• Decodes MPEG 1 & 2 audio layer III (CBR
+VBR +ABR); layers I & II optional;
MPEG4/2 AAC-LC-2.0.0.0 (+PNS);
WMA 4.0/4.1/7/8/9allprofiles (5-384 kbps);
WAV (PCM + IMA ADPCM);
General MIDI / SP-MIDI format 0 files
• Encodes IMA ADPCM from microphone
or line input
• Streaming support for MP3 and WAV
• EarSpeaker Spatial Processing
• Bass and treble controls
• Operates with a single clock 12..13 MHz
• Can also be used with 24..26 MHz clocks
• Internal PLL clock multiplier
• Low-power operation
• High-quality on-chip stereo DAC with no
phase error between channels
• Stereo earphone driver capable of driving a
30Ω load
• Quiet power-on and power-off
• I2S interface for external DAC
• Separate operating voltages for analog,
digital and I/O
• 5.5 KiB On-chip RAM for user code / data
• Serial control and data interfaces
• Can be used as a slave co-processor
• SPI flash boot for special applications
• UART for debugging purposes
• New functions may be added with software
and 8 GPIO pins
• Lead-free RoHS-compliantpackage(Green)
Description
VS1033 is a single-chip MP3/AAC/WMA/MIDI
audio decoder and ADPCM encoder. It contains
a high-performance, proprietary low-power DSP
processor core VS DSP4, working data memory,
5 KiB instruction RAM and 0.5 KiB data RAM
for user applications, serial control and input data
interfaces, upto 8 general purpose I/O pins, an
UART, as well as a high-quality variable-samplerate mono ADC and stereo DAC, followed by an
earphone amplifier and a common voltage buffer.
VS1033 receives its input bitstream through a serial input bus, which itlistenstoasasystemslave.
The input stream is decoded and passed through a
digital volume control to an 18-bit oversampling,
multi-bit, sigma-delta DAC. The decoding is controlled via a serial control bus. In addition to the
basic decoding, it is possible to add application
specific features,likeDSP effects, to theuserRAM
memory.
EarSpeaker spatial processing provides more natural sound in headphone listening conditions. It
widens the stereo image and positions the sound
sources outside the listener’s head.
MPEG Layer-3 audio decoding technology licensed from Fraunhofer IIS and Thomson.
Note: if you enable Layer I and Layer II decoding, you are liable for any patent issues that may
arise from using these formats. Joint licensing of MPEG 1.0 / 2.0 Layer III does not cover all patents
pertaining to layers I and II.
VS1033 contains WMA decoding technology from Microsoft.
This product is protected by certain intellectual property rights of Microsoft and cannot be used
or further distributed without a license from Microsoft.
VS1033 contains AAC technology (ISO/IEC 13818-7) which cannot be used without a proper license
from Via Licensing Corporation or individual patent holders.
To the best of our knowledge, if the end product does not play a specific format that otherwise would
require a customer license: MPEG 1.0/2.0 layers I and II, WMA, or AAC, the respective license should
not be required. Decoding of MPEG layers I and II are disabled by default, and WMA and AAC format
exclusion can be easily performed based on the contents of the SCI HDAT1 register.
2Disclaimer
This is a
preliminary
datasheet. All properties and figures are subject to change.
3Definitions
B Byte, 8 bits.
b Bit.
Ki “Kibi” = 210= 1024 (IEC 60027-2).
Mi “Mebi” = 220= 1048576 (IEC 60027-2).
VS DSP VLSI Solution’s DSP core.
W Word. In VS DSP, instruction words are 32-bit and data words are 16-bit wide.
Version 0.91, 2007-02-129
VLSI
Solution
y
VS1033c PRELIMINARY
4. CHARACTERISTICS&SPECIFICATIONS
4Characteristics & Specifications
4.1Absolute Maximum Ratings
ParameterSymbolMinMaxUnit
Analog Positive SupplyAVDD-0.33.6V
Digital Positive SupplyCVDD-0.32.7V
I/O Positive SupplyIOVDD-0.33.6V
Current at Any Digital Output±50mA
Voltage at Any Digital Input-0.3IOVDD+0.31V
Operating Temperature-40+85
Storage Temperature-65+150
1
Must not exceed 3.6 V
◦
◦
VS1033C
C
C
4.2Recommended Operating Conditions
ParameterSymbolMinTypMaxUnit
Ambient Operating Temperature-40+85◦C
Analog and Digital Ground
Positive AnalogAVDD2.52.83.6V
Positive DigitalCVDD2.42.52.7V
I/O VoltageIOVDDCVDD-0.6V2.83.6V
Input Clock Frequency
Internal Clock FrequencyCLKI1236.86455.3MHz
Internal Clock Multiplier
Master Clock Duty Cycle405060%
1
Must be connected together as close the device as possible for latch-up immunity.
2
The maximum sample rate that can be played with correct speed is XTALI/256.
Thus, XTALI must be at least 12.288 MHz to be able to play 48 kHz at correct speed.
3
Reset value is 1.0×. Recommended SC MULT=3.0×, SC ADD=1.0× (SCI CLOCKF=0x9000).
1
2
3
AGND DGND0.0V
XTALI1212.28813MHz
1.0×3.0×4.5×
Version 0.91, 2007-02-1210
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
4. CHARACTERISTICS&SPECIFICATIONS
4.3Analog Characteristics
Unless otherwisenoted: AVDD=2.5..2.85V, CVDD=2.4..2.7V, IOVDD=CVDD-0.6V..3.6V,TA=-40..+85◦C,
XTALI=12..13MHz, Internal Clock Multiplier 3.5×. DAC tested with 1307.894 Hz full-scale output
sinewave, measurement bandwidth 20..20000 Hz, analog output load: LEFT to GBUF 30Ω, RIGHT to
GBUF 30Ω. Microphone test amplitude 50 mVpp, fs=1 kHz, Line input test amplitude 1.1 V, fs=1 kHz.
ParameterSymbolMinTypMaxUnit
DAC Resolution18bits
Total Harmonic DistortionTHD0.10.3%
Dynamic Range (DAC unmuted, A-weighted)IDR90dB
S/N Ratio (full scale signal)SNR70dB
Interchannel Isolation (Cross Talk)5075dB
Interchannel Isolation (Cross Talk), with GBUF40dB
Interchannel Gain Mismatch-0.50.5dB
Frequency Response-0.10.1dB
Full Scale Output Voltage (Peak-to-peak)1.31.5
Deviation from Linear Phase5
Analog Output Load ResistanceAOLR1630
Analog Output Load Capacitance100pF
Microphone input amplifier gainMICG26dB
Microphone input amplitude501403mVpp AC
Microphone Total Harmonic DistortionMTHD0.020.10%
Microphone S/N RatioMSNR5062dB
Line input amplitude220028003mVpp AC
Line input Total Harmonic DistortionLTHD0.060.10%
Line input S/N RatioLSNR6068dB
Line and Microphone input impedances100kΩ
1
2
1.7Vpp
◦
Ω
1
3.0 volts can be achieved with +-to-+ wiring for mono difference sound.
2
AOLR may be much lower, but below Typical distortion performance may be compromised.
3
Above typical amplitude the Harmonic Distortion increases.
4.4Power Consumption
Tested with an MPEG 1.0 Layer-3 128 kbps sample and generated sine. Output at full volume. Internal
clock multiplier 3.0×.
ParameterMinTypMaxUnit
Power Supply Consumption AVDD, Reset0.65.0µA
Power Supply Consumption CVDD = 2.5V, Reset3.750.0µA
Power Supply Consumption AVDD, sine test, 30Ω + GBUF36.9mA
Power Supply Consumption CVDD = 2.5V, sine test8.2mA
Power Supply Consumption AVDD, no load7.0mA
Power Supply Consumption AVDD, output load 30Ω10.9mA
Power Supply Consumption AVDD, 30Ω + GBUF16.1mA
Power Supply Consumption CVDD = 2.5V14mA
Version 0.91, 2007-02-1211
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
4. CHARACTERISTICS&SPECIFICATIONS
4.5Digital Characteristics
ParameterSymbolMinTypMaxUnit
High-Level Input Voltage0.7×IOVDDIOVDD+0.31V
Low-Level Input Voltage-0.20.3×IOVDDV
High-Level Output Voltage at IO= -2.0 mA0.7×IOVDDV
Low-Level Output Voltage at IO= 2.0 mA0.3×IOVDDV
Input Leakage Current-1.01.0µA
SPI Input Clock Frequency
XRESET active time2XTALI
XRESET inactive to software ready20000500001XTALI
Power on reset, rise time to CVDD10V/s
1
DREQ rises when initialization is complete. You should not send any data or commands before that.
Version 0.91, 2007-02-1212
VLSI
Solution
y
VS1033C
1
48
A
B
C
D
E
F
G
1 2
3
4 5
6 7
TOP VIEW
0.80 TYP
4.80
7.00
1.10 REF
0.80 TYP
1.10 REF
4.80
7.00
A1 BALL PAD CORNER
VS1033c PRELIMINARY
5. PACKAGES AND PIN DESCRIPTIONS
5Packages and Pin Descriptions
5.1Packages
Both LPQFP-48 and BGA-49 are lead (Pb) free and also RoHS compliant packages. RoHS is a short
name of Directive 2002/95/EC on the restriction of the use of certain hazardous substances in electricaland electronic equipment.
5.1.1LQFP-48
Figure 1: Pin Configuration, LQFP-48.
LQFP-48 package dimensions are at http://www.vlsi.fi/ .
5.1.2BGA-49
BGA-49 package dimensions are at http://www.vlsi.fi/ .
XDCS / BSYNC
IOVDD114F3IOPWRI/O power supply
VCO15G2DOFor testing only (Clock VCO output)
DGND116F4DGNDCore & I/O ground
XTALO17G3AOCrystal output
XTALI18E4AICrystal input
IOVDD219G4IOPWRI/O power supply
IOVDD3F5IOPWRI/O power supply
DGND220DGNDCore & I/O ground
DGND321G5DGNDCore & I/O ground
DGND422F6DGNDCore & I/O ground
XCS23G6DIChip select input (active low)
CVDD224G7CPWRCore power supply
GPIO5 / I2S MCLK
RX26E6DIUART receive, connect to IOVDD if not used
TX27F7DOUART transmit
SCLK28D6DIClock for serial bus
SI29E7DISerial input
SO30D5DO3Serial output
CVDD331D7CPWRCore power supply
TEST32C6DIReserved for test, connect to IOVDD
GPIO0 / I2S SCLK
GPIO4 / I2S LROUT336A7DIOGeneral purpose IO 4 / I2S LROUT
AGND037C5APWRAnalog ground, low-noise reference
AVDD038B5APWRAnalog power supply
RIGHT39A6AORight channel output
AGND140B4APWRAnalog ground
AGND241A5APWRAnalog ground
GBUF42C4AOCommon buffer for headphones, do NOT connect to
AVDD143A4APWRAnalog power supply
RCAP44B3AIOFiltering capacitance for reference
AVDD245A3APWRAnalog power supply
LEFT46B2AOLeft channel output
AGND347A2APWRAnalog ground
LINEIN48A1AILine input
1
1
1
9E1DIOGeneral purpose IO 2 / serial input data bus clock
10F2DIOGeneral purpose IO 3 / serial data input
13E3DIData chip select / byte sync
3
25E5DIOGeneral purpose IO 5 / I2S MCLK
3
33C7DIOGeneral purpose IO 0 (SPIBOOT) / I2S SCLK
BGA
Ball
Pin
Type
Function
use 100 kΩ pull-down resistor
ground!
2
Version 0.91, 2007-02-1214
VLSI
Solution
y
VS1033c PRELIMINARY
1
First pin function is active in New Mode, latter in Compatibility Mode.
2
Unless pull-down resistor is used, SPI Boot is tried. See Chapter 9.5 for details.
3
If I2S CF ENA is ’0’ the pins are used for GPIO. See Chapter 10.13 for details.
