Glossary of Acronyms

ADC — Analog-to-Digital Converter
BOM — Bill Of Materials
EFD — Embedded Flash Drive
MLC — Multi-Level Cell
NAND — Not And (electronic logic gate)
NOR — Not Or (electronic logic gate)
SLC — Single-Level Cell
XIP — eXecute In Place

By Francois Kaplan, AVP Product Marketing, M-Systems Mobile Division

Multimedia downloads are whizzing along wider broadband networks to their final destinations: 3G handsets, personal media players, digital audio players, global positioning systems (GPS) and game consoles, just to mention a few. Users are shooting and storing more high-resolution photos on their camera phones and transmitting them over a number of wireless infrastructures. With all of this activity, high-density flash memory has become a key enabler of new next generation storage models in the consumer electronics market.

But for designers accustomed to working with NOR flash, the legacy solution for voice-centric cell phones, upgrading to high-density NAND flash can pose considerable obstacles — with potentially dire effects on time-to-market schedules.

click the image to enlarge

Figure 1. Comparison of different memory architectures..

Raw NAND flash media inherently has data reliability issues: the result of bit flips and bad blocks. As NAND flash technology becomes more cost-effective, it also becomes more complex. The latest NAND etching processes and MLC NAND, which stores twice the amount of data per cell than SLC NAND, compound data reliability issues. With new NAND flash designs coming at lightning speed to provide more memory in smaller sizes, chipset support is almost impossible to find. And as NAND flash popularity increases, its supply is problematic, particularly during cyclical times of flash allocation.

Dealing with these obstacles head-on can inflate development and integration time and budgets, while also impeding end-product design cycles. Using smarter NAND flash solutions can sidestep these obstacles by enabling use of multiple NAND sources, while at the same time supporting advanced NAND flash technologies without requiring design changes to the host chipset. Such an approach can help keep designers on track to meet aggressive time-to-market schedules.

Handset vendors using a multiple-source NAND solution can help solve issues of limited NAND flash supply and inflated costs, which can be particularly painful during flash allocation.

Supply continuity is complicated by advances in NAND flash technology and processes that evolve much more rapidly than chipsets, making it increasingly difficult to find chipset support for the latest NAND flash designs. Smartphone chipsets, for instance, began supporting small block SLC NAND flash at the end of 2003. Earlier that year, both large block SLC and more cost-effective MLC NAND had already been introduced. Large block NAND should finally be supported by a few chipsets in 2006, while MLC NAND is not expected to have chipset support. Despite NAND's overwhelming acceptance in the handset market, some lower-end chipsets targeting feature phones still do not support NAND.

Standardizing NAND flash devices to support advanced NAND technologies, without touching the host chipset, can solve the lag in chipset support.

When NAND media was first introduced to the consumer electronics market, successful implementation depended on using an EFD. These EFDs integrate the flash media along with a flash controller on the same chip, and come with specially developed flash management software. The very first EFD was M-Systems' DiskOnChip (also sold by Toshiba). EFDs offer access to the NAND media through a legacy, NOR-like interface. They also provide an XIP boot block, making it possible to completely remove the NOR flash from the host system and thereby reduce the BOM. Additional features include more efficient power consumption, better performance and security.

But the original EFD design, with the flash management software running on the host, could not support the newest, most cost-effective NAND technologies without updating the host software. Because this was a critical and time-consuming task requiring intensive testing and qualification, many designers chose to delay access to new technologies to the next generation platform.

New embedded flash drives, such as M-Systems' DOC H3, standardize the architecture by building the flash management software into the controller as firmware, rather than coupling it with the host (see Figure 1). By minimizing time-intensive host design efforts, new embedded flash drives enable handset vendors to move easily to the next NAND technology. New embedded flash drives can therefore be integrated as plug & play devices and use the most advanced and cost-effective NAND technologies.

As the consumer electronics market continues to offer more and increasingly sophisticated multimedia applications, the need for memory is exponentially increasing, justifying the move from NOR-based flash to NAND-based flash.

To answer new market challenges, both technical and business, consumer electronics vendors would be wise to implement complete, advanced flash memory solutions, such as second generation EFDs, rather than flash components that require heavier integration efforts and higher design costs. They will enable savvy handset manufacturers to cost-effectively support new NAND technologies with plug & play integration ease, helping them get their new, high-density handset models to market faster.