• High-speed analog-to-digital conversion for video signal
digitizing in 4 :1:1 format
• 100 Hz improved definition TV for all formats
(4/3, 16/9, 14/9 etc.).
GENERAL DESCRIPTION
The TDA8753A is a monolithic CMOS 8-bit video
low-power analog-to-digital conversion interface for YUV
signals. It converts the YUV analog input signal into 8-bit
binary coded digital words in format 4 :1:1 at a sampling
rate of 20 MHz. All analog signal inputs are clamped.
The device includes a digital sample rate converter for
variable compression with a factor 1 to 2.
Y71Y data output, bit 7 (MSB)
Y62Y data output, bit 6
Y53Y data output, bit 5
Y44Y data output, bit 4
Y35Y data output, bit 3
Y26Y data output, bit 2
Y17Y data output, bit 1
Y08Y data output, bit 0 (LSB)
V
V
DDD2
SSD2
9digital supply voltage 2, (+5 V)
10digital ground 2
U111U data output, bit 1 (n)
U012U data output, bit 0 (n − 1)
V113V data output, bit 1 (n)
V014V data output, bit 0 (n − 1)
V
The TDA8753 implements 3 independent CMOS 8-bit
analog-to-digital converters. The converters use a
multi-step approach with offset compensated
comparators.
Clamping
An internal clamping circuit is provided in each of the
3 analog channels. The analog pins INY, INV and INU are
switched to on-chip clamping levels during an active pulse
on the clamp input CLP.The clamping level in the
Y channel is code level 16. The clamping level in the U/V
channel is code level 128 (output code 0 in the
2's complement description) see Tables 3 and 4.
Sample rate converter
A sample rate converter is integrated in the TDA8753A to
facilitate programming of the horizontal aspect ratio which
can be varied from a factor 1 to 2.
This includes conversion from 16/9 to 14/9 and 4/3. In the
U/V channel a linear interpolation is sufficient because of
the four times oversampling.
The TDA8753A has three addressable control registers
which can be loaded via the signals UPDA and UPCL.
The format of this bus is fixed according to mode 0 of the
8051 family UART at 1 Mbaud (8 bits are transmitted, LSB
first).
Serial interface protocol
P
OWER-ON STATE
When powered up the SIO is in an unknown state and all
data in the registers is random. When signals are applied
to UPCL and UPDA in this state, the behaviour is
unpredictable. The only way to exit from this state to a
known state is apply a V50 signal to the TDA8753A.
I
NITIALIZATION STATE
From power-on or any other state, the INIT state is entered
(at the latest) one TDA8753A clock period after the end of
the V50 HIGH state. In this state the F0, F1 and F2
TDA8753A registers are loaded with the values that are in
the corresponding line buffers BF0, BF1 and BF2. The first
time V50 is issued after power-on, this data is unknown.
After a rising UPCL edge has been detected, the address
reception state is entered.
Discrete time oscillator (DTO)
A discrete time oscillator is used to calculate for every
sample of the phase delay that is needed for a given
compression factor.
Serial interface (SIO)
All controls are sent to the TDA8753A via a serial
microprocessor interface. Data from this interface will be
made active at the vertical input pulse V50.
handbook, halfpage
11110010
first data bit
of data value
for address F2
register
last address
bit received
(in this example address received is F2 hex)
DDRESS RECEPTION STATE
A
Bits are counted at each rising UPCL edge. The next 8 bits
received on UPDA line are considered as address bits.
The address reception is illustrated in Fig.3.
incoming stream
first bit
received
MBE426
1996 Jan 125
Fig.3 Address reception.
Philips SemiconductorsProduct specification
YUV 8-bit analog-to-digital interfaceTDA8753A
The TDA8753A registers have address F0, F1 and F2
hexadecimal notation. Whenever the received address
(decoded on the first 8 bits received) is one of these, the
event is recorded in such a way that the next data received
by the TDA8753A will be captured in the line
buffer BF0, BF1 and BF2 respectively.
When 8 bits have been received, the data reception state
is entered. The address reception state can also be exited
at any time when V50 goes HIGH. The F0, F1 and F2
registers may not be loaded properly if there is some
activity in progress on the incoming line.
handbook, halfpage
110XXXXX
first bit of next
address stream
Data value is F2 0:2 = 110(DEL 0:2 )
incoming stream
D
ATA RECEPTION STATE
The next 8 bits are considered as data bits according to
the format of Fig.4.
When 8 data bits have been received, the data is recorded
in the BF0, BF1 or BF2 line buffers if the previous address
recorded was F0 hex, F1 hex or F2 hex respectively.
The bit count is then reset to zero and the address
reception state is entered. This state may be ended any
time when V50 goes HIGH but in that condition F0, F1 and
F2 registers may not be loaded properly.
last address
bit received
first data bit of value
(e.g. for address F2 register)
MBE427
Fig.4 Data reception.
Table 1 Data allocation
ADDRESSPARAMETERFUNCTION
F0HCFcompression factor value will be (1 + cf/255)
NUMBER
OF BITS
87:0
POSITION
which results in a range from 1 to 2
F1HUV_CORINGcoring definition in U and V channels; see Table 521:0
UV_FILTER_TYPEnotch filter selection in U and V channels