slow power-up.
pop
2. Choose the configuration for output drivers, including common modes and output stage power
connections
3. Select the signals to be routed to headphones.
4. Power-up the blocks driving signals into HPL and HPR, but keep it muted
5. Unmute HPL and HPR and set the desired gain setting.
6. Power-on the HPL and HPR drivers.
7. Unmute the block driving signals to HPL and HPR after the Driver PGA flags are set to indicate
completion of soft-stepping after power-up. These flags can be read from Page 1, Register 63, Bits D7D6).
It is important to configure the Headphone Output driver depop control registers before powering up the
headphone; these register contents should not be changed when the headphone drivers are powered up.
Before powering down the HPL and HPR drivers, it is recommended that user read back the flags in Page
1, Register 63. For example. before powering down the HPL driver, ensure that bit D(7) = 1 and bit D(3) =
1 if INL is routed to HPL and bit D(1) = 1 if the Left Mixer is routed to HPL. The output driver should be
powered down only after a steady-state power-up condition has been achieved. This steady state powerup condition also must be satisfied for changing the HPL/R driver mute control in Page 1, Register 16 and
17, Bits D7), i.e. muting and unmuting should be done after the gain and volume controls associated with
routing to HPL/R finished soft-stepping.
In the differential configuration of HPL and HPR, when no coupling capacitor is used, the slow charging
method for pop-free performance need not be used. In the differential load configuration for HPL and
HPR, it is recommended to not use the output driver MUTE feature, because a pop may result.
During the power-down state, the headphone outputs are weakly pulled to ground using an approximately
50kΩ resistor to ground, to maintain the output voltage on HPL and HPR pins.
2.2.2.2Mono Differential DAC to Mono Differential Headphone Output
Figure 2-4. Low Power Mono DAC to Differential Headphone
This configuration supports the routing of the two differential outputs of the mono, left channel DAC to the
headphone amplifiers in differential mode (Page 1 / Register 12, D3 =1 and Page 1 / Register 13, D4 =1).
2.2.2.2.1 Offset Correction Scheme for Differential DAC to Differential Headphone Output
The TLV320DAC3203 offers an offset correction scheme which is based on calibration during power up.
This scheme will minimize differences in DC voltage between the HPL and HPR outputs.
The offset calibration happens after the headphones are powered up in differential mode. All other
headphone configurations like signal routings, gain settings and mute removal needs to be configured
before the power up of headphones. Any change in these settings while the headphones are powered up
may result in additional differential offsets and are best avoided.
The offset calibration block has a few programmable parameters which the user needs to control. The
user can either choose to calibrate the offset at each power-up of headphones or do it only for first power
up of headphone after system power up and hardware reset.
Programming Page 1 / Register 125, D(1:0) as “01” would cause the offset to be calibrated for each power
up of headphone. This is particularly useful when some headphone configurations like gain or signal
routings change between power ups.
Programming Page 1 / Register 125, D(1:0) as “10” would cause the offset to be calibrated for only the
first power-up of the headphone amplifiers after hardware reset. The calibration data will be stored in
internal memory until the next hardware reset or until AVDD power is removed. Since offset calibration is
not done every time the headphone amplifiers power up the turn on time is reduced by approximately 3.6
ms for subsequent powerups.
Programming Page 1 / Reg 125, D (1:0) as “00” (default) will disable offset correction block.
While the offset is being calibrated no signal should be applied to the headphone amplifier, i.e. the DAC
should be kept muted and analog bypass routing should be kept at highest attenuation setting of 78dB.
The user can read Page 1 / Regiseter 2, D2 to poll if calibration is completed (D2=”1” -> calibration is
completed).
Analog Audio I/O
Please see Section 4.5 for an example setup script enabling offset correction.
2.2.2.3Headphone Amplifier Class-D Mode
By default the headphone amplifiers in the TLV320DAC3203 work in Class-AB mode. By writing to Page
1, Register 3, Bits D7-D6) for the left headphone amplifier, and Page 1, Register 4, Bits D7-D6) with value
11, the headphone amplifiers enter a Class-D mode of operation.
In this mode a high frequency digital pulse-train representation of the DAC signal is fed to the load
connected to HPL and HPR outputs.
Because the output signal is a pulse train switching between Power Supply and Ground, the efficiency of
the amplifier is greatly improved. In this mode however, for good noise performance, care should be taken
to keep the analog power supply clean.
