Peripheral
Configuration Bus
Interrupt and
DMA Requests
16-Bit Shift Register
16-Bit Buffer
GPIO
Control
(all pins)
State
Machine
Clock
Control
SPIx_SIMO
SPIx_SOMI
SPIx_ENA
SPIx_SCS
SPIx_CLK
161
TMS320C6748
www.ti.com
SPRS590G – JUNE 2009 – REVISED JANUARY 2017
Submit Documentation Feedback
Product Folder Links:
TMS320C6748
Peripheral Information and Electrical Specifications
Copyright © 2009–2017, Texas Instruments Incorporated
6.17 Serial Peripheral Interface Ports (SPI0, SPI1)
Figure 6-36
is a block diagram of the SPI module, which is a simple shift register and buffer plus control
logic. Data is written to the shift register before transmission occurs and is read from the buffer at the end
of transmission. The SPI can operate either as a master, in which case, it initiates a transfer and drives
the SPIx_CLK pin, or as a slave. Four clock phase and polarity options are supported as well as many
data formatting options.
Figure 6-36. Block Diagram of SPI Module
The SPI supports 3-, 4-, and 5-pin operation with three basic pins (SPIx_CLK, SPIx_SIMO, and
SPIx_SOMI) and two optional pins (SPIx_SCS, SPIx_ENA).
The optional SPIx_SCS (Slave Chip Select) pin is most useful to enable in slave mode when there are
other slave devices on the same SPI port. The device will only shift data and drive the SPIx_SOMI pin
when SPIx_SCS is held low.
In slave mode, SPIx_ENA is an optional output. The SPIx_ENA output provides the status of the internal
transmit buffer (SPIDAT0/1 registers). In four-pin mode with the enable option, SPIx_ENA is asserted only
when the transmit buffer is full, indicating that the slave is ready to begin another transfer. In five-pin
mode, the SPIx_ENA is additionally qualified by SPIx_SCS being asserted. This allows a single
handshake line to be shared by multiple slaves on the same SPI bus.
In master mode, the SPIx_ENA pin is an optional input and the master can be configured to delay the start
of the next transfer until the slave asserts SPIx_ENA. The addition of this handshake signal simplifies SPI
communications and, on average, increases SPI bus throughput since the master does not need to delay
each transfer long enough to allow for the worst-case latency of the slave device. Instead, each transfer
can begin as soon as both the master and slave have actually serviced the previous SPI transfer.
