The speed of NAND flash memory has made a rapid progress. The ONFI v4.0 that is under development will use the 1.2VDDR3 transmission interface technology. Single channel flash transmission rate can reach the speed of 800 MB/s. As a result, solid state drive (SSD) products adopting SATAIII (6Gbps) as the transmission interface are about to face an insurmountable bottleneck.Instead, the PCIe interface technology is promising. The introduction of the new M.2, SATA Express, PCIe x16 add-in card, and SFF-8639 are all new generation SSD products that the storage device industry are banking on to transcend the performance limitations.SATA 6G reaching a bottleneckCc Wu , Innodisk's vice president of Embedded Flash Div. , discussed at a recent forum the application of the flash storage devices used by the industrial control sector as well as the evolution of the storage transmission interface specifications for IDE, SATA, and other interfaces. Wu indicated that the NAND flash manufacturing process has continued to progress each year from 60nm in 2006, 50nm in 2007, 42nm in 2008, 34nm in 2009, 27nm in 2010, 24nm in 2011, 21nm in 2012, and 19nm in 2013 and early 2014.The current industrial mainstream storage devices still adopt the rewritable and high tolerance SLC made using processes between 42nm and 21nm. MLC has also begun to penetrate into the industrial application sector since 2012, manufactured at the 21nm to 19nm nodes. Because TLC has a smaller number of rewritable times and the price difference between TLC and MLC is not as great as that between SLC and MLC, TLC applications are still mostly for consumer electronics, and the industrial market has yet to adopt them. When TLC will find its way into the industrial market remains to be seen.Judging from the evolution of flash performance, ONFI and Toggle mode have both become the mainstream flash interfaces. Under ONFI v3.1/Toggle mode 2.0, flash adopts the DDR2 interface transfer technology with transmission rates reaching 400MB/s (ONFI v3.1) or 533MB/s (ONFI v3.2), and the operating voltage reduced to merely 1.8V(SSTL_18). ONFI v4.0, which is being developed, will adopt the DDR3 transmission interface with the operating voltage dropping to 1.2V and the highest single-channel transmission rate reaching up to 800MB/s.In terms of performance, if an ONFI 3.2 NV-DDR2 goes with a 16KB Page MLC and Cache Read/Cache Program, the read speeds of single-channel 1-die, 2-die, 4-die, and 8-die stacks can reach as high as 533MB/s, and the respective write speeds can reach 20MB/s, 41MB/s, 82MB/s, and 164 MB/s. As for 16KB Page SLC chips, the write speeds can reach 73MB/s, 146MB/s, 291MB/s, and 533MB/s (in full load) under the same Cache Read/Cache Program conditions. But now the 600MB/s rate of SATAIII (SATA 6G) is about to become a transmission bottleneck for flash chips. To develop SSDs with higher transmission speeds, we must turn to the next generation of higher-speed transmission interface as early as possible.SATA remains as mainstream industrial interface, but set to give way to Mini PCIe/M.2In terms of flash applications in the industrial sector, a significant number of SLC has been used in embedded devices, acting as the embedded operating systems' boot storage with capacity mostly under 2GB. When more than 32GB in storage capacity is needed, MLC is a more cost-efficient alternative.When the storage capacity is less than 2GB and demand for storage is high, SLC can be used as an embedded OS hard disk to provide a higher degree of reliability and extend the service life for such products as industrial PCs, gaming machines, automation equipment, and military equipment. When the device needs storage of more than 32GB, MLC can be used as a data disks for storing application programs, enabling higher SSD capacity and providing better cost-efficiency for product segments including surveillance, POS, networking and digital signage.Take an industrial motherboard for an example. The designs mentioned above allow for numerous connection interfaces to connect the storage devices. They include: PATA CF/IDE DOM (Disk on Module) that can connect to the conventional 40pin IDE interface; SSD with 1.8 inch SATA or 2.5 inch