5. PACKAGES AND PIN DESCRIPTIONS
Pin types:
VS1033C
TypeDescription
DIDigital input, CMOS Input Pad
DODigital output, CMOS Input Pad
DIODigital input/output
DO3Digital output, CMOS Tri-stated OutputPad
AIAnalog input
In BGA-49, D4 is a no-connect ball.
TypeDescription
AOAnalog output
AIOAnalog input/output
APWRAnalog power supply pin
DGNDCore or I/O ground pin
CPWRCore power supply pin
IOPWRI/O power supply pin
Version 0.91, 2007-02-1215
VLSI
Solution
y
VS1033c PRELIMINARY
6Connection Diagram, LQFP-48
VS1033C
6. CONNECTION DIAGRAM, LQFP-48
Figure 3: Typical Connection Diagram Using LQFP-48.
The common buffer GBUF can be used for common voltage (1.24 V) for earphones. This will eliminate
the need for large isolation capacitors on line outputs, and thus the audio output pins from VS1033 may
be connected directly to the earphone connector.
GBUF must NOT be connected to ground in any circumstance. If GBUF is not used, LEFT and RIGHT
must be provided with coupling capacitors. See application notes for details.
Unused GPIO pins should have a pull-down resistor.
If UART is not used, RX should be connected to IOVDD and TX be unconnected.
Do not connect any external load to XTALO.
Note: This connection assumes SM SDINEW is active (see Chapter 8.7.1). If also SM SDISHARE is
used, xDCS should be tied low or high (see Chapter 7.2.1).
Version 0.91, 2007-02-1216
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
7. SPI BUSES
7SPI Buses
7.1General
The SPI Bus - that was originally used in some Motorola devices - has been used for both VS1033’s
Serial Data Interface SDI (Chapters 7.4 and 8.5) and Serial Control Interface SCI (Chapters 7.5 and 8.6).
7.2SPI Bus Pin Descriptions
7.2.1VS1002 Native Modes (New Mode)
These modes are active on VS1033 when SM SDINEW is set to 1 (default at startup). DCLK and
SDATA are not used for data transfer and they can be used as general-purpose I/O pins (GPIO2 and
GPIO3). BSYNC function changes to data interface chip select (XDCS).
SDI PinSCI PinDescription
XDCSXCSActive low chip select input. A high level forces the serial interface into
standby mode, ending the current operation. A high level also forces serial
output (SO) to high impedance state. If SM SDISHARE is 1, pin
XDCS is not used, but the signal is generated internally by inverting
XCS.
SCKSerial clock input. The serial clock is also used internally as the master
clock for the register interface.
SCK can be gated or continuous. In either case, the first rising clock edge
after XCS has gone low marks the first bit to be written.
SISerial input. If a chip select is active, SI is sampled on the rising CLK edge.
-SOSerial output. In reads, data is shifted out on the falling SCK edge.
In writes SO is at a high impedance state.
7.2.2VS1001 Compatibility Mode
This mode is active when SM SDINEW is set to 0. In this mode, DCLK, SDATA and BSYNC are active.
SDI PinSCI PinDescription
-XCSActive low chip select input. A high level forces the serial interface into
standby mode, ending the current operation. A high level also forces serial
output (SO) to high impedance state.
BSYNC-SDI data is synchronized with a rising edge of BSYNC.
DCLKSCKSerial clock input. The serial clock is also used internally as the master
SDATASISerial input. SI is sampled on the rising SCK edge, if XCS is low.
-SOSerial output. In reads, data is shifted out on the falling SCK edge.
clock for the register interface.
SCK can be gated or continuous. In either case, the first rising clock edge
after XCS has gone low marks the first bit to be written.
In writes SO is at a high impedance state.
Version 0.91, 2007-02-1217
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
7. SPI BUSES
7.3Data Request Pin DREQ
The DREQ pin/signal is used to signal if VS1033’s 2048-byte FIFO is capable of receiving data. If
DREQ is high, VS1033 can take at least 32 bytes of SDI data or one SCI command. DREQ is turned low
when the stream buffer is too full and for the duration of a SCI command.
Because of the 32-byte safety area, the sender may send upto 32 bytes of SDI data at a time without
checking the status of DREQ, making controlling VS1033 easier for low-speed microcontrollers.
Note: DREQ may turn low or high at any time, even during a byte transmission. Thus, DREQ should
only be used to decide whether to send more bytes. It does not need to abort a transmission that has
already started.
Note: In VS10XX products upto VS1002, DREQ was only used for SDI. In VS1003 and VS1033 DREQ
is also used to tell the status of SCI.
There are cases when you still want to send SCI commands when DREQ is low. Because DREQ is
shared between SDI and SCI, you can not determine if a SCI command has been executed if SDI is not
ready to receive. In this case you need a long enough delay after every SCI command to make certain
none of them is missed. The SCI Registers table in section 8.7 gives the worst-case handling time for
each SCI register write.
7.4Serial Protocol for Serial Data Interface (SDI)
7.4.1General
The serial data interface operates in slavemodesoDCLK signal must be generated by an externalcircuit.
Data (SDATA signal) can be clocked in at either the rising or falling edge of DCLK (Chapter 8.7).
VS1033 assumes its data input to be byte-sychronized. SDI bytes may be transmitted either MSb or LSb
first, depending of contents of SCI MODE (Chapter 8.7.1).
The firmware is able to accept the maximum bitrate the SDI supports.
7.4.2SDI in VS1002 Native Modes (New Mode)
In VS1002native modes(SM NEWMODE is 1), byte synchronizationisachieved by XDCS.Thestateof
XDCS may not change while a data byte transfer is in progress. To always maintain data synchronization
even if there may be glitches in the boards using VS1033, it is recommended to turn XDCS every now
and then, for instance once after every flash data block or a few kilobytes, just to keep sure the host and
VS1033 are in sync.
If SM SDISHARE is 1, the XDCS signal is internally generated by inverting the XCS input.
For new designs, using VS1002 native modes are recommended.
Version 0.91, 2007-02-1218
VLSI
Solution
y
VS1033C
BSYNC
SDATA
DCLK
D7D6D5D4D3D2D1D0
BSYNC
SDATA
DCLK
D7 D6D5 D4D3 D2 D1D0
D7 D6D5 D4D3 D2 D1D0
VS1033c PRELIMINARY
7. SPI BUSES
7.4.3SDI in VS1001 Compatibility Mode
Figure 4: BSYNC Signal - one byte transfer.
When VS1033 is running in VS1001 compatibility mode, a BSYNC signal must be generated to ensure
correct bit-alignment of the input bitstream. The first DCLK sampling edge (rising or falling, depending
on selected polarity), during which the BSYNC is high, marks the first bit of a byte (LSB, if LSB-first
order is used, MSB, if MSB-first orderisused). If BSYNC is ’1’ when the last bit is received,thereceiver
stays active and next 8 bits are also received.
7.4.4Passive SDI Mode
If SM NEWMODE is 0 and SM SDISHARE is 1, the operation is otherwise like the VS1001 compatibility mode, but bits are only received while the BSYNC signal is ’1’. Rising edge of BSYNC is still
used for synchronization.
7.5Serial Protocol for Serial Command Interface (SCI)
7.5.1General
The serial bus protocol for the Serial Command Interface SCI (Chapter 8.6) consists of an instruction
byte, address byte and one 16-bit data word. Each read or write operation can read or write a single
register. Data bits are read at the rising edge, so the user should update data at the falling edge. Bytes
are always send MSb first. XCS should be low for the full duration of the operation, but you can have
pauses between bits if needed.
The operation is specified by an 8-bit instruction opcode. The supported instructions are read and write.
See table below.
Figure 5: BSYNC Signal - two byte transfer.
Instruction
NameOpcodeOperation
READ0b0000 0011Read data
WRITE0b0000 0010Write data
Note: VS1033 sets DREQ low after each SCI operation. The duration depends on the operation. It is not
allowed to start a new SCI/SDI operation before DREQ is high again.
Version 0.91, 2007-02-1219
VLSI
Solution
y
7.5.2SCI Read
0 1 2 3 4 5 6 7 8 9 10 11 12 1330 3114 15 16 17
0 0 0 0 0 0 1 1 0 0 0 0
3 2 1 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
15 141 0
X
instruction (read)address
data out
XCS
SCK
SI
SO
don’t caredon’t care
DREQ
execution
0 1 2 3 4 5 6 7 8 9 10 11 12 1330 3114 15 16 17
0 0 0 0 0 0 10 0 0 0
3 2 1 0
1 0
X
address
XCS
SCK
SI
15 14
data out
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0SO
0 00 0
X
0
instruction (write)
DREQ
execution
VS1033c PRELIMINARY
Figure 6: SCI Word Read
VS1033C
7. SPI BUSES
VS1033 registers are read from using the following sequence, as shown in Figure 6. First, XCS line is
pulled low to select the device. Then the READ opcode (0x3) is transmitted via the SI line followed by
an 8-bit word address. After the address has been read in, any further data on SI is ignored by the chip.
The 16-bit data corresponding to the received address will be shifted out onto the SO line.
XCS should be driven high after data has been shifted out.
DREQ is driven low for a short while when in a read operation by the chip. This is a very short time and
doesn’t require special user attention.
7.5.3SCI Write
Version 0.91, 2007-02-1220
Figure 7: SCI Word Write
VS1033 registers are written from using the following sequence, as shown in Figure 7. First, XCS line
is pulled low to select the device. Then the WRITE opcode (0x2) is transmitted via the SI line followed
by an 8-bit word address.
VLSI
Solution
y
VS1033C
XCS
SCK
SI
SO
01151416
tXCSS
tXCSH
tWL tWH
tH
tSU
tV
tZ
tDIS
tXCS
30
31
VS1033c PRELIMINARY
7. SPI BUSES
After the word has been shifted in and the last clock has been sent, XCS should be pulled high to end the
WRITE sequence.
After the last bit has been sent, DREQ is driven low for the duration of the register update, marked “execution” in the figure. The time varies depending on the register and its contents (see table in Chapter 8.7
for details). If the maximum time is longer than what it takes from the microcontroller to feed the next
SCI command or SDI byte, it is not allowed to finish a new SCI/SDI operation before DREQ has risen
up again.
25ns is when pin loaded with 100pF capacitance. The time is shorter with lower capacitance.
Note: As tWL and tWH, as well as tH require at least 2 clock cycles, the maximum speed for the SPI
bus that can easily be used is 1/6 of VS1033’s internal clock speed CLKI. Slightly higher speed can be
achieved with very careful timing tuning. For details, see Application Notes for VS10XX.
Note: Although the timing is derived from the internal clock CLKI, the system always starts up in 1.0×
mode, thus CLKI=XTALI.
Note: Negative numbers mean that the signal can change in different order from what is shown in the
diagram.
Version 0.91, 2007-02-1221
VLSI
Solution
y
VS1033c PRELIMINARY
0
1233031
1010
000000
XX
XCS
SCK
SI
2
3233616263
SCI Write 1
SCI Write 2
DREQ
DREQ up before finishing next SCI write
123
XCS
SCK
SI
7654310765210
X
SDI Byte 1
SDI Byte 2
06789131415
DREQ
7.7SPI Examples with SM SDINEW and SM SDISHARED set
7.7.1Two SCI Writes
VS1033C
7. SPI BUSES
Figure 9: Two SCI Operations.
Figure 9 shows two consecutive SCI operations. Note that xCS must be raised to inactive state between
the writes. Also DREQ must be respected as shown in the figure.
7.7.2Two SDI Bytes
Figure 10: Two SDI Bytes.
SDI data is synchronized with a raising edge of xCS as shown in Figure 10. However, every byte doesn’t
need separate synchronization.
Version 0.91, 2007-02-1222
VLSI
Solution
y
VS1033C
0
1
XCS
SCK
SI
7
7651
00
076510
SDI Byte
SCI Operation
SDI Byte
893940414647
X
DREQ high before end of next transfer
VS1033c PRELIMINARY
7. SPI BUSES
7.7.3SCI Operation in Middle of Two SDI Bytes
Figure 11: Two SDI Bytes Separated By an SCI Operation.
