2.5.4. Address and Command Datapath
For LPDDR2, CA is four bits wide on the AFI interface. The least significant bit of CA is captured by a negative-edge triggered flop, and the most-significant bit of CA is captured by a positive-edge triggered flop, before multiplexing to provide maximum setup and hold margins for the AFI clock-to-MEM clock data transfer.
For DDR2/DDR3, the full-rate register and MUX architecture is similar to LPDDR2, but both phases are driven with the same signal. The DDR3 address and command signal is not clocked by the write/ADC clock as with LPDDR2, but by the inverted MEM clock, for better address and command margin.
Because the memory controller can drive both phases of cs_n in half-rate, the signal is fully exposed to the AFI side.
Did you find the information on this page useful?