Hi,
Everyone, when I configure the LPDDR2(Hynix 4Gb LPDDR2-S4B, H9TP17A8JDACNR-KGM) with IMX53(routing all signals the same length), some ESDCTL register parameters really confuse me especially in read process.
1, How the ESDCTL_ESDMISC RALAT field influence the read process, or does it have relationship with tCL filed in ESDCTL_ESDCFG0?
2, Even slow the frequency to 50MHz, using MRR command still could not get the right device id in ESDCTL_ESDMRR( all the field filled with 1), but I can use oscilloscope to get the right device ID on the ddr bus. So the receiving timing configure is wrong in IMX53, but how to?
3.FRC_MSR field in ESDCTL_MUR can or can't use in LPDDR2? For CA_DL_ABS_OFFSET field in ESDCTL_PDCMPR2, I think FRC_MSR have influence with LPDDR2, but the datasheet says only for ddr2/ddr3, not for lpddr2....confused....
4. I can use the ESDCTL to right configue all the timing need in LPDDR2, like BL8、RL6/WL3 etc.... and can get the right response by the oscilloscope, so
I think the write process (including the MRW command ) is right, because of the oscilloscope can show the timing, but the receive is a big problem.
I really don't know the read mechanism in IMX53, the datasheet do not make it clearly especially the parameter above, so can I get some help from this community?
best regards
from rain.
Solved! Go to Solution.
As for RALAT :
While internal Rx FIFO data-in are sampled by the DQS, the FIFO data-out have no
real time indication when data are ready in the FIFO. Internal data-out are
therefore delayed by a preset value of DDR device CAS latency plus the RALAT
number of ESDCTL logic cycles. RALAT should be set to a big (enough) value to
support the additional FIFO stage & board latencies. Optimal values for RALAT
are board dependent and should be found experimentally.
As for RALAT :
While internal Rx FIFO data-in are sampled by the DQS, the FIFO data-out have no
real time indication when data are ready in the FIFO. Internal data-out are
therefore delayed by a preset value of DDR device CAS latency plus the RALAT
number of ESDCTL logic cycles. RALAT should be set to a big (enough) value to
support the additional FIFO stage & board latencies. Optimal values for RALAT
are board dependent and should be found experimentally.
But how about tDQSCK be treated in iMX53? In RALAT field it says have extra 2 cycles for LPDDR2.