Figure 11 shows how an SCI operation is embedded in between SDI operations. xCS edges are used to
synchronize both SDI and SCI. Remember to respect DREQ as shown in the figure.
Version 0.91, 2007-02-1223
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
8. FUNCTIONAL DESCRIPTION
8Functional Description
8.1Main Features
VS1033 is based on a proprietary digital signal processor, VS DSP. It contains all the code and data
memory needed for MP3, AAC, WMA and WAV PCM + ADPCM audio decoding, MIDI synthesizer,
together with serial interfaces, a multirate stereo audio DAC and analog output amplifiers and filters.
Also ADPCM audio encoding is supported using a microphone amplifier and A/D converter. A UART
is provided for debugging purposes.
8.2Supported Audio Codecs
Conventions
MarkDescription
+Format is supported
-Format exists but is not supported
Format doesn’t exist
VS1033 decodes MPEG2-AAC-LC-2.0.0.0 and MPEG4-AAC-LC-2.0.0.0 streams. This means that the
low complexity profile with maximum of two channels can be decoded. If a stream contains more than
one element and/or element type, you can select which one to decode from the 16 single-channel, 16
channel-pair, and 16 low-frequency elements. The default is to select the first one that appears in the
stream.
Dynamic range control (DRC) is supported and can be controlled by the user to limit or enhance the
dynamic range of the material that has DRC information.
Both Sine window and Kaiser-Bessel-derived window are supported.
For MPEG4 pseudo-random noise substitution (PNS) is supported. Short frames (120 and 960 samples)
are not supported.
For AAC the streaming ADTS format is recommended. This format allows easy rewind and fast forward
because resynchronization is easily possible.
In addition to ADTS (.aac), MPEG2 ADIF (.aac) and MPEG4 AUDIO (.mp4 / .m4a) files are played,
but these formats are less suitable for rewind and fast forward operations. You can still implement these
features by using the safe jump points table and seek mechanism provided, or using slightly less robust
but much easier automatic resync mechanism (see Section 9.9).
Note: To be able to play the .mp4 and .m4a files, the mdat chunk must be the last chunk in the file.
64000 Hz, 88200 Hz, and 96000 Hz AAC files are played but with wrong speed.
2
Version 0.91, 2007-02-1226
Also all variable bitrate (VBR) formats are supported. Note that the table gives the maximum bitrate
allowed for two channels for a specific sample rate as defined by the AAC specification. The decoder
does not actually have a lower or upper limit.
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
8. FUNCTIONAL DESCRIPTION
8.2.5Supported WMA Formats
Windows Media Audio codec versions 2, 7, 8, and 9 are supported. All WMA profiles (L1, L2, and L3)
are supported. Previously streams were separated into Classes 1, 2a, 2b, and 3. The decoder has passed
Microsoft’s conformance testing program.
In addition to these expected WMA decoding profiles, all other bitrate and samplerate combinations are
supported, including variable bitrate WMA streams. Note that WMA does not consume the bitstream as
evenly as MP3, so you need a higher peak transfer capability for clean playback at the same bitrate.
Version 0.91, 2007-02-1227
VLSI
Solution
y
VS1033c PRELIMINARY
8. FUNCTIONAL DESCRIPTION
8.2.6Supported RIFF WAV Formats
The most common RIFF WAV subformats are supported.
General MIDI and SP-MIDI format 0 files are played. Format 1 and 2 files must be converted to format
0 by the user. The maximum simultaneous polyphony is 40. Actual polyphony depends on the internal
clock rate (which is user-selectable), the instruments used, whether the reverb effect is enabled, and the
possible global postprocessing effects enabled, such as bass and treble enhancers or EarSpeaker spatial
processing. The polyphony restriction algorithm makes use of the SP-MIDI MIP table, if present.
36.86 MHz (3.0× input clock) achieves 16-26 simultaneous sustained notes. The instantaneous amount
of notes can be larger. 36 MHz is a fair compromise between power consumption and quality, but higher
clocks can be used to increase the polyphony.
Reverb effect can be controlled by the user. In addition to reverb automatic and reverb off modes, 14
different decay times can be selected. These roughly correspond to different room sizes. Also, each
midi song decides how much effect each instrument gets. Because the reverb effect uses about 4 MHz of
processing power the automatic control enables reverb only when the internal clock is at least 3.0×.
8. FUNCTIONAL DESCRIPTION
When EarSpeaker spatial processing is active, MIDI reverb is not used.
New instruments have been implemented in addition to the 36 that are available in VS1003. VS1033c
now has unique instruments in the whole GM1 instrument set and one bank of GM2 percussions.
Version 0.91, 2007-02-1229
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
VS1033c Melodic Instruments (GM1)
1 Acoustic Grand Piano33 Acoustic Bass65 Soprano Sax97 Rain (FX 1)
2 Bright Acoustic Piano34 Electric Bass (finger)66 Alto Sax98 Sound Track (FX 2)
3 Electric Grand Piano35 Electric Bass (pick)67 Tenor Sax99 Crystal (FX 3)
4 Honky-tonk Piano36 Fretless Bass68 Baritone Sax100 Atmosphere (FX 4)
5 Electric Piano 137 Slap Bass 169 Oboe101 Brightness (FX 5)
6 Electric Piano 238 Slap Bass 270 English Horn102 Goblins (FX 6)
7 Harpsichord39 Synth Bass 171 Bassoon103 Echoes (FX 7)
8 Clavi40 Synth Bass 272 Clarinet104 Sci-fi (FX 8)
27 High Q43 High Floor Tom59 Ride Cymbal 275 Claves
28 Slap44 Pedal Hi-hat [EXC1]60 High Bongo76 Hi Wood Block
29 Scratch Push [EXC 7]45 Low Tom61 Low Bongo77 Low Wood Block
30 Scratch Pull [EXC 7]46 Open Hi-hat[EXC1]62 Mute Hi Conga78 Mute Cuica [EXC 4]
31 Sticks47 Low-Mid Tom63 Open Hi Conga79 Open Cuica [EXC 4]
32 Square Click48 High Mid Tom64 Low Conga80 Mute Triangle[EXC 5]
33 Metronome Click49 Crash Cymbal 165 High Timbale81 Open Triangle [EXC5]
34 Metronome Bell50 High Tom66 Low Timbale82 Shaker
35 Acoustic Bass Drum51 Ride Cymbal 167 High Agogo83 Jingle bell
Version 0.91, 2007-02-1230
36 Bass Drum 152 Chinese Cymbal68 Low Agogo84 Bell tree
37 Side Stick53 Ride Bell69 Cabasa85 Castanets
38 Acoustic Snare54 Tambourine70 Maracas86 Mute Surdo [EXC 6]
39 Hand Clap55 Splash Cymbal71 Short Whistle [EXC2]87 Open Surdo [EXC 6]
40 Electric Snare56 Cowbell72 Long Whistle [EXC 2]
41 Low Floor Tom57 Crash Cymbal 273 Short Guiro [EXC 3]
42 Closed Hi-hat [EXC1]58 Vibra-slap74 Long Guiro [EXC 3]
VLSI
Solution
y
VS1033c PRELIMINARY
Volume
control
Audio
FIFO
S.rate.conv.
and DAC
R
Bitstream
FIFO
SDI
L
SCI_VOL
SM_ADPCM=0
2048 stereo
samples
Bass
enhancer
SB_AMPLITUDE=0
SB_AMPLITUDE!=0
AIADDR = 0
AIADDR != 0
User
Application
MP3
WAV/ADPCM/
WMA / AAC /
MIDI decode
ST_AMPLITUDE=0
ST_AMPLITUDE!=0
Treble
enhancer
Ear
Speaker
8.3Data Flow of VS1033
VS1033C
8. FUNCTIONAL DESCRIPTION
Figure 12: Data Flow of VS1033.
First, depending on the audio data, and provided ADPCM encoding mode is not set, MP3, WMA, AAC,
PCM WAV, IMA ADPCM WAV, or MIDI data is received and decoded from the SDI bus.
After decoding, if SCI AIADDR is non-zero, application code is executed from the address pointed to
by that register. For more details, see Application Notes for VS10XX.
Then data may be sent to the Bass and Treble Enhancer depending on the SCI BASS register.
Next, headphone processing is performed, if the EarSpeaker spatial processing is active.
After that the signal is fed to the volume control unit, which also copies the data to the Audio FIFO.
The Audio FIFO holds the data, which is read by the Audio interrupt (Chapter 10.14.1) and fed to the
sample rate converter and DACs. The size of the audio FIFO is 2048 stereo (2×16-bit) samples, or 8
KiB.
The sample rate converter upsamples all different sample rates to XTALI/2, or 128 times the highest usable sample rate with 18-bit precision. This removes the need for complex PLL-based clocking schemes
and allows almost unlimited sample rate accuracy with one fixed input clock frequency. With a 12.288
MHz clock, the DA converter operates at 128 × 48 kHz, i.e. 6.144 MHz, and creates a stereo in-phase
analog signal. The oversampled output is low-pass filtered by an on-chip analog filter. This signal is then
forwarded to the earphone amplifier.
Version 0.91, 2007-02-1231
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
8. FUNCTIONAL DESCRIPTION
8.4EarSpeaker Spatial Processing
While listening to the headphones the sound has a tendency to be localized inside the head. The sound
field becomes flat and lacking the sensation of dimensions. This is unnatural, awkward and sometimes
even disturbing situation. This phenomenon is often referred in literature as ‘lateralization’, meaning
’in-the-head’ localization. Long-term listening to lateralized sound may lead to listening fatigue.
All real-life sound sources are external, leaving traces to the acoustic wavefront that arrives to the ear
drums. From these traces, the auditory system in the brain is able to judge the distance and angle of each
sound source. In loudspeaker listening the sound is external and these traces are available. In headphone
listening these traces are missing or ambiguous.
The EarSpeaker processing makes listening via headphones more like listening the same music from
real loudspeakers or live music. Once the EarSpeaker processing is activated, the instruments are moved
from inside to the outside of the head, making it easier to separate the different instruments (see figure
13). The listening experience becomes more natural and pleasant, and the stereo image is sharper as the
instruments are widely on front of the listener instead of being inside the head.
Figure 13: EarSpeaker externalized sound sources vs. normal inside-the-head sound
Note that EarSpeaker differs from any common spatial processing effects, such as echo, reverb, or bass
boost. EarSpeaker simulates accurately human auditory model and real listening environment acoustics.
Thus is does not change the tonal character of the music by introducing artificial effects.
EarSpeaker processing can be parameterized to a few different modes, each simulating a little different
type of acoustical situation and suiting for different personal preference and type of recording. See
section 8.7.1 for how to activate different modes.
Version 0.91, 2007-02-1232
• Off: Best option when listening through loudspeakers or if the audio to be played contains binaural
preprocessing
• minimal: Suits well for listening to normal musical scores with headphones, very subtle
• normal: Suits well for listening to normal musical scores with headphones, moves sound source
farther than minimal
• extreme: Suits well for old or ’dry’ recordings, or if the audio to be played is artificial, for example
generated MIDI
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
8. FUNCTIONAL DESCRIPTION
8.5Serial Data Interface (SDI)
The serial data interface is meant for transferring compressed MP3, WMA, or AAC data, WAV PCM and
ADPCM data as well as MIDI data.
If the input of the decoder is invalid or it is not received fast enough, analog outputs are automatically
muted.
Also several different tests may be activated through SDI as described in Chapter 9.
Version 0.91, 2007-02-1233
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
8. FUNCTIONAL DESCRIPTION
8.6Serial Control Interface (SCI)
The serial control interface is compatible with the SPI bus specification. Data transfers are always 16
bits. VS1033 is controlled by writing and reading the registers of the interface.
The main controls of the control interface are:
• control of the operation mode, clock, and builtin effects
• access to status information and header data
• access to encoded digital data
• uploading user programs
8.7SCI Registers
VS1033 sets DREQ low when it detects an SCI operation and restores it when it has processed the
operation. The duration depends on the operation. If DREQ is low when an SCI operation is performed,
it also stays low after SCI operation processing.