For using the Class-D mode of operation, the following clock-divider condition should be met:
MDAC = I × 4, where I = 1, 2, ..., 32
When a direct digital pulse train is driven out as a signal, high frequencies as a function of pulse train
frequency are also present which lead to power waste. To increase the efficiency and reduce power
dissipation in the load due to these high frequencies, an LC filter should be used in series with the output
and the load. The cutoff frequency of the LC filter should be adjusted to allow audio signals below 20kHz
to pass through, but highly attenuate the high-frequency signal content.
For using the headphones in the Class-D mode of operation, the headphones should first be powered up
in default Class-AB mode to charge the AC-coupling capacitor to the set common mode voltage. Once the
headphone amplifiers have been so powered up, the DAC should be routed to headphones and unmuted
before they can be switched to the Class-D mode. After Class D mode has been turned on, the linear,
Class AB mode amplifier must be turned off. For powering down the headphone amplifiers, the DAC
should first be muted.
See Section 4.3 for an example setup script enabling Class-D mode.
www.ti.com
Figure 2-5. Configuration for Using Headphone Amplifier in Class-D Mode
The TLV320DAC3203 includes a stereo decimation filter for digital microphone inputs. The stereo
recording path can be powered up one channel at a time, to support the case where only mono record
capability is required.
The digital microphone input path of the TLV320DAC3203 features a large set of options for signal
conditioning as well as signal routing:
•Stereo decimation filters (PDM input)
•Fine gain adjustment of digital channels with 0.1 dB step size
•Digital volume control with a range of -12 to +20dB
•Mute function
In addition to the standard set of stereo decimation filter features the TLV320DAC3203 also offers the
following special functions:
•Channel-to-channel phase adjustment
•Adaptive filter mode
2.3.1 Digital Microphone Interface
The TLV320DAC3203 digital-microphone interface is shown in Figure 2-6.
Digital Microphone Input/Decimation Filter
The TLV320DAC3203 outputs internal clock DIG_MIC_CLK on GPIO pin ( Page 0, Register 51, D(5:2)) or
MISO pin (Page 0, Register 55, D(4:1)). This clock can be connected to the external digital microphone
Figure 2-6. Digital Microphone in TLV320DAC3203
device. The single-bit output of the external digital microphone device can be connected to GPIO, DIN or
SCLK pins. Internally the TLV320DAC3203 latches the steady value of data on the rising edge of
DIG_MIC_CLK for the Left ADC channel, and the steady value of data on falling edge for the Right ADC
channel.
Figure 2-7. Timing Diagram for Digital Microphone Interface
The digital-microphone mode can be selectively enabled for only-left, only-right, or stereo channels. The
AOSR value for the ADC channel must be configured to select the desired decimation ratio to be achieved
based on the external digital microphone properties.
2.3.2 Digital Volume Control
The TLV320DAC3203 also has a digital volume-control block with a range from -12dB to +20dB in steps
of 0.5dB. It is set by programming Page 0, Register 83 and 84 respectively for left and right channels.
Table 2-4. Digital Volume Control for ADC
Desired GainLeft / Right Channel
dBPage 1, Register 83/84,
2.3.3 Digital Microphone Decimation Filtering and Signal Processing Overview
22
During volume control changes, the soft-stepping feature is used to avoid audible artifacts. The softstepping rate can be set to either 1 or 2 gain steps per sample. Soft-stepping can also be entirely
disabled. This soft-stepping is configured via Page 1, Register 81, D(1:0), and is common to soft-stepping
control for the analog PGA. During power-down of an ADC channel, this volume control soft-steps down to
-12.0dB before powering down. Due to the soft-stepping control, soon after changing the volume control
setting or powering down the ADC channel, the actual applied gain may be different from the one
programmed through the control register. The TLV320DAC3203 gives feedback to the user, through readonly flags Page 1, Reg 36, D(7) for Left Channel and Page 1, Reg 36, D(3) for the right channel.
Additionally, the gains in each of the channels is finely adjustable in steps of 0.1dB. This is useful when
trying to match the gain between channels. By programming Page 0, Register 82 the gain can be adjusted
from 0dB to -0.4dB in steps of 0.1dB. This feature, in combination with the regular digital volume control
allows the gains through the left and right channels be matched in the range of -0.5dB to +0.5dB with a
resolution of 0.1dB.
The TLV320DAC3203 includes a built-in digital decimation filter to process the oversampled data from the
PDM input to generate digital data at Nyquist sampling rate with high dynamic range. The decimation filter
can be chosen from three different types, depending on the required frequency response, group delay and
sampling rate.
The TLV320DAC3203 offers a range of processing blocks which implement various signal processing
capabilities along with decimation filtering. These processing blocks give users the choice of how much
and what type of signal processing they may use and which decimation filter is applied.
