Choosing the Right NAND Flash

print this page
NAND Flash

Overview

A 152% increase in NAND demand is expected in 2009, fueled by Flash storage cards, digital still cameras, mobile handsets, USB Flash drives, MP3 players, set-top boxes, handheld games, personal computers, enterprise servers, and other applications that are expected to collectively consume more than 8.7 trillion gigabytes of Flash memory.1Choosing the Right NAND Flash

In response to this boon in NAND demand, Flash memory manufacturers have begun to offer a variety of NAND Flash memory products—including some specialized NAND solutions—with significantly different performance capabilities and features across a number of process nodes. This NAND diversification means that designers cannot simply select any NAND Flash device for their applications, but rather need to have a basic understanding of the features and benefits of each type of NAND in order to select both the proper device and the proper supplier.

NAND Flash Basics

NAND Flash memory, like many other sorts of memory, stores data in a large array of cells where each cell holds one bit of data. In a typical NAND Flash device, “a high voltage of 18V is applied to the control gate to draw electrons from the substrate to tunnel through the gate oxide into a polysilicon floating gate layer. To store one bit, two charge levels in the floating gate layer can be stored to distinguish between a ‘1’ and a ‘0’.”2 Multilevel cells, discussed later, store additional charge levels within each bit cell.

Typically, a NAND Flash array is organized into many 128KB blocks. Each byte in one of these blocks can be individually written or programmed, but a single block represents the smallest erasable portion of the array. In an erased block, every bit is set to the binary 1. As an example, a monolithic 2Gb NAND Flash memory device usually consists of 2048 blocks with 64 pages per block. “Each page has 2112 bytes total, comprised of a 2048-byte data area and a 64-byte spare area. The spare area is typically used for ECC, wear-leveling information, and other software overhead functions, although it’s physically no different from the rest of the page. NAND devices are offered with either an 8- or 16-bit interface. Host data is connected to the NAND memory through a bidirectional data bus, 8 or 16 bits wide. In 16-bit mode, commands and addresses use only the lower 8 bits. The upper 8 bits are only used during data-transfer cycles.”3


NAND Flash Basics

Figure 1: A 2Gb NAND Flash memory device organized as 2048 blocks

SLC NAND Flash – Performance First

Best for High-Performance, Medium-Density Applications

Single-level cell (SLC) NAND Flash memory is, perhaps, NAND Flash at its simplest and best. As described above, SLC NAND stores one bit of data per memory cell. SLC NAND offers relatively fast read and write capabilities, good endurance, and relatively simple error correction algorithms. SLC NAND can be more expensive per bit when compared to other NAND technologies since each bit cell stores only one bit of data. As a result, designers have to make a choice between cost and performance. If an application needs speed—like a high-performance media card, some hybrid disk drives, solid state devices (SSDs), or some embedded applications4— SLC NAND may be the only choice.

MLC NAND Flash – Density and Value

Best for Higher-Density, Low-Cycle Applications

In contrast to SLC NAND, multilevel cell (MLC) NAND stores two or more bits per memory cell. To determine the state of each bit, a voltage is applied and the resulting current is detected. In an SLC device, only one voltage level is required. If current is detected, then the bit stored is 1; if no current is detected, then the bit is 0. For an MLC device, three different voltage levels are used to determine the state of both bits. Figure 2 illustrates the reference point voltages that are applied and the resulting bit values for both SLC and MLC devices.

Generally, MLC NAND offers twice the capacity as SLC NAND in the same size device and comes at a significantly lower cost-per-bit. While designers will have to make some trade-offs in terms of performance and reliability (since SLC NAND is about three times as fast as MLC NAND and offers up to 10 times the endurance); but for most applications, MLC NAND offers the right combination of price and performance. In fact, MLC NAND represents nearly 80% of all NAND Flash shipments. And MLC NAND will soon become the dominant Flash memory of choice for SSDs because its performance is superior to magnetic hard disk drives.

SLC: One Bit per Cell
SLC NAND stores 2 states per memory cell and allows 1 bit programmed/read per memory cell.

SLC NAND


MLC: Two Bits per Cell
MLC NAND stores 4 states per memory cell and allows 2 bits programmed/read per memory cell.

MLC NAND

Figure 2: SLC NAND vs. MLC NAND

High-Speed NAND Interface – Up to 200 MB/s Read and Write Speeds

Ultra-Fast Read/Write Throughput for Demanding Storage Applications

Essentially, the high-speed NAND Flash interface is a synchronous DDR interface with speed-optimized read and write logic. In fact, the high-speed NAND interface can achieve up to five times the performance of a standard NAND Flash interface. Our high-speed NAND interface is designed to the Open NAND Flash Interface (ONFI) 2.1 standard, which also provides easy design-in and backwards compatibility.

The performance features of our high-speed NAND interface make it a great fit for performance SSDs, USB 3.0, and other speed-intensive applications.

High-Endurance NAND

Best for Intensive, Enterprise Applications

Enterprise NAND is a high-endurance NAND product family that offers significantly longer cycle endurance than standard NAND.

