Hello, I belong to a distributor of Freescale Japan.
Our customer have a question about i.MX6 NAND progamming.
Actually, they are using kobs-ng to program a boot image to NAND in prototype, but they want to use NAND gang programmer for mass production.
In this case, they will dump the NAND data by "nanddump" from a prototype and program it to other NAND devices for mass production with gang programmer.
But if kobs-ng manage bad block, they think taht they cannot use this method.
So would you let me know this NAND programming procedure is OK or NG?
If it is NG, is there any other method?
Best Regards,
Satoshi Shimoda
Solved! Go to Solution.
Hi Satoshi,
By my understanding, kobs-ng manages bad block with using DBBT but it's not mandatory.
If you specify the page address of DBBT is zero, iROM ignores it and uses bad block marker in OOB
to identify bad block.
However, IMHO, I think your method is not good idea. Because the dumped image might contain
bit errors that would be caused by the prototype's chip. It means you will get twice of
bit errors and bad blocks for mass production and that will make to reduce yield rate.
My suggestions are
1) to make whole NAND image(I mean it contains ECC for normal pages and FCB page) by software.
2) to use mfgtools
.
BR,
Nozomi SATO
Hi Satoshi,
By my understanding, kobs-ng manages bad block with using DBBT but it's not mandatory.
If you specify the page address of DBBT is zero, iROM ignores it and uses bad block marker in OOB
to identify bad block.
However, IMHO, I think your method is not good idea. Because the dumped image might contain
bit errors that would be caused by the prototype's chip. It means you will get twice of
bit errors and bad blocks for mass production and that will make to reduce yield rate.
My suggestions are
1) to make whole NAND image(I mean it contains ECC for normal pages and FCB page) by software.
2) to use mfgtools
.
BR,
Nozomi SATO