CY7C68013A/CY7C68014A
CY7C68015A/CY7C68016A
Document #: 38-08032 Rev. *K
Page 11 of 60
3.13
External FIFO Interface
3.13.1
Architecture
The FX2LP slave FIFO architecture has eight 512-byte blocks
in the endpoint RAM that directly serve as FIFO memories,
and are controlled by FIFO control signals (such as IFCLK,
SLCS#, SLRD, SLWR, SLOE, PKTEND, and flags).
In operation, some of the eight RAM blocks fill or empty from
the SIE, while the others are connected to the I/O transfer
logic. The transfer logic takes two forms, the GPIF for internally
generated control signals, or the slave FIFO interface for
externally controlled transfers.
3.13.2
Master/Slave Control Signals
The FX2LP endpoint FIFOS are implemented as eight physi-
cally distinct 256x16 RAM blocks. The 8051/SIE can switch
any of the RAM blocks between two domains, the USB (SIE)
domain and the 8051-I/O Unit domain. This switching is done
virtually instantaneously, giving essentially zero transfer time
between “USB FIFOS” and “Slave FIFOS.” Since they are
physically the same memory, no bytes are actually transferred
between buffers.
At any given time, some RAM blocks are filling/emptying with
USB data under SIE control, while other RAM blocks are
available to the 8051 and/or the I/O control unit. The RAM
blocks operate as single-port in the USB domain, and
dual-port in the 8051-I/O domain. The blocks can be
configured as single, double, triple, or quad buffered as previ-
ously shown.
The I/O control unit implements either an internal-master (M
for master) or external-master (S for Slave) interface.
In
Master
(M)
mode,
the
GPIF
internally
controls
FIFOADR[1..0] to select a FIFO. The RDY pins (two in the
56-pin package, six in the 100-pin and 128-pin packages) can
be used as flag inputs from an external FIFO or other logic if
desired. The GPIF can be run from either an internally derived
clock or externally supplied clock (IFCLK), at a rate that
transfers data up to 96 Megabytes/s (48-MHz IFCLK with
16-bit interface).
In Slave (S) mode, the FX2LP accepts either an internally
derived clock or externally supplied clock (IFCLK, max.
frequency 48 MHz) and SLCS#, SLRD, SLWR, SLOE,
PKTEND signals from external logic. When using an external
IFCLK, the external clock must be present before switching to
the external clock with the IFCLKSRC bit. Each endpoint can
individually be selected for byte or word operation by an
internal configuration bit, and a Slave FIFO Output Enable
signal SLOE enables data of the selected width. External logic
must insure that the output enable signal is inactive when
writing data to a slave FIFO. The slave interface can also
operate asynchronously, where the SLRD and SLWR signals
act directly as strobes, rather than a clock qualifier as in
synchronous mode. The signals SLRD, SLWR, SLOE and
PKTEND are gated by the signal SLCS#.
3.13.3
GPIF and FIFO Clock Rates
An 8051 register bit selects one of two frequencies for the
internally supplied interface clock: 30 MHz and 48 MHz. Alter-
natively, an externally supplied clock of 5 MHz–48 MHz
feeding the IFCLK pin can be used as the interface clock.
IFCLK can be configured to function as an output clock when
the GPIF and FIFOs are internally clocked. An output enable
bit in the IFCONFIG register turns this clock output off, if
desired. Another bit within the IFCONFIG register will invert
the IFCLK signal whether internally or externally sourced.
3.14
GPIF
The GPIF is a flexible 8- or 16-bit parallel interface driven by a
user-programmable finite state machine. It allows the
CY7C68013A/15A to perform local bus mastering, and can
implement a wide variety of protocols such as ATA interface,
printer parallel port, and Utopia.
The GPIF has six programmable control outputs (CTL), nine
address outputs (GPIFADRx), and six general-purpose ready
inputs (RDY). The data bus width can be 8 or 16 bits. Each
GPIF vector defines the state of the control outputs, and deter-
mines what state a ready input (or multiple inputs) must be
before proceeding. The GPIF vector can be programmed to
advance a FIFO to the next data value, advance an address,
etc. A sequence of the GPIF vectors make up a single
waveform that will be executed to perform the desired data
move between the FX2LP and the external device.
3.14.1
Six Control OUT Signals
The 100- and 128-pin packages bring out all six Control Output
pins (CTL0-CTL5). The 8051 programs the GPIF unit to define
the CTL waveforms. The 56-pin package brings out three of
these signals, CTL0–CTL2. CTLx waveform edges can be
programmed to make transitions as fast as once per clock
(20.8 ns using a 48-MHz clock).
3.14.2
Six Ready IN Signals
The 100- and 128-pin packages bring out all six Ready inputs
(RDY0–RDY5). The 8051 programs the GPIF unit to test the
RDY pins for GPIF branching. The 56-pin package brings out
two of these signals, RDY0–1.
3.14.3
Nine GPIF Address OUT Signals
Nine GPIF address lines are available in the 100- and 128-pin
packages, GPIFADR[8..0]. The GPIF address lines allow
indexing through up to a 512-byte block of RAM. If more
address lines are needed, I/O port pins can be used.
3.14.4
Long Transfer Mode
In master mode, the 8051 appropriately sets GPIF transaction
count registers (GPIFTCB3, GPIFTCB2, GPIFTCB1, or
GPIFTCB0) for unattended transfers of up to 232 transactions.
The GPIF automatically throttles data flow to prevent under or
overflow until the full number of requested transactions
complete. The GPIF decrements the value in these registers
to represent the current status of the transaction.
[+] Feedback
[+] Feedback
English ▼
English
Chinese
German
Japanese
Russian
Korean
Spanish
French
Italian
Portuguese
Polish
Vietnamese
Indian
Mexican
British
New Zealand