SATA; SATA DOM, mSATA, SATA Slim, Cfast, or CF-SATA memory cards that connect to the SATA interface; USB modules that connect to a 10-pin USB connector; SD/micro SD memory cards that can be read through a card reader; and the eMMC and uSSD single-chip disks that are built into the motherboards.These various interfaces must have at least five years of product life in order to serve industrial purposes. In the past, newly designed motherboards would often keep the old transmission connection interfaces in use for several more years to give clients more time to transition to new interfaces. At the same time, multiple interfaces can cover a wide range of differentiated applications. Small-size connection interfaces are also required due to system space constraints. SATA remains as the mainstream transmission interface for industrial applications.PCIe to become the ultimate solutionCc Wu provided a chronicle of the SSD evolution in terms of form factors and transmission interfaces. In 1995, the CompactFlash (CF) had a transmission rate of only 8.3 MByte/s (PIO mode 2). In 1999, the Secure Digital (SD) specification was jointly developed by SanDisk, Matsushita and Toshiba. The SATA Rev 1.0a specification was launched in January 2003, which increased the transmission rate to 150MB/s (1.5Gbps). In 2003, the PCI-SIG organization premiered the PCIe 1.0a single plane (1x plane) achieving the transmission rate of 250MB/s (2.5Gbps). The SATA Rev 2.0 was unveiled in 2006, which improved the transmission rate to 300MB/s (3Gbps). In January 2007, PCI-SIG released the PCIe 2.0 specification whereby the single plane (1x plane) with a transmission rate up to 500MB/s (5Gbps).In June 2008, the SATA Rev 3.0 interface boosted the transmission rate to 600MB/s (6Gbps). In 2009, JEDEC developed SATA Slim (MO-297A). In September 2009, the SATA-IO organization introduced the mSATA interface. In 2010, the PCIe Gen3 enabling single-plane 1GB/s (8GT/s) was launched. The SATA Express interface appeared in 2011 and Intel released the M.2 (NGFF) interface in 2012. The high-speed transmission interface evolution continues.The PCI Express (PCIe) interface evolution started with PCIe 1.0,which was based on the 8b/10b encoding scheme with x1 transmission rate reaching 2.5GT/s (250MB/s) and the x16 transmission rate at 4GB/s. Then came PCIe 2.0, which was based on the 8b/10b encoding scheme boosting the x1 transmission rate to 5GT/s (500MB/s) and the x16 transmission rate to 8GB/s. PCIe 3.0 adopted the 128b/130b encoding principles, raising the x1 transmission rate to reach 8GT/s (1GB/s), and the x16 transmission rate to 16GB/s. PCIe 4.0 is expected to have an x1 transmission rate exceeding 16GT/s (2GB/s), and the transmission rate can reach 32GB/s in the x16 mode.PCIe interface form factors and market trendsPCIe has diverse interface form factors. A Mini PCIe connector has 52 pins and works well with cloud computing and data centers; PCIe x16 is used in graphic cards and PCIe SSD emphasizes high IOPS; M.2(NGFF) launched by Intel in 2012 features three sizes - 2242, 2260 and 2280 - that are currently used in Ultrabooks.The 2.5-inch SSD has adopted the SATAIII (600MB/s) interface. The SATA Express combines the conventional SATA and PCIe interfaces and provides x2 dual planes. PCIe 2.0/3.0 x2 can achieve 1GB/s and 2GB/s. Asustek has showcased motherboards adopting SATA Express, and their interfaces are compatible with conventional SATA cables and PCIe connectors.Finally, the SFF-8639 interface is currently used in the back panels of the enterprise-grade 2.5-inch storage devices. It can connect to PCIe, SATA as well as SAS storage devices, and is compatible with SATA Express. SFF-8639 offers six PCLe lanes, but only a maximum of four can be used at a time. PCIe 2.0/3.0 can raise the transmission rate to 2GB/s and 4GB/s, respectively.In terms of the industry support for SFF-8639, Dell's PowerEdge R820/R720 1U Servers are connected through PCIe HBA adapters at the backplane connection and expansion slots, and it is adopted by Apple's MacBook Air as well as some MacBook Pro models. However, it is still rarely seen in the industrial computers/motherboards. Micron has premiered the P320h 