Post

2 followers Follow
0
Avatar

SPI Memory for N6000

My company is evaluating the Neuron 6000. But, the use of SPI Flash memory does not give us enough read/write/erase cycles to be used as persistent memory. There is a SPI MRAM chip having the same footprint and uses the same commands except it doesn't have a Chip Erase or 4KB Sector Erase. Assuming that this MRAM chip ignores these commands, I wonder if the N6000 will have problems using this memory. Or, will Echelon provide support for chips like this in the near future? The chip in question is the MR20H40 by Everspin Technologies.

MrGuru

Official comment

Avatar

The chip erase is only ever used as a fallback if there’s no system image found by the bootloader, sending it back to V19.
The sector erase would result in just a quick return since the chip would reply immediately with the status bit 0 = 0 (at least I’m assuming that’s the case).
The kicker is the write protect. The bootloader will try and set BP bits to something based on the size of the part reported by the JEDEC ID command – which this part doesn’t seem to support. So we don’t know how to set up two important variables based on the chip capacity.
So in order to support this we’d probably have to make an exception for it in the bootloader code, make an assumption about the size, and not write protect the chip since we require lower-half write protection. This leaves the bootloader and system image somewhat vulnerable to corruption.

Note that the 6000 does have the concept of user-defined persistent data that gets wear-leveled through FLASH based on blob size and allocated size. This is documented in the NC include file “Logger.h”.

Kamal Mehta
Comment actions Permalink

Please sign in to leave a comment.

1 comment