Garbz
No longer a newbie, moving up!
- Joined
- Oct 26, 2003
- Messages
- 9,713
- Reaction score
- 203
- Location
- Brisbane, Australia
- Website
- www.auer.garbz.com
- Can others edit my Photos
- Photos NOT OK to edit
So there's quite a few people on this forum who advocate many smaller flash chips rather than one big one. That's all good and fine. But recently I saw something that's got me thinking of the durability of modern EEPROM flash.
So CF cards have been run over by cars, have gone through the wash (I've personally had 2 CF cards, 1 USB stick and 1 SD card go through the wash and one CF card even made it through the drier. Nothing has died on me yet. CF cards have to be one of the most mechanically durable cards on the market save for the hundreds of pins in the socket. But they all use one underlying identical technology. They are all the same kind of flash chip.
SD cards address the flash chip directly using I2C and so do xD cards. The memory card reader needs a controller to decide how to write to the cards. CF cards mimic an ATA interface to the flash and each chip contains its own onboard controller (this is one of the reasons for their huge size and number of pins). You can even get basic adapters that make SD cards look like a UDMA harddisk, and plug into the standard parallel interface.
In the past this underlying flash chip has always had a limited number of writes before they break. Note that nearly every failure case is during writing and not during reading. So a dead card can nearly always be recovered (assuming the flash chip died and not the onboard controller). Early this number was often 100000 re-writes. The parts of the chips that always wore out were the file allocation tables which often get overwritten constantly during normal operation.
Last year I looked at this the number quoted was "up to" 1000000 write operations for an EEPROM Atmel serial EEPROM chip that I used in my thesis. Looking at Microchip and Atmels website shows most newer chips have a change in wording of "at least" 1million write/erase cycles permitted.
One of my more favourite providers of elec engineering toys Dangerous Prototypes decided to test just one of these new chips. By creating a small program that writes verifies and erases the chip over and over again and counting the number of times the verification is successful they are testing just how durable flash is.
It looks like it's going to hit 4million write/erase cycles this week. There's a webcam which you can use to watch the device in action.
Dangerous Prototypes Prototype: Flash_Destroyer
So looks like it may be time to consider physical abuse being the killer of flash rather than having old cards which didn't even outlive the shutter on a camera.
So CF cards have been run over by cars, have gone through the wash (I've personally had 2 CF cards, 1 USB stick and 1 SD card go through the wash and one CF card even made it through the drier. Nothing has died on me yet. CF cards have to be one of the most mechanically durable cards on the market save for the hundreds of pins in the socket. But they all use one underlying identical technology. They are all the same kind of flash chip.
SD cards address the flash chip directly using I2C and so do xD cards. The memory card reader needs a controller to decide how to write to the cards. CF cards mimic an ATA interface to the flash and each chip contains its own onboard controller (this is one of the reasons for their huge size and number of pins). You can even get basic adapters that make SD cards look like a UDMA harddisk, and plug into the standard parallel interface.
In the past this underlying flash chip has always had a limited number of writes before they break. Note that nearly every failure case is during writing and not during reading. So a dead card can nearly always be recovered (assuming the flash chip died and not the onboard controller). Early this number was often 100000 re-writes. The parts of the chips that always wore out were the file allocation tables which often get overwritten constantly during normal operation.
Last year I looked at this the number quoted was "up to" 1000000 write operations for an EEPROM Atmel serial EEPROM chip that I used in my thesis. Looking at Microchip and Atmels website shows most newer chips have a change in wording of "at least" 1million write/erase cycles permitted.
One of my more favourite providers of elec engineering toys Dangerous Prototypes decided to test just one of these new chips. By creating a small program that writes verifies and erases the chip over and over again and counting the number of times the verification is successful they are testing just how durable flash is.
It looks like it's going to hit 4million write/erase cycles this week. There's a webcam which you can use to watch the device in action.
Dangerous Prototypes Prototype: Flash_Destroyer
So looks like it may be time to consider physical abuse being the killer of flash rather than having old cards which didn't even outlive the shutter on a camera.