2.5-inch PCIe SSD that adopts the SFF-8639 connection interface. It supports the x4 PCIe Gen2 specification and its sequential read and write speeds can reach 1.75GB/s and 1.1GB/s, respectively.SSDs developed with the PCIe interface in the form of x16 add-on cards include: OCZ's Z-Drive R4(PCIe-SATA+SATA SSD); Micron's P420m; Fusion IO's ioDrive; Seagate's X8 (Virident); and OCZ's Z-Drive R5 native PCIe SSD cards. They provide high IOPS that is most needed by servers. Their transmission rates can reach up to 1.5GB/s to 3GB/s, or even up to 6.5GB/s.Industrial motherboards have miniPCIe slots of the 50.8mm (L) x 29.8mm (W) x 4.4mm (H) size. In the past, these slots were slated to accept Intel's Wi-Fi wireless LAN modules. Manufacturers have also modified them into mSATA slots with similar pin counts. M.2 (NGFF) can serve as a new generation of slot/interface specifications compatible with miniPCIe, USB, SDIO, UART, PCM, I2C, and SATA(mSATA) add-on cards. There are Socket 1 Type 1630, 2230, 3030; Socket 2 Type 2242, 3042, which support PCIe x2/SATA; and Socket 3 Type 2260, 2280, 22110, which support PCIe x4/SATA.PCIe is poised to replace the existing SATAIII to become the future mainstream specification for storage device. Judging from the history of storage devices where a transition would usually take three to five years, Cc Wu expects that PCIe SSDs will become the mainstream by 2015. In addition, judging from the fact that Apple's MacBook Air and MacBook Pro have already adopted M.2 and SFF-8639, the transition of the SSD industry will continue to accelerate. Various SSD products adopting PCIe will start to appear in the market in the second half of 2014.Cc Wu, Innodisk's vice president of Embedded Flash Div.

In the industrial PC market, there is strong demand for high-performance, highly reliable and ruggedized SSDs (solid-state drives). Apacer Technology has overcome SATA III technology barriers and developed the ultra-performance Combo SATA Drive (CSD), and Corepower technology protects data in cache from impact of power disruption. Their water- and dust-proof capabilities are improved by special coating and molding technologies. Apacer also provides industrial- and military-class protection/erase methods and cloud computing-based monitoring software to enhance data protection and remote maintenance capability.Big changes in storage device interfacesRobert Lee, Director of R&D Dept. at Apacer outlined the development of interface technology for storage devices, their specifications, and SSD applications in industrial PC. Serial ATA International Organization (SATA-IO) released SATA Rev 2.5 in October 2005, SATA Rev 2.6 in February 2007, SATA Rev 3.0 in June 2009, SATA Rev 3.1 in July 2011. In August 2013, the SATA-IO released SATA Rev 3.2 and several new interface specifications such as SATA Express (derived from PCI Express) and M.2 (NGFF), specified definition of microSSD and removable SSD modules (universal storage modules, USM) and optimized solid-state hybrid drives.The specification of SATA Express, a new 2.5-inch storage device interface added to SATA Rev 3.2, keeps the original SATA Connector, with definition changed as two sets of 7-pin SATA Port connectors + 3/4-pin PCIe signal. SATA Express is compatible with existing SATA connectors, allowing use of two connection cables. If a SATA Express male connector is plugged in, the interface switches from SATA Express to PCIe. SATA Express is designed to be 2-lane (x2), two-way multiplexing of serial signals with a maximum transfer speed of 2GB/s (PCIe 3.0).M.2 originated from Next Generation Form Factor (NGFF) unveiled by Intel at its 2012 IDF and is specifically designed for add-on cards of Ultrabook and tablet SSDs. M.2 is used in SSD cache, Type 2242 SSD Cache Slot B measuring 22mm x 42mm, and SSD add-on cards, Type 2260 SSD Slot C measuring 22mm x 60mm and Type 2280 SSD Slot C measuring 22mm x 80mm and Type 22110 SSD Slot C measuring 22mm x 110mm.microSSD (uSSD) is SATA-IO's specification (JEDEC MO-276) specific to SATA SSD, a single chip through integrated packaging of NAND flash and the controller die. A microSSD chip measures 20mm x 16mm in FPBGA (fine pitch ball grid array) package