SLUFD512MU1UI-B STEC, SLUFD512MU1UI-B Datasheet - Page 9

• Manufacturer Part Number: SLUFD512MU1UI-B
• Manufacturer: STEC
• Type: USB Flash Driver

Specifications of SLUFD512MU1UI-B

• Density: 512MByte
• Operating Supply Voltage (typ): 5V
• Operating Temperature (min): -40C
• Operating Temperature (max): 85C
• Package Type: Not Required
• Pin Count: 4
• Operating Temperature Classification: Industrial
• Operating Supply Voltage (min): 4.75V
• Operating Supply Voltage (max): 5.25V
• Programmable: Yes
• Lead Free Status / RoHS Status: Compliant

SLUFDxxx(M/G)U1U(I)-y
Industrial Grade USB Flash Drive
Datasheet
3.2.2 Wear Leveling
The SLC NAND flash devices that are being used in the USB Flash Drive are guaranteed for 100,000
Write/Erase cycles per block. This means that after approximately 100,000 erase cycles, the erase block
has a higher probability for errors than the error rate that is typical to the flash. While 100,000 write/erase
cycles may be good for consumer data storage, such as digital cameras, MP3 players, etc., it is not
sufficient for industrial and embedded applications where data is constantly written to the device and long
product life is required.
For example, operating systems that use a file system, will update the File Allocation Table (FAT) every
time a write is done to the device. Without any wear leveling in place, the area on the flash where the FAT
table is located would wear out faster than other areas, reducing the lifetime of the entire flash device.
To overcome this limitation, the flash management algorithm needs to make sure that each block in the
device ages, i.e. is “worn out”, at the same rate. The built-in wear leveling scheme makes sure that with
every write to the flash, the youngest block is used. This ensures that the full flash media is used
uniformly, so that one area of the flash will not reach the endurance limits prematurely before other areas.
The implemented wear leveling algorithm ensures a minimum of 2 million write/erase cycles for the entire
flash media.
3.2.3 Error Correction/Detection
The USB 2.0 controller implements an advanced Error Correction/Detection scheme, based on the Reed-
Solomon algorithm. The ECC engine can detect up to 5 bytes and correct up to 4 bytes per 512 bytes
(symbol based). To ensure the fastest performance, both detection and correction are done on-the-fly, in
hardware only.
Each time the host application writes a sector of 512 bytes to the Industrial Grade UFD, a unique ECC
signature is created by the ECC engine and written together with the data to the flash. When the data is
read back by the host, the ECC engine creates again a unique ECC signature. It will then compare the
original written signature with the newly created signature, and sets an error flag if the two signatures are
not the same. Correction of the data is done on-the-fly when the error flag is set, and the data presented
to the host will be the same as the original written data. This powerful Error Correction/Detection scheme
14
results in an overall error rate of less than 1 in 10
bits, read.
3.2.4 Power Failure Management
The embedded flash management software uses algorithms that ensure data integrity, even during power
failures. After each write, a verify write flag is set in the extra area of the flash page, while a dismount flag
is set for the whole device during regular power-down. When a power failure occurs, both the verify write
and the dismount flag will not be set, indicating a power failure during a write or erase operation. Upon the
next power-up, the mapping tables will be reconstructed from the information stored in the flash memory,
and the last version of the sector with a correct verify write flag will be used.
This mechanism ensures complete data integrity. When writing, copying, or erasing the flash device, the
data format remains valid at all intermediate stages. Old data is never erased until the verify write flag of
the new sector has been set. Therefore, a data sector cannot exist in a partially written state. The
operation is either successfully completed (verify write flag is set) and the new data is valid, or the
operation has not yet been completed or has failed, in which case the old sector contents remain valid.
61000-04536-108, April 2008
9