If DREQ is high before a SCI operation, do not start a new SCI/SDI operation before DREQ is high
again. If DREQ is low before a SCI operation because the SDI can not accept more data, make certain
there is enough time to complete the operation before sending another.
SCI registers, prefix SCI
RegTypeResetTime1Abbrev[bits]Description
0x0rw0x80070 CLKI4MODEMode control
0x1rw0x0C
3
40 CLKISTATUSStatus of VS1033
0x2rw02100 CLKIBASSBuilt-in bass/treble enhancer
0x3rw011000 XTALI5CLOCKFClock freq + multiplier
0x4rw040 CLKIDECODE TIMEDecode time in seconds
0x5rw03200 CLKIAUDATAMisc. audio data
0x6rw080 CLKIWRAMRAM write/read
0x7rw080 CLKIWRAMADDRBase address for RAM write/read
0x8r0-HDAT0Stream header data 0
0x9r0-HDAT1Stream header data 1
0xArw03200 CLKI2AIADDRStart address of application
0xBrw02100 CLKIVOLVolume control
0xCrw050 CLKI2AICTRL0Application control register 0
0xDrw050 CLKI2AICTRL1Application control register 1
0xErw050 CLKI2AICTRL2Application control register 2
0xFrw050 CLKI2AICTRL3Application control register 3
1
This is the worst-case time that DREQ stays low after writing to this register. The user may choose to
skip the DREQ check for those register writes that take less than 100 clock cycles to execute.
2
In addition, the cycles spent in the user application routine must be counted.
3
Firmware changes the value of this register immediately to 0x58, and in less than 100 ms to 0x50.
4
When mode register write specifies a software reset the worst-case time is 20000 XTALI cycles.
5
Writing to this register may force internal clock to run at 1.0 × XTALI for a while. Thus it is not a
good idea to send SCI or SDI bits while this register update is in progress.
Version 0.91, 2007-02-1234
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
8. FUNCTIONAL DESCRIPTION
8.7.1SCI MODE (RW)
SCI MODE is used to control the operation of VS1033 and defaults to 0x0800 (SM SDINEW set).
BitNameFunctionValueDescription
0SM DIFFDifferential0normal in-phase audio
1left channel inverted
1SM LAYER12Allow MPEG layers I & II0no
1yes
2SM RESETSoft reset0no reset
1reset
3SM OUTOFWAVJump out of WAV decoding0no
1yes
4SM EARSPEAKER LOEarSpeaker low setting0off
1active
5SM TESTSAllow SDI tests0not allowed
1allowed
6SM STREAMStream mode0no
1yes
7SM EARSPEAKER HIEarSpeaker high setting0off
1active
8SM DACTDCLK active edge0rising
1falling
9SM SDIORDSDI bit order0MSb first
1MSb last
10SM SDISHAREShare SPI chip select0no
1yes
11SM SDINEWVS1002 native SPI modes0no
1yes
12SM ADPCMADPCM recording active0no
1yes
13SM ADPCM HPADPCM high-pass filter active0no
1yes
14SM LINE INADPCM recording selector0microphone
1line in
15SM CLK RANGEInput clock range012..13 MHz
124..26 MHz
When SM DIFF is set, the player inverts the left channel output. For a stereo input this creates virtual
surround, and for a mono input this creates a differential left/right signal.
SM LAYER12 enables MPEG 1.0 and 2.0 layer I and II decoding in addition to layer III. If you enableLayer I and Layer II decoding, you are liable for any patent issues that may arise. Joint licensing
of MPEG 1.0 / 2.0 Layer III does not cover all patents pertaining to layers I and II.
Software reset is initiated by setting SM RESET to 1. This bit is cleared automatically.
If you want to stop decoding a WAV, WMA, or MIDI file in the middle, set SM OUTOFWAV, and send
data honouring DREQ until SM OUTOFWAV is cleared. SCI HDAT1 will also be cleared. For WMA
and MIDI it is safest to continue sending the stream, send zeroes for WAV.
Bits SM EARSPEAKER LO and SM EARSPEAKER HI control the EarSpeaker spatial processing. If
both are 0, the processing is not active. Other combinations activate the processing and select 3 different
effect levels: LO = 1, HI = 0 selects minimal, LO = 0, HI = 1 selects normal, and LO = 1, HI = 1 selects
extreme. EarSpeaker takes approximately 6 MIPS at 44.1 kHz sample rate. EarSpeaker is automatically
disabled with AAC files.
Version 0.91, 2007-02-1235
VLSI
Solution
y
VS1033C
05001000150020002500300035004000
−20
−15
−10
−5
0
5
VS1023 AD Converter with and Without HP Filter
Frequency / Hz
Amplitude / dB
No High−Pass
High−Pass
VS1033c PRELIMINARY
8. FUNCTIONAL DESCRIPTION
If SM TESTS is set, SDI tests are allowed. For more details on SDI tests, look at Chapter 9.10.
SM STREAM activates VS1033’s stream mode. In this mode, data should be sent with as even intervals
as possible and preferable in blocks of less than 512 bytes, and VS1033 makes every attempt to keep its
input buffer half full by changing its playback speed upto 5%. For best quality sound, the average speed
error should be within 0.5%, the bitrate should not exceed 160 kbit/s and VBR should not be used. For
details, see Application Notes for VS10XX. This mode only works with MP3 and WAV files.
SM DACT defines the active edge of data clock for SDI. When ’0’, data is read at the rising edge, when
’1’, data is read at the falling edge.
When SM SDIORD is clear, bytes on SDI are sent MSb first. By setting SM SDIORD, the user may
reverse the bit order for SDI, i.e. bit 0 is received first and bit 7 last. Bytes are, however, still sent in the
default order. This register bit has no effect on the SCI bus.
Setting SM SDISHARE makes SCI and SDI share the same chip select, as explained in Chapter 7.2, if
also SM SDINEW is set.
Setting SM SDINEW will activateVS1002 native serial modes as described in Chapters 7.2.1 and 7.4.2.
Note, that this bit is set as a default when VS1033 is started up.
By activating SM ADPCM andSM RESET at the sametime,theuserwill activateIMAADPCMrecording mode (see section 9.4). If SM ADPCM HP is set (use only for 8 kHz sample rate), ADPCM mode
will start with a high-pass filter. This may help intelligibility of speech when there is lots of background
noise. The difference created to the ADPCM encoder frequency response is as shown in Figure 14.
Figure 14: ADPCM Frequency Responses with 8 kHz sample rate.
SM LINE IN is used to select the input for ADPCM recording. If ’0’, microphone input pins MICP and
MICN are used; if ’1’, LINEIN is used.
SM CLK RANGE activates a clock divider in the XTAL input. When SM CLK RANGE is set, from
the chip’s point of view e.g. 24 MHz becomes 12 MHz. SM CLK RANGE should be set as soon as
possible after a chip reset.
Version 0.91, 2007-02-1236
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
8.7.2SCI STATUS (RW)
SCI STATUS contains information on the current status of VS1033 and lets the user shutdown the chip
without audio glitches.
NameBitsDescription
SS VER6:4Version
SS APDOWN23Analog driver powerdown
SS APDOWN12Analog internal powerdown
SS AVOL1:0Analog volume control
SS VER is 0 for VS1001, 1 for VS1011, 2 for VS1002, 3 for VS1003, and 5 for VS1033.
SS APDOWN2controlsanalog driverpowerdown. Normally thisbitis controlled by thesystemfirmware.
However, if the user wants to powerdown VS1033 with a minimum power-off transient, turn this bit to
1, then wait for at least a few milliseconds before activating reset.
8. FUNCTIONAL DESCRIPTION
SS APDOWN1 controls internal analog powerdown. This bit is meant to be used by the system firmware
only.
SS AVOL is the analog volume control: 0 = -0 dB, 1 = -6 dB, 3 = -12 dB. This register is meant to be
used automatically by the system firmware only. Use SCI VOL to control the analog powerdown.
8.7.3SCI BASS (RW)
NameBitsDescription
ST AMPLITUDE15:12Treble Control in 1.5 dB steps (-8..7, 0 = off)
ST FREQLIMIT11:8Lower limit frequency in 1000 Hz steps (1..15)
SB AMPLITUDE7:4Bass Enhancement in 1 dB steps (0..15, 0 = off)
SB FREQLIMIT3:0Lower limit frequency in 10 Hz steps (2..15)
The Bass Enhancer VSBE is a powerful bass boosting DSP algorithm, which tries to take the most out
of the users earphones without causing clipping.
VSBE is activated when SB AMPLITUDE is non-zero. SB AMPLITUDE should be set to the user’s
preferences, and SB FREQLIMIT to roughly 1.5 times the lowest frequency the user’s audio system can
reproduce. For example setting SCI BASS to 0x00f6 will have 15 dB enhancement below 60 Hz.
Note: Because VSBE tries to avoid clipping, it gives the best bass boost with dynamical music material,
or when the playback volume is not set to maximum. It also does not create bass: the source material
must have some bass to begin with.
Treble Control VSTC is activated when ST AMPLITUDE is non-zero. For example setting SCI BASS
to 0x7a00 will have 10.5 dB treble enhancement at and above 10 kHz.
Bass Enhancer uses about 2.1 MIPS and Treble Control 1.2 MIPS at 44100 Hz sample rate. Both can be
on simultaneously.
Version 0.91, 2007-02-1237
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
8. FUNCTIONAL DESCRIPTION
8.7.4SCI CLOCKF (RW)
The operation of SCI CLOCKF is different in VS1003 and VS1033 than in VS10x1 and VS1002. For
general applications with 12.288 MHz clock use 0x9000 for 3.0 × ..4.0×, or 0xa800 for 3.5 × ..4.0×.
SC ADD tells, howmuchthe decoder firmware is allowedtoadd to the multiplier specified bySC MULT
if more cycles are temporarily needed to decode a WMA stream. The values are:
SC ADDMASKMultiplier addition
00x0000No modification is allowed
10x08000.5×
20x10001.0×
30x18001.5×
SC FREQ is used to tell if the input clock XTALI is running at something else than 12.288 MHz. XTALI
is set in 4 kHz steps. The formula for calculating the correct value for this register is
XT ALI−8000000
4000
(XTALI is in Hz).
Note: The default value 0 is assumed to mean XTALI=12.288 MHz.
Note: because maximum sample rate is
XT ALI
256
, all sample rates are not available if XTALI < 12.288
MHz.
Note: Automatic clock change can only happen when decoding WMA files. Automatic clock change is
done one 0.5× at a time. This does not cause a drop to 1.0× clock and you can use the same SCI and
SDI clock throughout the WMA file.
Example: If SCI CLOCKF is 0x9BE8, SC MULT = 4, SC ADD = 3 and SC FREQ = 0x3E8 = 1000.
This means thatXTALI = 1000×4000+8000000 = 12 MHz. The clock multiplier issetto3.0×XTALI=
36 MHz, and the maximum allowed multiplier that the firmware may automatically choose to use is
(3.0 + 1.5)×XTALI = 54 MHz.
Version 0.91, 2007-02-1238
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
8.7.5SCI DECODE TIME (RW)
When decoding correct data, current decoded time is shown in this register in full seconds.
The user may change the value of this register. In that case the new value should be written twice.
SCI DECODE TIME is reset atevery softwareresetandalso when WAV (PCM or IMA ADPCM), AAC,
WMA, or MIDI decoding starts or ends.
8.7.6SCI AUDATA (RW)
When decoding correct data, the current sample rate and number of channels can be found in bits 15:1
and 0 of SCI AUDATA, respectively. Bits 15:1 contain the sample rate divided by two, and bit 0 is 0 for
mono data and 1 for stereo. Writing to SCI AUDATA will change the sample rate directly.
Example: 44100 Hz stereo data reads as 0xAC45 (44101).
Example: 11025 Hz mono data reads as 0x2B10 (11024).
