A thumb drive, also known as a USB flash drive, stores data using flash memory. Flash memory is a type of electronic memory that can be erased and reprogrammed. It gets its name from the fact that data can be flashed, or written and erased, rapidly.
How does flash memory work?
Flash memory is made up of semiconductor-based memory chips that store data electronically. Each memory chip contains thousands or millions of tiny transistors or gates. These gates hold electric charges that represent bits of data – either a 1 or a 0.
The gates are separated into blocks or pages. To write data, a programming voltage is applied to the gates, forcing electrons to tunnel through an oxide layer and get trapped on the other side. This programs the gates to a charged “on” state to represent a 1 or an uncharged “off” state to represent a 0. To erase the data, a reverse voltage is applied to discharge the gates back to their unprogrammed state.
What type of flash memory is used in USB drives?
Most USB flash drives today utilize a type of flash memory called NAND flash. NAND flash gets its name from the NAND logic gates that make up the core memory cell. Here’s a quick overview of how NAND flash works:
- Each memory cell is made up of a floating gate transistor.
- The floating gate can hold a charge, representing a 1, or no charge, representing a 0.
- To read the cell, a voltage is applied to detect if it is charged or not.
- To write to the cell, a high voltage is applied to inject electrons into the floating gate, charging it to a 1 state.
- To erase the cell, a reverse voltage is applied to drain the electrons, discharging the floating gate back to a 0 state.
- Cells are arranged into pages and blocks. Entire blocks are erased together in NAND flash.
Compared to earlier forms of flash memory, NAND flash offers higher densities and lower cost per bit. This has allowed it to become the dominant flash memory type used for solid-state storage like USB drives.
How is data organized on a thumb drive?
Data is organized and managed on a USB flash drive using a flash translation layer (FTL). Key responsibilities of the FTL include:
- Page mapping – Logically addressing data at the page level
- Block management – Erasing and rewriting blocks as needed
- Wear leveling – Ensuring even wear across all blocks
- Bad block management – Detecting and mapping out any bad blocks
- Read/write algorithms – Algorithms to handle read and write requests
- Garbage collection – Reclaiming unused pages and consolidating data
The FTL makes the flash memory appear like a standard block storage device to the host computer. When you save a file to a USB drive, the FTL determines where to write the data and manages all the underlying complexity of flash memory programming and erasing.
What is the architecture of a typical USB flash drive?
Here are the key internal components that make up a USB flash drive:
- USB connector – Connects the drive to a host computer.
- USB mass storage controller – Implements USB protocol and transfers data between the host and flash memory.
- Flash memory chip(s) – Stores data electronically in gates/cells organized into pages and blocks.
- Crystal oscillator – Generates clock signals for the controller.
- LED light – Indicates activity status.
- Resistor – Helps pull-up USB data lines.
- Controller firmware – Software that runs on the controller to manage the FTL, USB protocol, security features, etc.
The USB mass storage controller and flash memory chip(s) are the core components. The controller bridges the communication between the host and flash memory. Different USB drive capacities are achieved by varying how many flash memory chips are included.
How is data written to a USB flash drive?
Here is a general overview of how data gets written to a flash drive:
- The host computer sends a write command over USB along with the data to store.
- The USB controller receives the data, breaks it into pages, and begins filling up its internal buffer.
- When the buffer is full, the controller starts programming pages into the flash memory array.
- The flash memory is organized into blocks containing multiple pages. Pages are written sequentially within a block.
- If the block being written to is full, a new block is allocated and writing continues.
- The FTL maintains metadata on the logical to physical page mapping as data is written.
- Once all the data is buffered and written, the drive indicates the write is complete.
The controller hardware manages the entire process of buffering data from the host and then programming it to the right locations in flash memory. The FTL software controls the logical to physical mapping.
Write cycle
Here is a more detailed look at the series of steps to write a single page of data to flash memory:
- The selected page is erased by discharging all gates to a 0 state.
- The write voltage is applied to the gates, injecting electrons to charge them to represent the data bits.
- The data is read back and compared to the input data to verify it programmed correctly.
- If verification fails, the erase and write steps are repeated.
This erase-write-verify cycle ensures the data is reliably programmed to the target threshold voltage levels. The FTL manages the mapping of logical pages from the host to physical page locations.
How is data read from a USB flash drive?
Here are the basic steps to read data from a flash drive:
- The host sends a read command over USB indicating the logical address to read from.
- The USB controller looks up the physical page location via the FTL mapping table.
- A low voltage is applied to the cell gates in the target page and the presence or absence of charge is sensed.
- The page data is read into the controller’s internal buffer.
- The controller transfers the data from its buffer back to the host over USB.
The controller handles all the low-level flash memory read operations and data transfers. The FTL provides the mapping between logical and physical addresses invisible to the host. The host simply issues read and write requests by logical address.
How is data erased from a USB flash drive?
Flash memory must be erased prior to being rewritten. Here is the typical process to erase data:
- The host sends a delete command or a rewrite command that requires erasing.
- The FTL marks any referenced logical pages as invalid.
- The physical block containing the pages is selected for erasing.
- An erase voltage is applied to discharge all the cell gates in the block to the 0 state.
