New Small Form Factor Storage Standard Targets Embedded Systems

To achieve small size, low power consumption and fast time to market requirements, embedded systems designers often look to chipsets found in cell phone handsets or mobile internet devices (MIDs) to cost-effectively meet their design requirements. These components, whether they are off-the-shelf chipsets from Intel, AMD or Freescale, or FPGA’s from Xilinx, Altera, or Actel, that later migrate to custom ASICS, often define the available storage interfaces. These chipsets are widely understood and supported and more often than not, make use of USB, SD, MMC or some other type of serial programmable interface that is not usually defined with traditional storage such as PATA or SATA.

Likewise, traditional flash cards (MMC, SD and their mini/and micro counterparts) or USB thumb drives that connect to these interfaces were originally designed for consumer applications, which have a completely different usage model than an embedded system. Consumer application storage devices are designed to be removable and transportable; they are not designed for the rigors of embedded systems.

Many cell phone or MIDs architectures are moving away from these traditional flash cards because:

1. Their biggest support costs revolve around the fact that their customers tend to buy the wrong card.
2. They find that their customers do not upgrade their memory cards every two years; they upgrade their entire MIDs device.

The one thing that all hardware designers know is that whatever the hardware giveth, the software taketh away. So embedded systems designers have traditionally needed to make a value judgment:

1. Design with a removable, upgradeable storage solution that has inferior environmental characteristics, which means a mounting solution must be developed so that the wireless base station or embedded server can be upgraded three to five years into a 10- year deployment, or
2. Design with an embedded module like eMMC or any one of the multitude of similar products that needs to be soldered down on the motherboard to meet environmental requirements, but is not upgradeable.

Both of the above solutions are further complicated by the fact that consumer- centric products only have about a one-year lifecycle, so using either solution will necessitate multiple system requalification cycles during a multi-year product deployment.

The only smaller form factor that has gained true adoption in the embedded and enterprise system OEM space has been CompactFlash (CF). CF became the most popular, widely-used form factor in the embedded space for two main reasons. First, people forget that CF was originally developed to be an industrial standard as a smaller alternative to PCMCIA cards. Only with the advent of digital cameras did it become a storage form factor for consumer elect ronics. Second, CF has stayed around a long time because of the interface modes – mainly IDE (PATA) and PCMCIA memory modes. These modes are well supported by computing chipsets deployed in embedded and military systems. Now, however, even CF is facing major challenges for new designs, primarily because:

1. Most new chipsets are not supporting PATA interfaces because they require too many pins. The industry has gone away from parallel interfaces in general, preferring the faster serial interface.
2. The trend in embedded computing is to migrate to smaller form factor boards such as COMs (computer on modules) where CompactFlash is actually too big for the computing platform.