Example: Writing 0xAC80 sets sample rate to 44160Hz, stereo mode does not change.
8. FUNCTIONAL DESCRIPTION
To reduce the digital power consumption when in idle, you can write a low samplerate to SCI AUDATA.
8.7.7SCI WRAM (RW)
SCI WRAM is used to upload application programs and data to instruction and data RAMs. The start
address must be initialized by writing to SCI WRAMADDR prior to the first write/read of SCI WRAM.
As 16 bits of data can be transferred with one SCI WRAM write/read, and the instruction word is 32 bits
long, twoconsecutive writes/reads areneededforeach instruction word. The byte orderisbig-endian(i.e.
most significant words first). After each full-word write/read, the internal pointer is autoincremented.
8.7.8SCI WRAMADDR (W)
SCI WRAMADDR is used to set the program address for following SCI WRAM writes/reads. Address
offset of 0 is used for X, 0x4000 for Y, and 0x8000 for instruction memory. Peripheral registers can also
be accessed.
SM WRAMADDRDest. addr.Bits/Description
Start.. .EndStart.. .EndWord
0x1800.. .0x187F0x1800...0x187F16X data RAM
0x5800.. .0x587F0x1800...0x187F16Y data RAM
0x8030.. .0x84FF0x0030...0x04FF32Instruction RAM
0xC000.. .0xFFFF0xC000...0xFFFF16I/O
Only user areas in X, Y, and instruction memory are listed above. Other areas can be accessed, but should
not be written to unless otherwise specified.
Version 0.91, 2007-02-1239
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
8. FUNCTIONAL DESCRIPTION
8.7.9SCI HDAT0 and SCI HDAT1 (R)
For WAV files, SCI HDAT1 contains 0x7665 (“ve”). SCI HDAT0 contains the data rate in double word
increments for all supported RIFF WAVE formats: mono and stereo 8-bit or 16-bit PCM, mono and
stereo IMA ADPCM. To get the byte rate of the file, multiply the value by 4. To get the bit rate of the
file, multiply the value by 32. Note: usage of SCI HDAT0 with WAV files has changed from VS1003.
For AAC ADTS streams, SCI HDAT1 contains 0x4154 (“AT”). For AAC ADIF files, SCI HDAT1 contains 0x4144 (“AD”). For AAC .mp4 / .m4a files, SCI HDAT1 contains 0x4D34 (“M4”). SCI HDAT0
contains the average data rate in bytes per second. To get the bit rate of the file, multiply the value by 8.
For WMA files, SCI HDAT1 contains 0x574D(“WM”)andSCI HDAT0 contains the data rate measured
in bytes per second. To get the bit rate of the file, multiply the value by 8.
for MIDI files, SCI HDAT1 contains 0x4D54 (“MT”) and SCI HDAT0 contains the average data rate in
bytes per second. To get the bit rate of the file, multiply the value by 8. Note: usage of SCI HDAT0 with
MIDI has changed from VS1003.
For MP3 files, SCI HDAT1 is between 0xFFE0 and 0xFFFF. SCI HDAT1 / 0 contain the following:
0stereo
HDAT0[5:4]extensionsee ISO 11172-3
HDAT0[3]copyright1copyrighted
0free
HDAT0[2]original1original
0copy
HDAT0[1:0]emphasis3CCITT J.17
2reserved
150/15 microsec
0none
Version 0.91, 2007-02-1240
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
When read, SCI HDAT0 and SCI HDAT1 contain header information that is extracted from MP3 stream
currently being decoded. After reset both registers are cleared, indicating no data has been found yet.
The “sample rate” field in SCI HDAT0 is interpreted according to the following table:
SCI AIADDR indicates the start address of the application code written earlier with SCI WRAMADDR
and SCI WRAM registers. If no application code is used, this register should not be initialized, or it
should be initialized to zero. For more details, see Application Notes for VS10XX.
Version 0.91, 2007-02-1241
8.7.11SCI VOL (RW)
SCI VOL is a volume control for the player hardware. The most significant byte of the volume register
controls the left channel volume, the low part controls the right channel volume. The channel volume
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
sets the attenuation from the maximum volume level in 0.5 dB steps. Thus, maximum volume is 0x0000
and total silence is 0xFEFE.
Note, that after hardware reset the volume is set to full volume. Resetting the software does not reset the
volume setting.
Setting SCI VOL to 0xFFFF will activate analog powerdown mode.
Example: for a volume of -2.0 dB for the left channel and -3.5dB for the right channel: (2.0/0.5) = 4,
3.5/0.5 = 7 → SCI VOL = 0x0407.
Example: SCI VOL = 0x2424 → both left and right volumes are 0x24 * -0.5 = -18.0 dB
8.7.12SCI AICTRL[x] (RW)
8. FUNCTIONAL DESCRIPTION
SCI AICTRL[x] registers ( x=[0 .. 3] ) can be used to access the user’s application program.
Version 0.91, 2007-02-1242
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
9. OPERATION
9Operation
9.1Clocking
VS1033 operates on a single, nominally 12.288 MHz fundamental frequency master clock. This clock
can begeneratedbyexternal circuitry (connectedtopinXTALI)orby the internal clock chrystal interface
(pins XTALI and XTALO).
VS1033 can also use 24..26 MHz clocks when SM CLK RANGE is set to 1. From the chip’s point of
view the input clock is then 12..13 MHz.
9.2Hardware Reset
When the XRESET -signal is driven low, VS1033 is reset and all the control registers and internal states
are set to the initial values. XRESET-signal is asynchronous to any external clock. The reset mode
doubles as a full-powerdown mode, where both digital and analog parts of VS1033 are in minimum
power consumption stage, and where clocks are stopped. Also XTALO is grounded.
When XRESET is asseted, all output pins go to their default states. All input pins will go to highimpedance state (to input state), except SO, which is still controlled by the XCS.
After a hardware reset (or at power-up) DREQ will stay down for around 20000 clock cycles, which
means an approximate 1.6 ms delay if VS1033 is run at 12.288 MHz. After this the user should set
such basic software registers as SCI MODE, SCI BASS, SCI CLOCKF, and SCI VOL before starting
decoding. See section 8.7 for details.
If the input clock is 24..26 MHz, SM CLK RANGE should be set as soon as possible after a chip reset
without waiting for DREQ.
Internal clock can be multiplied with a PLL. Supported multipliers through the SCI CLOCKF register
are 1.0 × . . . 4.5× the input clock. Reset value for Internal Clock Multiplier is 1.0×. If typical values
are wanted, the Internal Clock Multiplier needs to be set to 3.0× after reset. Wait until DREQ rises, then
write value 0x9800 to SCI CLOCKF (register 3). See section 8.7.4 for details.
9.3Software Reset
In some cases the decoder software hastobereset. This is done by activating bit 2 in SCI MODE register
(Chapter 8.7.1). Then wait for at least 2 µs, then look at DREQ. DREQ will stay down for at least 20000
clock cycles, which means an approximate 1.6 ms delay if VS1033 is run at 12.288 MHz. After DREQ
is up, you may continue playback as usual.
If you want to make sure VS1033 doesn’t cut the ending of low-bitrate data streams and you want to do
a software reset, it is recommended to feed 2048 zeros (honoring DREQ) to the SDI bus after the file and
before the reset. This is especially important for MIDI files.
If youwantto interrupt the playingofaWAV,AAC, WMA,orMIDIfile in the middle, setSM OUTOFWAV
Version 0.91, 2007-02-1243
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
in themoderegister, and senddatahonouring DREQ (with athree-secondtimeout)until SM OUTOFWAV
is cleared (SCI HDAT1 will also be cleared) before continuing with a software reset. For WMA and
MIDI it is safest to continue sending the stream, send zeroes for WAV. MP3 can be interrupted without
SM OUTOFWAV by just sending zero bytes, because it is a stream format.
9. OPERATION
Version 0.91, 2007-02-1244
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
9. OPERATION
9.4ADPCM Recording
This chapter explains how to create RIFF/WAV file with IMA ADPCM format. This is a widely supported ADPCM format and many PC audio playback programs can play it. IMA ADPCM recording
gives roughly a compression ratio of 4:1 compared to linear, 16-bit audio. This makes it possible to
record 8 kHz audio at 32.44 kbit/s.
9.4.1Activating ADPCM mode
IMA ADPCM recording mode is activated by setting bits SM RESET and SM ADPCM in SCI MODE.
Optionally a high-pass-filter can be enabled for 8kHzsamplerateby also setting SM ADPCM HP at the
same time. Line input is used instead of mic if SM LINE IN is set. Before activatingADPCMrecording,
user must write a clock divider value to SCI AICTRL0 and gain to SCI AICTRL1.
The differences of using SM ADPCM HP are presented in figure 14 (page 36). As a general rule, audio
will be fuller and closer to original if SM ADPCM HP is not used. However, speech may be more
intelligible with the high-pass filter active. Use the filter only with 8kHz sample rate.
Before activating ADPCM recording, user should write a clock divider value to SCI AICTRL0. The
F
sampling frequency is calculated from the following formula: f
s
=
c
, where Fcis the internal clock
256×d
(CLKI) and d is the divider value in SCI AICTRL0. The lowest valid value for d is 4. If SCI AICTRL0
contains 0, the default divider value 12 is used.
Examples:
F
= 2.0 × 12.288 MHz, d = 12. Now f
c
F
= 2.5 × 14.745 MHz, d = 18. Now f
c
F
= 2.5 × 13 MHz, d = 16. Now f
c
=
s
2.0×12288000
=
s
s
2.5×13000000
256×12
2.5×14745000
=
256×18
256×16
= 8000 Hz.
= 8000 Hz.
= 7935 Hz.
Also, before activating ADPCM mode, the user has to set linear recording gain control to register
SCI AICTRL1. 1024 is equal to digital gain 1, 512 is equal to digital gain 0.5 and so on. If the user
wants to use automatic gain control (AGC), SCI AICTRL1 should be set to 0. Typical speech applications usually are better off using AGC, as this takes care of relatively uniform speech loudness in
recordings.
Since VS1033c SCI AICTRL2 controls the maximum AGC gain. If SCI AICTRL2 is zero, the maximum gain is 65535 (64×), i.e. whole range is used. This is compatible with previous operation.
9.4.2Reading IMA ADPCM Data
After IMA ADPCM recording has been activated, registers SCI HDAT0 and SCI HDAT1 have new
functions.
The IMA ADPCM sample buffer is 1024 16-bit words. The fill status of the buffer can be read from
Version 0.91, 2007-02-1245
SCI HDAT1. If SCI HDAT1 is greater than 0, you can read as many 16-bit words from SCI HDAT0. If
the data is not read fast enough, the buffer overflows and returns to empty state.
Note: if SCI HDAT1 ≥ 896, it may be better to wait for the buffer to overflow and clear before reading
samples. That way you may avoid buffer aliasing.
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
9. OPERATION
Each IMA ADPCM block is 128 words, i.e. 256 bytes. If you wish to interrupt reading data and possibly
continue later, please stop at a 128-word boundary. This way whole blocks are skipped and the encoded
stream stays valid.
9.4.3Adding a RIFF Header
To make your IMA ADPCM file a RIFF / WAV file, you have to add a header before the actual data.
Note that 2- and 4-byte values are little-endian (lowest byte first) in this format:
File OffsetField NameSizeBytesDescription
0ChunkID4"RIFF"
4ChunkSize4F0 F1 F2 F3File size - 8
If we have n audio blocks, the values in the table are as follows:
F = n × 256 + 52
R = F
B =
(see Chapter 9.4.1 to see how to calculate Fs)
s
Fs×256
505
S = n × 505. D = n × 256
If you know beforehand how much you are going to record, you may fill in the complete header before
any actual data. However, if you don’t know how much you are going to record, you have to fill in the
header size datas F , S and D after finishing recording.
The 128 words (256 bytes) of an ADPCM block are read from SCI HDAT0 and written into file as
follows. The high 8 bits of SCI HDAT0 should be written as the first byte to a file, then the low 8 bits.