- The block is now erased and ready to be rewritten.
- Garbage collection and wear leveling routines eventually reuse the erased block.
The erase process resets an entire block of flash memory in preparation for rewriting. The FTL manages logical to physical mapping and block selection. The controller hardware controls the erase voltage pulses.
What is wear leveling and garbage collection?
Wear leveling and garbage collection are two important processes the FTL uses to optimize performance and flash memory lifespan:
Wear leveling
Wear leveling aims to spread out erases evenly so that no single block prematurely fails from excessive program/erase cycles relative to other blocks. Tactics include:
- Dynamic block rotation – Regularly exchanging static and dynamic data blocks
- Start-gap – Ensuring even block usage by writing to blocks with the lowest erase counts
Garbage collection
Garbage collection consolidates data to free up space by:
- Identifying invalid/stale pages not being referenced
- Copying valid pages from a block into a new block
- Erasing the original block to reclaim unused space
These processes enable sustained performance and lifespan from flash memory devices.
What gives USB flash drives fast performance?
Here are some of the key factors that enable fast USB flash drive performance:
- High degree of parallelism – Multiple NAND flash chips and channels can be read/written simultaneously.
- Advanced controller – Uses techniques like native command queuing and out-of-order execution.
- No in-place writes – Out-of-place writes avoid erase times.
- DRAM cache – On-board DRAM provides low latency buffering.
- Optimized algorithms – Page mapping, wear leveling, garbage collection, etc. are optimized for speed.
USB 3.2 and USB4 interfaces also allow for much higher interface bandwidth compared to older USB 2.0 drives. Combined, these factors enable modern high-speed flash drives.
How reliable is data storage on USB flash drives?
USB flash drives use reliable high-endurance NAND flash memory designed for thousands to hundreds of thousands of program/erase cycles. However, flash memory does have limitations:
- Block/chip failures – A block or entire chip can wear out and fail.
- Read disturbs – Repeated reads can cause bit errors.
- Write disturbs – Writing one cell can affect nearby cells.
To improve reliability:
- ECC checks – Error correcting codes detect and repair bit errors.
- Data integrity testing – Controllers verify successful writes.
- Wear leveling – Evens out wear to avoid failure weak spots.
- Bad block management – Maps out failed or weak blocks.
Overall USB flash drive reliability is very good, with typical failure rates under 1-2% per year with everyday consumer use. But higher endurance drives designed for intensive workloads can support full drive writes per day for 5 years or more.
What are the capacity limits of USB flash drives?
Early USB flash drives only had capacities up to a few gigabytes. But capacities have grown tremendously over the years. Here are some of the key factors influencing capacity limits:
- NAND lithography process – Smaller process dimensions allow for higher densities.
- Bits per cell – MLC and TLC (2 and 3 bits per cell) provide higher densities than SLC (1 bit per cell).
- Die stacking – Manufacturers can stack multiple dies in a single package for more capacity.
- Number of chips – More flash memory dies can be added for higher capacities but with tradeoffs.
Current maximum capacities include:
| USB Drive Type | Maximum Capacity |
|---|---|
| Consumer USB 2.0 | 1 TB |
| Consumer USB 3.0/3.1 | 2 TB |
| Ruggedized | 1-2 TB |
| High-end Professional | 8 TB |
As NAND flash lithographies continue to shrink below 20nm, maximum capacities will continue increasing. But tradeoffs around cost, performance, and endurance will remain.
How does a USB drive interface with a host computer?
A USB flash drive uses the standard USB mass storage class protocol to interface with host devices like computers and smartphones. Key aspects include:
- USB Specification – Supports USB specs like USB 3.2 Gen2 for high speed.
- USB Mass Storage Class – Uses the mass storage device class protocol.
- Logical Block Addressing – Data is addressed using logical block addresses (LBAs).
- SCSI Commands – SCSI transparent commands like READ and WRITE are transported over USB.
- Plug-and-Play – Automatic detection and configuration when plugged into a USB port.
The operating system interacts with the drive using standard mass storage drivers built into the OS kernel. This allows the USB drive to appear as just another storage drive without specialized drivers.
What are the main components of USB flash drive firmware?
The firmware that runs on the controller has the following major roles:
USB Interface
Implements the USB specification and mass storage class protocol for host communication.
Flash Translation Layer
Manages the logical to physical mapping of data to flash memory pages and blocks.
Read/Write/Erase
Controls algorithms for reading, writing (programming), and erasing flash memory.
Wear leveling
Evens out wear by spreading out erases across all available blocks.
Bad Block Management
Detects weak or failed blocks and maps them out of use.
Garbage Collection
Reclaims unused space by consolidating data and erasing blocks no longer in use.
Error Correction
Implements ECC and error handling to detect and recover from bit errors.
Together, these firmware components manage all the complexity of making unreliable flash storage appear as a reliable and high performance mass storage device to the host computer system.
Conclusion
USB flash drives use NAND flash memory to provide a compact, high capacity, high speed, and reliable storage medium. The controller and firmware hide the complexity of flash memory management behind a standard USB mass storage interface. Continued advancements in NAND flash technology and controllers will yield USB drives with even more impressive capabilities in the future.