Table 2-5 gives an overview of the available processing blocks and their properties.
The signal processing blocks available are:
•First-order IIR
•Scalable number of biquad filters
•Variable-tap FIR filter
The processing blocks are tuned for common cases and can achieve high anti-alias filtering or low-group
delay in combination with various signal processing effects such as audio effects and frequency shaping.
The available first order IIR, BiQuad and FIR filters have fully user-programmable coefficients. The
Resource Class Column (RC) gives an approximate indication of power consumption.
Depending on the selected processing block, different types and orders of digital filtering are available. A
1st-order IIR filter is always available, and is useful to efficiently filter out possible DC components of the
signal. Up to 5 biquad section or alternatively up to 25-tap FIR filters are available for specific processing
blocks. The coefficients of the available filters are arranged as sequentially indexed coefficients in two
banks. If adaptive filtering is chosen, the coefficient banks can be switched on-the-fly. For more details on
adaptive filtering see Section 2.3.3.2.3 below.
The coefficients of these filters are each 24-bits wide, in two's-complement and occupy 3 consecutive 8-bit
registers in the register space. For default values please see Section 5.17.
2.3.3.1.2.1 1stOrder IIR Section
The transfer function for the first order IIR Filter is given by
The frequency response for the 1storder IIR Section with default coefficients is flat at a gain of 0dB.
Details on coefficient default values are given in Section 5.17.
Table 2-6. First-Order IIR Filter Coefficients
Digital Microphone Input/Decimation Filter
(2)
FilterFIlterCoefficient Left ChannelCoefficient Right Channel
The transfer function of each of the Biquad Filters is given by
The frequency response for each of the biquad section with default coefficients is flat at a gain of 0dB.
Details on coefficient default values are given in Section 5.17.
www.ti.com
(3)
Table 2-7. Biquad Filter Coefficients
FilterFIlterCoefficient Left ChannelCoefficient Right Channel
Six of the available processing blocks offer FIR filters for signal processing. PRB_R9 and PRB_R12
feature a 20-tap FIR filter while the processing blocks PRB_R3, PRB_R6, PRB_R15 and PRB_R18
feature a 25-tap FIR filter
The coefficients of the FIR filters are 24-bit 2’s complement format and correspond to the coefficient space
as listed below. There is no default transfer function for the FIR filter. When the FIR filter gets used all
applicable coefficients must be programmed.
FilterFIlter Coefficient Left ChannelFilter Coefficient Right Channel
ADCChannelResponseforDecimationFilter A
(Redlinecorrespondsto –73dB)
G013
Digital Microphone Input/Decimation Filter
2.3.3.1.3 Decimation Filter
The TLV320DAC3203 offers 3 different types of decimation filters. The integrated digital decimation filter
removes high-frequency content and down samples the audio data from an initial sampling rate of
AOSR*Fs to the final output sampling rate of Fs. The decimation filtering is achieved using a higher-order
CIC filter followed by linear-phase FIR filters. The decimation filter cannot be chosen by itself, it is implicitly
set through the chosen processing block.
The following subsections describe the properties of the available filters A, B and C.
2.3.3.1.3.1 Decimation Filter A
This filter is intended for use at sampling rates up to 48kHz. When configuring this filter, the oversampling
ratio of the can either be 128 or 64. For highest performance the oversampling ratio must be set to 128.
Please also see the PowerTune chapter for details on performance and power in dependency of AOSR.
Filter A can also be used for 96kHz at an AOSR of 64.
ParameterConditionValue (Typical)Units
AOSR = 128
Filter Gain Pass Band0…0.39 Fs0.062dB
Filter Gain Stop Band0.55…64Fs–73dB
Filter Group Delay17/FsSec.
Pass Band Ripple, 8 ksps0…0.39 Fs0.062dB
Pass Band Ripple, 44.1 ksps0…0.39 Fs0.05dB
Pass Band Ripple, 48 ksps0…0.39 Fs0.05dB
AOSR = 64
Filter Gain Pass Band0…0.39 Fs0.062dB
Filter Gain Stop Band0.55…32Fs–73dB
Filter Group Delay17/FsSec.
Pass Band Ripple, 8 ksps0…0.39 Fs0.062dB
Pass Band Ripple, 44.1 ksps0…0.39 Fs0.05dB
Pass Band Ripple, 48 ksps0…0.39 Fs0.05dB
Pass Band Ripple, 96 ksps0…20kHz0.1dB
www.ti.com
Table 2-9. Decimation Filter A, Specification
30
TLV320DAC3203 ApplicationSLAU434–May 2012
Figure 2-17. Decimation Filter A, Frequency Response