Post a Comment
Feature
Memory options: flash may dominate
Microcontrollers that integrate flash memory are beginning to flood the market, bringing with them advantages for engineers.
By Kenji Tsuda, Editorial Director, EDN Japan -- EDN, 11/9/2006
For example, NEC Electronics reports that shipments of flash-based microcontrollers grew 29.6% in fiscal 2005 and will grow 91.4% in fiscal 2006. Renesas Technology also reports significant growth; the company shipped 700 million units from 2001 to June 2005 and expects to reach 1 billion units by March 2007. Fujitsu, Toshiba, and Panasonic also offer flash-based devices, and the trend extends beyond the Japanese vendors, with Freescale Semiconductor, STMicroelectronics, Atmel, and others also strengthening their flash-based offerings.
Flash microcontrollers offer a number of advantages centered on programmability. You can program the devices before or after mounting on a pc board and before or after shipment into the market. Flash microcontrollers, however, have been 20 to 30% more expensive than conventional mask-ROM or OTP (one-time-programmable) ROM versions. The cost premium resulted in IC vendors offering fewer families and derivatives relative to the other product types.
So what has prompted the recent expansion in the flash-microcontroller market? Certainly, a shrinking cost premium for flash is partly responsible. But designers also value the flexibility and shorter time to market that flash-based devices can deliver. In addition, a growing set of development tools from third parties supports quicker design cycles.
Semiconductor process technology also is swinging in the favor of flash. Compared with past devices, today's microcontrollers feature greatly increased transistor density, which allows the integration of broad sets of peripherals. The die area dedicated to memory is shrinking relative to the other functions. Though mask-ROM-based devices were once easier to produce, that advantage has shrunk considerably.
In the current era, when manufacturers offer a wide variety of OEM products with relatively small manufacturing volumes, programmable microcontrollers prove the best choice for many manufacturers. Both the manufacturer and the IC vendor can program the devices. The turnaround time is effectively zero: OEM manufacturers don't have to wait for the mask-program step.
|
Chip vendors also enjoy many advantages. They can produce flash-microcontroller products in increasing volumes and reduce the price premium. And when OEM customers cease production of a particular product, the IC vendors aren't left with unusable inventory.
Flash devices may also be more reliable than mask-ROM devices. According to one OEM, the first mask-ROM products produced for a new design sometimes exhibited insufficient noise immunity—a severe problem that could result in a revision in the design. Meanwhile, a vendor ships the same proven flash-based designs to every OEM customer.
Renesas recently reviewed its flash-microcontroller strategy. The company had already developed a family of flash microcontrollers branded Z-TAT (zero turnaround time). The Renesas portfolio, however, includes a variety of microcontrollers that integrate flash memory, OTP memory, and mask ROM—targeting architectures to specific applications. Like NEC, Fujitsu is shifting to focus on flash microcontrollers, moving away from conventional OTP and mask-ROM products.
OEM designers may also see further reduction in flash-microcontroller prices, driven by new players with their own IP (intellectual property). Upstarts such as Impinj from the United States and e-Memory of Taiwan are entering the market. The Impinj Flash IP features no additional mask steps, whether targeting microcontrollers, LCD drivers, or electronic-ID applications. The Impinj IP design, however, is suitable only for relatively low-density, 2- or 4-kbit products. The e-Memory IP requires just three masks but delivers 2-Mbit or higher density flash that you can integrate on a microcontroller chip. Renesas has licensed the e-Memory IP.