The ability to store data over a number of WRITE/ERASE cycles is often described as endurance. A typical MLC NAND Flash memory device built on a mature process can endure approximately 5,000 WRITE/ERASE cycles, while a standard SLC NAND Flash device endures 100,000 cycles—which is more than enough for most applications. However, for intensive enterprise applications that require significantly longer cycle endurance, Enterprise NAND is a highly reliable, high-density, high-endurance solution.

Enterprise, or high-endurance, NAND is optimized for high-transaction data processing and high-speed server functions. Offering up to a six-fold and three-fold increase for MLC and SLC NAND, respectively, it far surpasses standard cycle rates and markedly improves product life and performance.

Serial NAND Flash

A Lower-Cost, Higher-Density Alternative to NOR

Serial NAND Flash relies on the standard serial peripheral interface (SPI) often used for basic, low pin count communication between microcontrollers and system peripherals. SPI typically uses four pins, including a serial clock, serial data-in, serial data-out, and chip select.

EEPROMs and NOR Flash have dominated SPI memory solutions, but these technologies only offer low-densities options—256Mb at best—that won’t work with some of the newest designs. Boosting serial NOR Flash densities to match demand is theoretically possible but could become cost prohibitive. Enter Serial NAND Flash. Serial NAND Flash has several advantages over earlier SPI memory solutions. First, it’s available in much greater densities, starting at 1Gb, and it offers a lower cost-per-megabit than NOR. On-chip error correction code (ECC) and faster write speeds than NOR are further benefits.

As with all technologies, there are trade-offs. Serial NAND cannot read data as quickly as parallel NAND solutions—a problem serial NOR shares—and designers will have to make software changes to manage NAND’s cache reading, but even with these minor trade-offs, Serial NAND is the right choice for many applications that have traditionally relied on NOR.

NAND for MCPs

Flexibility for Mobile Devices

More and more NAND Flash memory is being used in multichip packages (MCPs) where it is paired with Mobile LPDRAM in a variety of form factors. NAND/LPDRAM MCPs are offered in densities of 1GB to 4GB for SLC NAND and 1GB to 8GB for e-MMC™ Embedded Memory.

A Managed NAND Solution: e·MMC™

Easy to Design In

Managed NAND solutions—like e·MMC Embedded Memory—are a great design choice if ease of development is a key factor. In essence, embedded memory “transforms a program/erase/read device with bad blocks and bad bits (NAND) into a simple write/read memory. This managed interface addresses potential NAND design concerns internally, using error correction code (ECC), wear leveling, and bad block management technology. Handling errors internally takes the burden off the host controller and increases speed, providing higher system performance.”4

NAND Flash memory requires a controller to aid with array management, ECC, wear leveling, and other functions, but where that controller resides and works can and should vary, based on a specific design’s needs.

e·MMC embedded memory includes the NAND Flash part, NAND translation layer, and ECC in a single JEDEC-compliant, MultiMediaCard (MMC) package. And with its low-profile BGA package, e·MMC can be the right choice for designers who prefer an MMC interface with application-to-application interoperability.


NAND Flash Comparison

Figure 3: Comparison of a host controller for a NAND Flash vs. a managed NAND device

NAND Flash SSDs

Preferred for Their Density and as HDD Replacements

SSDs group together many NAND Flash devices and use advanced controllers to manage the arrays so designers who need NAND Flash performance and reliability can also have greater density. SSDs can be developed with high-speed or high-endurance NAND for some enterprise or gaming applications, or can be developed with less expensive MLC NAND for most notebook or consumer applications.

An SSD can emulate a magnetic HDD, including matching disk drive interfaces and protocols like SATA, while offering far greater performance and reliability. Because an SSD has no moving parts, there are no spinning platters or head actuators that can break. And SSDs will not suffer from the superparamagnetic effects that will likely limit future generations of HDDs.

Embedded USB Solutions

A Low-Cost, Embedded Storage Option

Embedded USBs bring the density and reliability of an SSD to networking and embedded applications with a simple universal serial bus (USB) connector. An embedded USB is physically smaller than a 1.8-inch HDD, costs far less to implement than even the cheapest hard drive, draws a mere 330mW of power when it’s actively reading or writing data, and will boot much faster than most hard drives. These embedded USB SSDs mount directly to the industry-standard USB headers found on most PC motherboards. For rugged applications, the additional mounting standoff secures the embedded USB drive to its host, all without cables, brackets, or mounting rails. Additionally, embedded USB drives are available in smaller capacities ideally tailored for embedded operating systems and applications.

Conclusion

A basic knowledge of the various kinds NAND Flash-based memory solutions available can help designers make informed decisions about which NAND Flash device to specify for a particular design. This article's general description of each of the NAND Flash options, along with a snapshot of their features and functionality, helps illustrate what makes them better suited for some applications than others. More technical information about each memory type can be found in the Products section of our Web site.

References:
  1. “Data Flash Market Tracker–Q3 2008,” iSuppli Corp. (October 2008)
  2. Gregory Wong, “Solid State Drives: A Closer Look Report No. FI-NHL-SSD-1008,” Forward Insights, (October 2008): page 15.
  3. Jim Cooke, “Flash Memory 101: An Introduction to NAND Flash,” CommsDesign (March 20, 2006).
  4. nand.com.