You have to register before you can post. To keep this forum out of spammers, every registration is manually approved.
If you know answers on any questions on this forum, please feel free to answer them. (PS: I try to answer at least once a week or when possible, - Robert)
The picture is for DDR3, but it is similar for DDR2. Check out especially the DATA BYTES (yellow and green block) - you need to have your databanks length matched correctly. And because your UDM and LDM signals were swapped, I am just pointing out, that they may be length matched according to different data group.
Did you match all the data signals (data, strobe, mask) to the same length? Or you length match them based on banks?
I am asking as I am trying to understand if the signals are correctly split into databanks.
How did you swap the pins, in your FPGA code or on the board? Also, how the length matching is done (when you swapped the UDM & LDM, is the data bank length matching correct)?
Yes sir. with pin swapping......fitter error is removed. Now i run Memory test through NIOS processor. We are using Altera CycloneIII 484 BGA FPGA. But the test fails at the inital address. Can u have any experiencing with that?
I have assigned FPGA Bank 7 for DDRII_D0-D7, DDRII_LDQS, DDRII_UDQS, DDRII_UDM & DDRII_LDM..........& Bank 8 for DDRII_A0-A12, CAS, RAS,WE,CS,CKE,CK_P,CK_N,ODT & Bank Addresses.
How are you connecting them to the memory chip and how did you create the BANK groups (e.g. BANK0: DDR1D0-7, DDR1_LDM, DDR1_LDQS)? UDM is upper byte (8-15), LDM is lower byte (0-7). From the screenshot which you attached it looks to me, that you may have UDM and LDM swapped - but I would need to see the chip connection.
Hmm, only what I can think about from schematic/layout point of view is if the mask pins are correctly grouped with associated data bytes ... I am not sure about implementation in FPGA code.
Leave a comment: