New Hardware Revision
From nookDevs
It appears Barnes and Noble has introduced a new hardware revision or another change which is causing units to become bricks after running the official Barnes and Noble 1.0.0 firmware image. These units can be identified by the 10031 serial #.
A label on the bottom of the nook box will indicate what version your nook is.
If the serial # begins with 10031, the unit MAY OR MAY NOT be rootable. If the serial # is greater than 10031, it almost certainly NOT rootable.
If the device comes preloaded with firmware version 1.4.0 or earlier, it is likely rootable, regardless of serial number. If it comes preloaded with 1.4.1 or greater, it is almost certainly NOT rootable.
Teardown of the new nook hardware
The Main difference of the new nook hardware is the removal of the SD card and holder. This has been replaced by a single chip solution most likely to save a few dollars in production.
The Chip on the nook I tore down was a KLM261DEHE-B101
Which is a standard Samsung moviNAND chip, see
This is wired as an MMC chip using a 4bit data bus to the host controller.
The data lines are wired with 4 10k resistors for pullup next to the memory chip R153-R156.
The other lines needed to read the chip are the clock and command lines. I dont know where these end up.
However since it is most likely driven directly by the micro controller, it may be risky to drive the mmc card with an external controller and not damage the CPU on the nook. Perhaps one could connect a standard mmc reader to the lines and then power up the mmc chip and hold the main processor in reset, thus allowing the contents of the mmc card to be dumped or restored using a standard reader. (Unless the mmc chip is also locked with a password)
From the mmc 4.3 controller specification:
Product moviNAND
Size 2gigbyte
Composition nand1+micro
Density = 16gig MLC
VCC VCCQ MODE
3.3V(2.5V~3.6V) 1.8V(1.65V~1.95V) Dual
3.3V(2.5V ~ 3.6V)
Unspecified controller
6th generation
Package = FBGA
Revesion = EVT1
Basically it should be identical to standard mmc flash chip. Thus it should not be much different compared to the original hardware revision.
However from manufacturing it would not make sense to have to program even a boot loader using the test jig, so there might be a way to load the OS onto the mmc card in an easier way. New theory is there is a way to force the cpu to boot from the external SD card instead of the internal MMC card. Thus if we could write a bootloader for the device that ran from the SD card we could then load the kernel and environment.