Note that this is contrary to the default operation of some 16-bit microcontrollers, and you may have to
take extra care to do this right.
A way to see if you have written the file in the right way is to check bytes 2 and 3 (the first byte counts
as byte 0) of each 256-byte block. Byte 3 should always be zero.
Version 0.91, 2007-02-1246
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
9. OPERATION
9.4.4Playing ADPCM Data
In order to play back your IMA ADPCM recordings, you have to have a file with a header as described
in Chapter 9.4.3. If this is the case, all you need to do is to provide the ADPCM file through SDI as you
would with any audio file.
9.4.5Sample Rate Considerations
VS10xx chips that support IMA ADPCM playback are capable of playing back ADPCM files with
any sample rate. However, some other programs may expect IMA ADPCM files to have some exact
sample rates, like 8000 or 11025Hz. Also, some programs or systems do not support sample rates below
8000Hz.
However, if you don’t have an appropriate clock, you may not be able to get an exact 8kHz sample rate.
If you have a 12 MHz clock, the closest sample rate you can get with 2.0 × 12 MHz and d = 12 is
f
= 7812.5Hz. Because the frequency error is only 2.4%, it may be best to set fs= 8000Hz to the
s
header if the same file is also to be played back with an PC. This causes the sample to be played back a
little faster (one minute is played in 59 seconds).
Note, however, that unless absolutely necessary, sample rates should not be tweaked in the way described
here.
If you want better quality with the expense of increased data rate, you can use higher sample rates, for
example 16 kHz.
9.4.6Example Code
The following code initializes IMA ADPCM encoding on VS1003b/VS1023 and shows how to read the
data.
} while (w < 256 || w >= 896); /* wait until 512 bytes available */
while (idx < 512) {
w = ReadMp3SpiReg(SCI_HDAT0);
db[idx++] = w>>8;
db[idx++] = w&0xFF;
}
idx = 0;
write_block(datasector++, db); /* Write output block to disk */
}
.../* Fix WAV header information*/
.../* Then update FAT information */
ResetMP3();/* Normal reset, restore default settings */
SetMp3Vol(vol);
}
Version 0.91, 2007-02-1248
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
9. OPERATION
9.5SPI Boot
If GPIO0 is set with a pull-up resistor to 1 at boot time, VS1033 tries to boot from external SPI memory.
SPI boot redefines the following pins:
Normal ModeSPI Boot Mode
GPIO0xCS
GPIO1CLK
DREQMOSI
GPIO2MISO
The memory has to be an SPI Bus Serial EEPROM with 16-bit addresses (i.e. at least 1 KiB). The serial
speed used by VS1033 is 245 kHz with the nominal 12.288 MHz clock. The first three bytes in the
memory have to be 0x50, 0x26, 0x48.
9.6Play/Decode
This is the normal operation mode of VS1033. SDI data is decoded. Decoded samples are converted to
analog domain by the internal DAC. If no decodable data is found, SCI HDAT0 and SCI HDAT1 are set
to 0 and analog outputs are muted.
When there is no input for decoding, VS1033 goes into idle mode (lower power consumption than during
decoding) and actively monitors the serial data input for valid data.
All different formats can be played back-to-back without software reset in-between. Send at least 2052
zeros after each stream. However, using software reset between streams may still be a good idea, as it
guards againstbrokenfiles. In this case you shouldtwaitfor the completion of the decoding (SCI HDAT1
and SCI HDAT0 become zero) before issuing software reset.
9.7Feeding PCM data
VS1033 can be used as a PCM decoder by sending a WAV file header. If the length sent in the WAV
header is 0 or 0xFFFFFFF, VS1033 will stay in PCM mode indefinitely (or until SM OUTOFWAV has
been set). 8-bit linear and 16-bit linear audio is supported in mono or stereo.
Version 0.91, 2007-02-1249
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
9. OPERATION
9.8Extra Parameters
The following structure is in X memory at address 0x1940 and can be used to change some extra parameters or get various information. The chip ID is also easily available.
u_int32 chipID;/*1940/41 Initialized at reset for your convenience */
u_int16 version;/*1942 - structure version */
u_int16 midiConfig; /*1943 */
u_int16 config1;/*1944 */
u_int16 config2;/*1945 configs are not cleared between files */
u_int32 jumpPoints[16]; /*1946..65 file byte offsets */
u_int16 latestJump; /*1966 index to lastly updated jumpPoint */
s_int16 seek1;/*1967 file data inserted/removed bytes -32768..32767*/
s_int16 seek2;/*1968 file data inserted/removed kB -32768..32767*/
s_int16 resync;/*1969 > 0 for automatic m4a, ADIF, WMA resyncs */
union {
struct {
u_int32 curPacketSize;
u_int32 packetSize;
} wma;
struct {
u_int16 sceFoundMask; /* SCE’s found since last clear */
u_int16 cpeFoundMask; /* CPE’s found since last clear */
u_int16 lfeFoundMask; /* LFE’s found since last clear */
u_int16 playSelect;/* 0 = first any, initialized at aac init */
s_int16 dynCompress; /* -8192=1.0, initialized at aac init */
Notice that reading two-word variables through the SCI WRAMADDR and SCI WRAM interface is
not protected in any way. The variable can be updated between the read of the low and high parts. The
problem arises when both the low and high parts change values. To determine if the value is correct, you
should read the value twice and compare the results.
The following exampleshowswhathappenswhenbytesLeftisdecreasedfrom0x10000to0xffffand
the update happens between low and high part reads or after high part read.
Read Invalid
AddressValue
0x196a0x0000 change after this
0x196b0x0000
0x196a0xffff
0x196b0x0000
AddressValue
0x196a0x0000
0x196b0x0001 change after this
0x196a0xffff
0x196b0x0000
You can see that in the invalid read the low part wraps from 0x0000 to 0xffffwhile the high part stays the
same. In this case the second read gives a valid answer, otherwise always use the value of the first read.
The second read is needed when it is possible that the low part wraps around, changing the high part, i.e.
when the low part is small. bytesLeft is only decreased by one at a time, so a reread is needed only
if the low part is 0.
Version 0.91, 2007-02-1250
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
9.8.1Common Parameters
ParameterAddressUsage
chipID0x1940/41Fuse-programmed unique ID (copy)
version0x1942Structure version – 0x0001
jumpPoints[16]0x1946-65Packet offsets for WMA and AAC
latestJump0x1966Index to latest jumpPoint
seek10x1967Seek amount in bytes
seek20x1968Seek amount in kilobytes
resync0x1969Automatic resync selector
The fuse-programmed ID is read at startup and copied into the chipID field. The version field can
be used to determine the layout of the rest of the structure. The version number is changed when the
structure is changed.
jumpPoints contain 32-bit file offsets. Each valid (non-zero) entry indicates a start of a packet for
WMA or start of a raw data block for AAC (ADIF, .mp4 / .m4a). latestJump contains the index of
the entry that was updated last. If you only read entry pointed to by latestJump you do not need to
read the entry twice to ensure validity. Jump point information can be used to implement perfect fast
forward and rewind for WMA and AAC (ADIF, .mp4 / .m4a).
9. OPERATION
seek1 and seek2 fields are used when music data is skipped or inserted. Negative values mean that
data has been added (for example in rewind operation), positive values mean that data has been skipped.
Youcanuseeitherseek1,whichgives the seek amount in bytes, or seek2 whichgives the seek amount
in kilobytes, or you can use both. The field value is zeroed when the firmware has detected the seek.
resync field is usedtoforcearesynchronizationto the stream for WMA and AAC(ADIF, .mp4 / .m4a).
This field can be used to implement almost perfect fast forward and rewind for WMA and AAC (ADIF,
.mp4 / .m4a). The user should set this field before performing data seeks if they are not in packet or data
block boundaries. The field value tells how many tries are allowed before giving up. The value 32767
gives infinite tries, in which case the user must use SM OUTOFWAV or software reset to end decoding.
In every case remember to use seek1 and/or seek2 fields to indicate the skipped/inserted data.
Note: WMA, ADIF, and .mp4 / .m4a files begin with a metadata section, which must be fully processed
before any fast forward or rewindoperation. When the first jumpPoint appears it is safe to perform seeks.
You can also detect the start of decoding from SCI DECODE TIME.
9.8.2WMA
ParameterAddressUsage
curPacketSize0x196a/6bThe size of the packet being processed
packetSize0x196c/6dThe packet size in ASF header
The ASF header packet size is availableinpacketSize. With this information and a packet start offset
from jumpPoints you can parse the packet headers and skip packets in ASF files.
Version 0.91, 2007-02-1251
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
9.8.3AAC
ParameterAddressUsage
sceFoundMask0x196aSingle channel elements found
cpeFoundMask0x196bChannel pair elements found
lfeFoundMask0x196cLow frequency elements found
playSelect0x196dPlay element selection
dynCompress0x196eCompress coefficient for DRC, -8192=1.0
dynBoost0x196fBoost coefficient for DRC, 8192=1.0
playSelect determines which element to decode if a stream has multiple elements. The value is
set to 0 each time AAC decoding starts, which causes the first element that appears in the stream to be
selected for decoding. Other values are: 0x01 - select first single channel element (SCE), 0x02 - select
first channel pair element (CPE), 0x03 - select first low frequency element (LFE), S ∗ 16 + 5 - select
SCE number S, P ∗ 16 + 6 - select CPE number P, L ∗ 16 + 7 - select LFE number L. When automatic
selection has been performed, playSelect reflects the selected element.
9. OPERATION
sceFoundMask, cpeFoundMask, and lfeFoundMask indicate which elements have been found
in an AACstream since the variables havelastbeencleared. The values can be usedtopresentanelement
selection menu with only the available elements.
dynCompress and dynBoost changethebehaviorof the dynamic range control (DRC) that is present
in some AAC streams. These are also initialized when AAC decoding starts.
SCI HDAT0 contains the average bitrate in bytes per second, is updated once per second and it can be
used to calculate an estimate of the remaining playtime.
bytesLeft0x196a/6bThe number of bytes left in this track
midiConfig controls the reverb effect and play speed.
SCI HDAT0 contains the average bitrate in bytes per second, is updated once per second and it can be
used together with bytesLeft to calculate an estimate of the remaining playtime.
Version 0.91, 2007-02-1252
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
9. OPERATION
9.9Fast Forward / Rewind
9.9.1MP3
MPEG1.0 and MPEG2.0 layer 3 defines a stream format suitable for random-access. When you want to
skip forward or backwards in the file, first send 2048 zeros, then continue sending the file from the new
location.
By sending zeros you make certain a partial frame does not cause loud artefacts in the sound. The normal
file type checking then finds a new MP3 header and continues decoding.
9.9.2AAC - ADTS
MPEG2.0 Advanced Audio Coded (AAC) defines a stream format suitable for random-access (ADTS).
When you want to skip forward or backwards in the file, first send 2048 zeros, then continue sending the
file from the new location.
By sending zeros you make certain a partial frame does not cause loud artefacts in the sound. The normal
file type checking then finds a new ADTS header and continues decoding.
9.9.3AAC - ADIF, MP4
MPEG4.0 Advanced Audio Codec (AAC) specifies a multimedia file format (.mp4 / .m4a) but does not
specify a stream format and MPEG2.0 AAC specifies a file format (ADIF) in addition to the streamable
ADTS format. ADIF and .mp4 / .m4a are not suitable for random-access and it is recommended that
they are converted to ADTS format for playback.
However, it is also possible to implement fast forward and rewind for ADIF and .mp4 / .m4a files. The
easiest way is to use the resync field (see section 9.8.1):
• Write 8192 to resync
– Write 0x1969 to SCI WRAMADDR, Write 0x2000 to SCI WRAM
• Send 2048 zeroes
• Make a seek X in the file (X > 0 for forward seek)
• Indicate the low part of the seek amount by writing to seek1
– Write 0x1967 to SCI WRAMADDR, Write (X − 2048)&1023 to SCI WRAM
• Indicate the high part of the seek amount by writing to seek2
– Write 0x1968 to SCI WRAMADDR, Write (X − 2048)/1024 to SCI WRAM
• Continue sending the file from the new location
Perfect fast forward and rewind can be implemented by using the jumpPoints table and making seeks
only on packet or data block boundaries.
