Philips Semiconductors
Product data
PCK2002
0–300 MHz I
2
C 1:18 clock buffer
2001 Jul 19
6
Consider the command code and the byte count bytes required as the first two bytes of any transfer. The command code is software
programmable via the controller, but will be specified as 0000 0000 in the clock specification. The byte count byte is the number of additional
bytes required to transfer, not counting the command code and byte count bytes. Additionally, the byte count byte is required to be a minimum of
1 byte and a maximum of 32 bytes to satisfy the above requirement.
For example:
Byte count byte
Notes:
MSB
LSB
0000
0000
Not allowed. Must have at least one byte.
0000
0001
Data for functional and frequency select register (currently byte 0 in spec)
0000
0010
Reads first two bytes of data. (byte 0 then byte 1)
0000
0011
Reads first three bytes (byte 0, 1, 2 in order)
0000
0100
Reads first four bytes (byte 0, 1, 2, 3 in order)
0000
0101
Reads first five bytes (byte 0, 1, 2, 3, 4 in order)
0000
0110
Reads first six bytes (byte 0, 1, 2, 3, 4, 5 in order)
0000
0111
Reads first seven bytes (byte 0, 1, 2, 3, 4, 5, 6 in order)
0010
0000
Max byte count supported = 32
A transfer is considered valid after the acknowledge bit corresponding to the byte count is read by the controller. The serial controller interface
can be simplified by discarding the information in both the command code and the byte count bytes and simply reading all the bytes that are
sent to the clock driver after being addressed by the controller. It is expected that the controller will not provide more bytes than the clock driver
can handle. A clock vendor may choose to discard any number of bytes that exceed the defined byte count.
8) Clock stretching: The clock device must not hold/stretch the SCLOCK or SDATA lines low for more than 10 ms. Clock stretching is
discouraged and should only be used as a last resort. Stretching the clock/data lines for longer than this time puts the device in an error/time-out
mode and may not be supported in all platforms. It is assumed that all data transfers can be completed as specified without the use of
clock/data stretching.
9) General Call: It is assumed that the clock driver will not have to respond to the “general call.”
10) Electrical Characteristics: All electrical characteristics must meet the standard mode specifications found in section 15 of the I
2
C
specification.
a) Pull-Up Resistors: Any internal resistors pull-ups on the SDATA and SCLOCK inputs must be stated in the individual datasheet. The use of
internal pull-ups on these pins of below 100 k
is discouraged. Assume that the board designer will use a single external pull-up resistor for
each line and that these values are in the 5–6 k
range. Assume one I
2
C device per DIMM (serial presence detect), one I
2
C controller, one
clock driver plus one/two more I
2
C devices on the platform for capacitive loading purposes.
(b) Input Glitch Filters: Only fast mode I
2
C devices require input glitch filters to suppress bus noise. The clock driver is specified as a standard
mode device and is not required to support this feature.
11) PWR DWN: If a clock driver is placed in PWR DWN mode, the SDATA and SCLK inputs must be Tri-Stated and the device must retain all
programming information. I
dd
current due to the I
2
C circuitry must be characterized and in the data sheet.
For specific I
2
C information consult the Philips I
2
C Peripherals Data Handbook IC12 (1997).