EEPROMs 24LC256 24LC512 24C512B 24C512C and wearout issues
Cluso99
Posts: 18,069
Looking at the fact that DigiKey announced the 24C512B was becoming non-stocked I thought I had to look into what was happening.
First the B version is being replaced with the C version and as yet DigiKey do not stock it.
Second, I noticed that the 512 C version (and realised the B does the same) is 128 byte blocks whereas the 256 is 64 byte blocks. First impression was will it now work with the propeller. Yes, because the newer versions also allow byte writing. So, it is legal to write 64 bytes at a time. But, if you write two consecutive 64 byte blocks will that actually perform two writes to the 512 B or C. I think so, so you have just halved your number of writes. However, you have 1,000,000 cycles!! So, the software would have to determine what EEPROM is fitted to avoid this. The prop bootloader cannot be changed, but a driver could.
Also, it brings up another point. Because it is possible to program a byte at a time, this fact could be useful for someone datalogging where they do not want to lose any data on power down (i.e. they do not want to buffer data). As long as they realise they are using 64 or 128 writes to achieve this, so it will wear out quicker.
As for the difference between LC and C (24LC256), it seems that Atmel are changing their part numbers to remove the LC. I could not find any reference on their website to the apparently older 24LC512.
Just thought I would let you folks know what I found.
First the B version is being replaced with the C version and as yet DigiKey do not stock it.
Second, I noticed that the 512 C version (and realised the B does the same) is 128 byte blocks whereas the 256 is 64 byte blocks. First impression was will it now work with the propeller. Yes, because the newer versions also allow byte writing. So, it is legal to write 64 bytes at a time. But, if you write two consecutive 64 byte blocks will that actually perform two writes to the 512 B or C. I think so, so you have just halved your number of writes. However, you have 1,000,000 cycles!! So, the software would have to determine what EEPROM is fitted to avoid this. The prop bootloader cannot be changed, but a driver could.
Also, it brings up another point. Because it is possible to program a byte at a time, this fact could be useful for someone datalogging where they do not want to lose any data on power down (i.e. they do not want to buffer data). As long as they realise they are using 64 or 128 writes to achieve this, so it will wear out quicker.
As for the difference between LC and C (24LC256), it seems that Atmel are changing their part numbers to remove the LC. I could not find any reference on their website to the apparently older 24LC512.
Just thought I would let you folks know what I found.