Version 0.91, 2007-02-1253
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
9.9.4WMA
Windows Media Audio (WMA) is enclosed as data packets into Advanced Systems Format (ASF) files.
This file format is not suitable for random-access.
However, it is also possible to implement fast forward and rewind for WMA files. The easiest way is to
use the resync field (see Section 9.9.3), perfect fast forward and rewind can be implemented by using
the jumpPoints table and making seeks only on packet or data block boundaries.
9.9.5Midi
Midi is not at all suitable for random-access. You can implement fast forward using the playSpeed
bits of the midiConfig field to select 1-128× play speed. SCI DECODE TIME also speeds up.
If necessary, rewind can be implemented by restarting the decoding of a MIDI file and fast forwarding
to the appropriate place. SCI DECODE TIME can be used to decide when the right place has been
reached. This is best suited for soundless rewind.
9. OPERATION
Version 0.91, 2007-02-1254
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
9. OPERATION
9.10SDI Tests
There are several test modes in VS1033, which allow the user to perform memory tests, SCI bus tests,
and several different sine wave tests.
All tests are started in a similar way: VS1033 is hardware reset, SM TESTS is set, and then a test
command is sent to the SDI bus. Each test is started by sending a 4-byte special command sequence,
followed by 4 zeros. The sequences are described below.
9.10.1Sine Test
Sine test is initialized with the 8-byte sequence 0x53 0xEF 0x6E n 0 0 0 0, where n defines the sine test
to use. n is defined as follows:
The frequency of the sine to be output can now be calculated from F = F
S
×
128
.
s
Example: Sine test is activated with value 126, which is 0b01111110. Breaking n to its components,
F
Idx = 0b011 = 3 and thus Fs= 22050H z. S = 0b11110 = 30, and thus the final sine frequency
s
F = 22050H z ×
30
≈ 5168Hz.
128
To exit the sine test, send the sequence 0x45 0x78 0x69 0x74 0 0 0 0.
Note: Sine test signals go through the digital volume control, so it is possible to test channels separately.
Version 0.91, 2007-02-1255
9.10.2Pin Test
Pin test is activated with the 8-byte sequence 0x50 0xED 0x6E 0x54 0 0 0 0. This test is meant for chip
production testing only.
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
9.10.3Memory Test
Memory test mode is initialized with the 8-byte sequence 0x4D 0xEA 0x6D 0x54 0 0 0 0. After this
sequence, wait for 500000 clock cycles. The result can be read from the SCI register SCI HDAT0, and
’one’ bits are interpreted as follows:
Bit(s)MaskMeaning
150x8000Test finished
14:7Unused
60x0040Mux test succeeded
50x0020Good I RAM
40x0010Good Y RAM
30x0008Good X RAM
20x0004Good I ROM
10x0002Good Y ROM
00x0001Good X ROM
0x807fAll ok
9. OPERATION
Memory tests overwrite the current contents of the RAM memories.
9.10.4SCI Test
Sci test is initialized with the 8-byte sequence 0x53 0x70 0xEE n 0 0 0 0, where n − 48 is the register
number to test. The content of the given register is read and copied to SCI HDAT0. If the register to be
tested is HDAT0, the result is copied to SCI HDAT1.
Example: if n is 48, contents of SCI register 0 (SCI MODE) is copied to SCI HDAT0.
Version 0.91, 2007-02-1256
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
10. VS1033 REGISTERS
10VS1033 Registers
10.1Who Needs to Read This Chapter
User software is required when a user wishes to add some own functionality like DSP effects to VS1033.
However, most users of VS1033 don’t need to worry about writing their own code, or about this chapter,
including those who only download software plug-ins from VLSI Solution’s Web site.
10.2The Processor Core
VS DSP is a 16/32-bit DSP processor core that also had extensive all-purpose processor features. VLSI
Solution’s free VSKIT Software Package contains all the tools and documentation needed to write, simulate and debug Assembly Language or Extended ANSI C programs for the VS DSP processor core.
VLSI Solution also offers a full Integrated Development Environment VSIDE for full debug capabilities.
10.3VS1033 Memory Map
VS1033’s Memory Map is shown in Figure 15.
10.4SCI Registers
SCI registers described in Chapter 8.7 can be found here between 0xC000..0xC00F. In addition to these
registers, there is one in address 0xC010, called SCI CHANGE.
SCI registers, prefix SCI
RegTypeResetAbbrev[bits]Description
0xC010r0CHANGE[5:0]Last SCI access address
SCI CHANGE bits
NameBitsDescription
SCI CH WRITE41 if last access was a write cycle
SCI CH ADDR3:0SCI address of last access
10.5Serial Data Registers
SDI registers, prefix SER
RegTypeResetAbbrev[bits]Description
0xC011r0DATALast received 2 bytes, big-endian
0xC012w0DREQ[0]DREQ pin control
Version 0.91, 2007-02-1257
VLSI
Solution
y
VS1033c PRELIMINARY
00000000
Instruction (32−bit)Y (16−bit)X (16−bit)
System Vectors
User
Instruction
RAM
X DATA
RAM
Y DATA
RAM
00300030
Y DATA
ROM
X DATA
ROM
40004000
Instruction
ROM
C000
C100C100
C000
05000500
80008000
StackStack
1940
1880
18001800
1880
1940
2800
2000
2800
2000
FFFFFFFF
I/O
UserUser
VS1033C
10. VS1033 REGISTERS
10.6DAC Registers
Figure 15: User’s Memory Map.
DAC registers, prefix DAC
RegTypeResetAbbrev[bits]Description
0xC013rw0FCTLLDAC frequency control, 16 LSbs
0xC014rw0FCTLHDAC frequency control 4MSbs, PLL control
0xC015rw0LEFTDAC left channel PCM value
0xC016rw0RIGHTDAC right channel PCM value
Every fourth clock cycle, an internal 26-bit counter is added to by (DAC FCTLH & 15) × 65536 +
DAC FCTLL. Whenever this counter overflows, valuesfrom DAC LEFT and DAC RIGHT are read and
a DAC interrupt is generated.
Version 0.91, 2007-02-1258
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
10. VS1033 REGISTERS
10.7GPIO Registers
GPIO registers, prefix GPIO
RegTypeResetAbbrev[bits]Description
0xC017rw0DDR[7:0]Direction
0xC018r0IDATA[7:0]Values read from the pins
0xC019rw0ODATA[7:0]Values set to the pins
GPIO DIR is used to set the direction of the GPIO pins. 1 means output. GPIO ODATA remembers its
values even if a GPIO DIR bit is set to input.
GPIO registers don’t generate interrupts.
Note that in VS1033 the VSDSP registers can be read and written through the SCI WRAMADDR and
SCI WRAM registers. You can thus use the GPIO pins quite conveniently.
Version 0.91, 2007-02-1259
VLSI
Solution
y
VS1033c PRELIMINARY
10. VS1033 REGISTERS
10.8Interrupt Registers
Interrupt registers, prefix INT
RegTypeResetAbbrev[bits]Description
0xC01Arw0ENABLE[7:0]Interrupt enable
0xC01Bw0GLOB DIS[-]Write to add to interrupt counter
0xC01Cw0GLOB ENA[-]Write to subtract from interrupt counter
0xC01Drw0COUNTER[4:0]Interrupt counter
INT ENABLE controls the interrupts. The control bits are as follows:
INT ENABLE bits
NameBitsDescription
INT EN TIM17Enable Timer 1 interrupt
INT EN TIM06Enable Timer 0 interrupt
INT EN RX5Enable UART RX interrupt
INT EN TX4Enable UART TX interrupt
INT EN MODU3Enable AD modulator interrupt
INT EN SDI2Enable Data interrupt
INT EN SCI1Enable SCI interrupt
INT EN DAC0Enable DAC interrupt
VS1033C
Note: It may take upto 6 clock cycles before changing INT ENABLE has any effect.
Writing any value to INT GLOB DIS adds one to the interrupt counter INT COUNTER and effectively
disables all interrupts. It may take upto 6 clock cycles before writing to this register has any effect.
Writing any value to INT GLOB ENA subtracts one from the interrupt counter (unless INT COUNTER
already was 0). If the interrupt counter becomes zero, interrupts selected with INT ENABLE are restored. An interrupt routine should always write to this register as the last thing it does, because interrupts automatically add one to the interrupt counter, but subtracting it back to its initial value is the
responsibility of the user. It may take upto 6 clock cycles before writing this register has any effect.
By reading INT COUNTER the user may check if the interrupt counter is correct or not. If the register
is not 0, interrupts are disabled.
Version 0.91, 2007-02-1260
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
10. VS1033 REGISTERS
10.9A/D Modulator Registers
Interrupt registers, prefix AD
RegTypeResetAbbrev[bits]Description
0xC01Erw0DIVA/D Modulator divider
0xC01Frw0DATAA/D Modulator data
AD DIV bits
NameBitsDescription
ADM POWERDOWN151 in powerdown
ADM DIVIDER14:0Divider
ADM DIVIDER controls the AD converter’s sampling frequency. To gather one sample, 128 × n clock
cycles are used (n is value of AD DIV). The lowest usable value is 4, which gives a 48 kHz sample rate
when CLKI is 24.576 MHz. When ADM POWERDOWN is 1, the A/D converter is turned off.
AD DATA contains the latest decoded A/D value.
Version 0.91, 2007-02-1261
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
10. VS1033 REGISTERS
10.10Watchdog v1.0 2002-08-26
The watchdog consist of a watchdog counter and some logic. After reset, the watchdog is inactive.
The counter reload value can be set by writing to WDOG CONFIG. The watchdog is activated by writing 0x4ea9 to register WDOG RESET. Every time this is done, the watchdog counter is reset. Every
65536’th clock cycle the counter is decremented by one. If the counter underflows, it will activate vsdsp’s internal reset sequence.
Thus, after the first 0x4ea9 write to WDOG RESET, subsequent writes to the same register with the
same value must be made no less than every 65536×WDOG CONFIG clock cycles.
Once started, the watchdog cannot be turned off. Also, a write to WDOG CONFIG doesn’t change the
counter reload value.
After watchdoghasbeen activated,anyread/write operation from/to WDOG CONFIG or WDOG DUMMY
will invalidate the next write operation to WDOG RESET. This will prevent runaway loops from resetting the counter, even if they do happen to write the correct number. Writing a wrong value to
WDOG RESET will also invalidate the next write to WDOG RESET.
Reads from watchdog registers return undefined values.
RS232 UART implements a serial interface using rs232 standard.
Figure 16: RS232 Serial Interface Protocol
When the line is idling, it stays in logic high state. When a byte is transmitted, the transmission begins
with a start bit (logic zero) and continues with data bits (LSB first) and ends up with a stop bit (logic
high). 10 bits are sent for each 8-bit byte frame.
10.11.1Registers
UART registers, prefix UARTx
RegTypeResetAbbrevDescription
0xC028r0STATUS[4:0]Status
0xC029r/w0DATA[7:0]Data
0xC02Ar/w0DATAH[15:8]Data High
0xC02Br/w0DIVDivider
10.11.2Status UARTx STATUS
A read from the status register returns the transmitter and receiver states.
UARTx STATUS Bits
NameBitsDescription
UART ST FRAMEERR4Framing error (stop bit was 0)
UART ST RXORUN3Receiver overrun
UART ST RXFULL2Receiver data register full
UART ST TXFULL1Transmitter data register full
UART ST TXRUNNING0Transmitter running
UART ST FRAMEERR is set if the stop bit of the received byte was 0.
UART ST RXORUN is set if a received byte overwrites unread data when it is transferred from the
receiver shift register to the data register, otherwise it is cleared.
UART ST RXFULL is set if there is unread data in the data register.
UART ST TXFULL is set if a write to the data register is not allowed (data register full).
