AM3505 NAND Flash Interface

Hello,

You can connect an 8b NAND device to a single CS, but not 2 8b NANDs to a single CS.  If you want to use 2 8b NAND devices you will need to use 2 CS. 

The GPMC does not provide the A0 byte address line required to random-byte addressable 8b interface, so for 8b NAND devices you will need to connect the 8b data to the gpmc_d[7:0] lines and the data will be byte aliased. 

Please see the GPMC subsystem secion of the Technical Reference manual for more info.  Also, below are a couple of good wiki links for the GPMC:

http://processors.wiki.ti.com/index.php/AM3517/05_GPMC_Subsystem

http://processors.wiki.ti.com/index.php/Tips_for_configuring_OMAP35x_and_AM-DM37x_GPMC_registers

Hi Jeff,

So if I am to understand correctly.  You can only have 8b R/W access even if you have 2 8b NANDs because by using 2 CS's you will effectively double your memory.  I was hoping you could use 2 NAND devices and have 16b R/W access to the AM3517/05 basically making it look like 1 16b instead of 2 8b with independent access. 

The only reason this crossed my mind is that there is only 1 (Micron) mass market supplier for 16b NAND Flash.  Whereas, there are at least 3 for 8b NAND Flash.  I guess we'll try 8b and 16b NAND on our prototypes to see what the application implications will be between 8b and 16b R/W access.

hello Davey,

is there a particular reason you need the 16b interface? 

The reason that you cannot connect 2 NANDs to one CS is because you would then be sending the same command to both NANDs for the read/write processes, so both NANDs would try to execute.  Since NAND is not a random access device,  you should be able to get the memory depth you need with a single NAND device.

Jeff,

I had a16b NAND Flash in my initial schematic because all of the AM3517 eval boards I'd seen had them.  After doing some research I was very uncomfortable using a sole source (Micron) for 16b NAND.   Whereas 8b NAND Flash is availabe through distribution by at least 3 suppliers.  Trying to make sure I didn't have any avoidable gotcha's on the prototyes I decided to poll our firmware group about the change.  The one concern was giving up 16b access, basically doubling R/W access times by going to 8b.  In application we are going to make many R/W accesses.  

As I said previously I will populate both 8b and 16b NANDs on the prototypes and this difference (access times) can be tested.

Davey,

Ok, sounds good.  Another thing to consider is the fact that Spansion has recently announced that they will enter the NAND flash market and they will have both x8 and x16 devices like Micron.  Spansion has good distribution like Micron and they plan on having very long life time parts, so you should be able to have 2 sources for a x16 devices.