UART ST TXRUNNING is set if the transmitter shift register is in operation.
Version 0.91, 2007-02-1263
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
10. VS1033 REGISTERS
10.11.3Data UARTx DATA
A read from UARTx DATA returns the received byte in bits 7:0, bits 15:8 are returned as ’0’. If there is
no more data to be read, the receiver data register full indicator will be cleared.
A receive interrupt will be generated when a byte is moved from the receivershift register to the receiver
data register.
A write to UARTx DATA sets a byte for transmission. The data is taken from bits 7:0, other bits in the
written value are ignored. If the transmitter is idle, the byte is immediately moved to the transmitter shift
register, a transmit interrupt request is generated, and transmission is started. If the transmitter is busy,
the UART ST TXFULL will be set and the byte remains in the transmitter data register until the previous
byte has been sent and transmission can proceed.
10.11.4Data High UARTx DATAH
The same as UARTx DATA, except that bits 15:8 are used.
10.11.5Divider UARTx DIV
UARTx DIV Bits
NameBitsDescription
UART DIV D115:8Divider 1 (0..255)
UART DIV D27:0Divider 2 (6..255)
The divider is set to 0x0000 in reset. The ROM boot code must initialize it correctly depending on the
master clock frequency to get the correct bit speed. The second divider (D2) must be from 6 to 255.
f
The communication speed f =
m
(D1+1)×(D2)
, where fmis the master clock frequency, and f is the
TX/RX speed in bps.
Divider values for common communication speeds at 26 MHz master clock:
Transmitter operates as follows: After an 8-bit word is written to the transmit data register it will be
transmitted instantly if the transmitter is not busy transmitting the previous byte. When the transmission
begins a TX INTR interrupt will be sent. Status bit [1] informs the transmitter data register empty (or
full state) and bit [0] informs the transmitter (shift register) empty state. A new word must not be written
to transmitter data register if it is not empty (bit [1] = ’0’). The transmitter data register will be empty
as soon as it is shifted to transmitter and the transmission is begun. It is safe to write a new word to
transmitter data register every time a transmit interrupt is generated.
Receiver operates as follows: It samples the RX signal line and if it detects a high to low transition, a
start bit is found. After this it samples each 8 bit at the middle of the bit time (using a constant timer),
and fills the receiver (shift register) LSB first. Finally the data in the receiver is moved to the reveive
data register, the stop bit state is checked (logic high = ok, logic low = framing error) for status bit[4],
the RX INTR interrupt is sent, status bit[2] (receive data register full) is set, and status bit[2] old state is
copied to bit[3] (receive data overrun). After that the receiver returns to idle state to wait for a new start
bit. Status bit[2] is zeroed when the receiver data register is read.
10. VS1033 REGISTERS
RS232 communication speed is set using two clock dividers. The base clock is the processor master
clock. Bits 15-8 in these registers are for first divider and bits 7-0 for second divider. RX sample
frequency is the clock frequency that is input for the second divider.
Version 0.91, 2007-02-1265
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
10. VS1033 REGISTERS
10.12Timers v1.0 2002-04-23
There are two 32-bit timers that can be initialized and enabled independently of each other. If enabled,
a timer initializes to its start value, written by a processor, and starts decrementing every clock cycle.
When the value goes past zero, an interrupt is sent, and the timer initializes to the value in its start value
register, and continues downcounting. A timer stays in that loop as long as it is enabled.
A timer has a 32-bit timer register for down counting and a 32-bit TIMER1 LH register for holding the
timer start value written by the processor. Timers have also a 2-bit TIMER ENA register. Each timer is
enabled (1) or disabled (0) by a corresponding bit of the enable register.
TIMER CF CLKDIV is the master clock divider for all timer clocks. The generated internal clock
f
frequency f
m
=
i
, where fmis the master clock frequency and c is TIMER CF CLKDIV. Example:
c+1
Witha 12 MHz master clock, TIMER CF DIV=3 divides the master clock by 4, and the output/sampling
clock would thus be f
i
=
12MHz
3+1
= 3MHz.
Version 0.91, 2007-02-1266
VLSI
Solution
y
VS1033c PRELIMINARY
10.12.3Configuration TIMER ENABLE
TIMER ENABLE Bits
NameBitsDescription
TIMER EN T11Enable timer 1
TIMER EN T00Enable timer 0
10.12.4Timer X Startvalue TIMER Tx[L/H]
VS1033C
10. VS1033 REGISTERS
The 32-bit start value TIMER Tx[L/H] sets the initial counter value when the timer is reset. The timer
f
interrupt frequency f
t
=
i
where fiis the master clock obtained with the clock divider (see Chap-
c+1
ter 10.12.2 and c is TIMER Tx[L/H].
Example: Witha 12 MHz master clock andwithTIMER CF CLKDIV=3, the master clock fi= 3MHz.
If TIMER TH=0, TIMER TL=99, then the timer interrupt frequency f
3MHz
=
t
99+1
= 30kHz.
10.12.5Timer X Counter TIMER TxCNT[L/H]
TIMER TxCNT[L/H] contains the current counter values. By reading this register pair, the user may get
knowledge of how long it will take before the next timer interrupt. Also, by writing to this register, a
one-shot different length timer interrupt delay may be realized.
10.12.6Interrupts
Each timer has its own interrupt, which is asserted when the timer counter underflows.
Version 0.91, 2007-02-1267
VLSI
Solution
y
VS1033c PRELIMINARY
MCLK
SDATA
SCLK
LROUT
MSBLSBMSB
Left Channel WordRight Channel Word
10.13I2S DAC Interface
The I2S Interface makes it possible to attach an external DAC to the system.
I2S CF ENA enables the I2S interface. After reset the interface is disabled and the pins are used for
GPIO.
I2S CF MCLK ENA enables the MCLK output. The frequency is either directly the input clock (nominal 12.288 MHz), or half the input clock when mode register bit SM CLK RANGE is set to 1 (2426 MHz input clock).
I2S CF SRATE controls the output samplerate. When set to 48 kHz, SCLK is MCLK divided by 8,
when 96 kHz SCLK is MCLK divided by 4, and when 192 kHz SCLK is MCLK divided by 2.
Figure 17: I2S Interface, 192 kHz.
To enable I2S first write 0xc017 to SCI WRAMADDR and 0x33 to SCI WRAM, then write 0xc040 to
SCI WRAMADDR and 0x0c to SCI WRAM.
See application notes for more information.
Version 0.91, 2007-02-1268
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
10. VS1033 REGISTERS
10.14System Vector Tags
The System Vector Tags are tags that may be replaced by the user to take control over several decoder
functions.
10.14.1AudioInt, 0x20
Normally contains the following VS DSP assembly code:
jmpi DAC_INT_ADDRESS,(i6)+1
The user may, at will, replace the first instruction with a jmpi command to gain control over the audio
interrupt.
10.14.2SciInt, 0x21
Normally contains the following VS DSP assembly code:
jmpi SCI_INT_ADDRESS,(i6)+1
The user may, at will, replacetheinstructionwitha jmpi command to gaincontrolover the SCIinterrupt.
10.14.3DataInt, 0x22
Normally contains the following VS DSP assembly code:
jmpi SDI_INT_ADDRESS,(i6)+1
The user may, atwill,replacetheinstruction with a jmpi commandtogain control over theSDIinterrupt.
10.14.4ModuInt, 0x23
Normally contains the following VS DSP assembly code:
Version 0.91, 2007-02-1269
jmpi MODU_INT_ADDRESS,(i6)+1
The user may, at will, replace the instruction with a jmpi command to gain control over the AD Modulator interrupt.
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
10.14.5TxInt, 0x24
Normally contains the following VS DSP assembly code:
jmpi EMPTY_INT_ADDRESS,(i6)+1
The user may, at will, replace the instruction with a jmpi command to gain control over the UART TX
interrupt.
10.14.6RxInt, 0x25
Normally contains the following VS DSP assembly code:
jmpi RX_INT_ADDRESS,(i6)+1
10. VS1033 REGISTERS
The user may, at will, replace the first instruction with a jmpi command to gain control over the UART
RX interrupt.
10.14.7Timer0Int, 0x26
Normally contains the following VS DSP assembly code:
jmpi EMPTY_INT_ADDRESS,(i6)+1
The user may, at will, replace the first instruction with a jmpi command to gain control over the Timer
0 interrupt.
10.14.8Timer1Int, 0x27
Normally contains the following VS DSP assembly code:
jmpi EMPTY_INT_ADDRESS,(i6)+1
The user may, at will, replace the first instruction with a jmpi command to gain control over the Timer
1 interrupt.
Version 0.91, 2007-02-1270
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
10.14.9UserCodec, 0x0
Normally contains the following VS DSP assembly code:
jr
nop
If the user wants to take control away from the standard decoder, the first instruction should be replaced
with an appropriate j command to user’s own code.
Unless the user is feeding MP3 or WMA data at the same time, the system activates the user program
in less than 1 ms. After this, the user should steal interrupt vectors from the system, and insert user
programs.
10. VS1033 REGISTERS
10.15System Vector Functions
The System Vector Functions are pointers to some functions that the user may call to help implementing
his own applications.
10.15.1WriteIRam(), 0x2
VS DSP C prototype:
void WriteIRam(registeri0 u int16 *addr, registera1 u int16 msW, registera0 u int16 lsW);
This is the preferred way to write to the User Instruction RAM.
10.15.2ReadIRam(), 0x4
VS DSP C prototype:
u int32 ReadIRam(registeri0 u int16 *addr);
This is the preferred way to read from the User Instruction RAM.
A1 contains the MSBs and A0 the LSBs of the result.
10.15.3DataBytes(), 0x6
VS DSP C prototype:
Version 0.91, 2007-02-1271
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
u int16 DataBytes(void);
If the user has taken over the normal operation of the system by switching the pointer in UserCodec
to point to his own code, he may read data from the Data Interface through this and the following two
functions.
This function returns the number of data bytes that can be read.
10.15.4GetDataByte(), 0x8
VS DSP C prototype:
u int16 GetDataByte(void);
Reads and returns one data byte from the Data Interface. This function will wait until there is enough
data in the input buffer. Audio interrupts must be enabled for this function to work.
10. VS1033 REGISTERS
10.15.5GetDataWords(), 0xa
VS DSP C prototype:
void GetDataWords(registeri0 y u int16 *d, register a0 u int16 n);
Read n data byte pairs and copy them in big-endian format (first byte to MSBs) to d. This function will
wait until there is enough data in the input buffer. Audio interrupts must be enabled for this function to
work.
Version 0.91, 2007-02-1272
VLSI
Solution
y
VS1033C
VS1033c PRELIMINARY
11. DOCUMENT VERSION CHANGES
11Document Version Changes
This chapter describes the most important changes to this document.
11.1Version 0.91 for VS1033c, 2007-02-12
• GBUF connection in Connection diagram changed (figure 3 in section 6).
GBUF must have 10Ω and 47 nF to ground.
• Mention of the 8 kHz Phone Application removed. Other features have replaced this code in
VS1033c.
11.2Version 0.9 for VS1033c, 2006-08-15
• EarSpeaker documentation added.
11.3Version 0.8 for VS1033b, 2006-05-19
• Mention of quiet power-off added to feature list.
11.4Version 0.6 for VS1033a, 2006-01-05
• ADPCM recording section added (section 9.4)
11.5Version 0.5, 2005-10-21
• More detailed info about fast forward / rewind.
Version 0.91, 2007-02-1273
VLSI
Solution
y
VS1033c PRELIMINARY
12Contact Information
VLSI Solution Oy
Entrance G, 2nd floor
Hermiankatu 8
FIN-33720 Tampere
FINLAND
Fax: +358-3-3140-8288
Phone: +358-3-3140-8200
Email: sales@vlsi.fi
URL: http://www.vlsi.fi/
VS1033C
12. CONTACT INFORMATION
Version 0.91, 2007-02-1274
Loading...
+ hidden pages
You need points to download manuals.
1 point = 1 manual.
You can buy points or you can get point for every manual you upload.