with pitch of 0.5mm. microSSD can be directly soldered on the motherboard of a small embedded system to save space of the PCB.Device Sleep (DEVSLP) is an advanced power management protocol for SATA, but has not been used in all industrial PC equipment or general motherboards. In the past, power management for SATA was done through software command, with operation from active state to partial state and then to slumber state. The addition of DEVSLP definition to SATA Rev 3.2 enables SATA peripheral components to go into DEVSLP state for further power saving.PCIe to become mainstream, but application to industrial PC may see longer waitAlthough a shift from SATA to PCI Express for storage devices will be an unstoppable trend, such a shift will not happen in the near future because motherboards used in industrial PC need much time to design and obtain certification, Lee said. PCI Express interface has progressed from 8b/10b coding, single-lane (x1) transfer speed of 2.5-5GT/s (250-500MB/s) and 16-lane speed of 4-8GB/s for PCIe Gen1 & Gen2 to 128b/130b coding, single-lane (x1) transfer speed of 8GT/s (1 GB/s) and 16-lane speed of 16GB/s for PCIe Gen3.In terms of SSD device sizes, M.2 (NGFF), with 67/75 Connector and 22m x 42/60/80/110mm in dimensions, allows for SATA and PCIe interfaces. MO-300 (mSATA), whose 52-pin mSATA slot is derived from miniPCIe slot, can support SATA and PCIe. A 2.5-inch SSD measuring 100mm x70 mm can support SATA and PCIe (SATA Express/SFF-8639), while MO-297 (SATA Slim) storage modules, despite their 7+15 SATA Connector, so far support only SATA but not PCIe.SSD solutions for embedded systemsSSDs have been widely adopted by embedded systems including POS (point of sale/service), medical devices, servers, games consoles and tablets for booting and storage purposes. For industrial PC purposes, the order of priority in selecting SSDs was previously: reliability>performance>cost. But the priorities have now changed depending on different market segments. The emphasis may be on performance: performance>reliability>cost; or, more or less similar to consumer devices, the order may become: cost> reliability>performance. All of them, Apacer will add more kinds of application for industrial PC.For application to industrial PC, SSDs have faced serious challenges to meet requirements of very high speed, cache data protection upon power disruption and ruggedized design, such as water- and dust-proof. In addition, enhanced information security is a must. Users may also want to constantly keep track of the SSDs' conditions and service life, and they may also need smaller and lighter SSD products.SSD for industrial PCThe Apacer-developed 2.5-inch Combo SATA Drive (CSD) is a SSD device consisting of two SSD systems with two SSD controllers on a single SSD PCB. Under RAID 0 operations, the motherboard of CSD, specifically for industrial PC, supports 2-port SATA and reaches nearly PCIe Gen2 x2 performance (1GB/s). Based on CrystalDiskMark v3 testing, the CSD has sequential read and write speeds of 1,024MB/s and 602MB/s respectively. Apacer offers CSD models of SLC SSD/MLC SSD hybrid design, with SLC for OS and MLC for data storage.Apacer CSD uses a SFF-8482 29-pin SAS male connector which, based on notch design, can be connected with a SAS female connector at the backplane of a server or network-attached storage device, but not a SATA female connector. System products with SAS female connectors on motherboards can be connected with, in addition to SATA SSD/hard disk drives, Apacer CSDs for quicker operations and backup.DRAM is added to most SSDs as data cache to increase read/write speeds, but this runs the risk of losing data on DRAM cache if power supply is disrupted. To solve the problem, Apace adopts CorePower technology, using special power protect circuit which, upon power disruption, supply power for controller ICs to write back data from cache to non-volatile NAND flash to prevent loss of data.For SSD reliability protection design, Apacer adopts low-pressure molding technology to completely clothe ICs and pins on PCBs to avoid oxidation and bending, and this also increases shock-resistance as well. In addition, Apacer adopts inexpensive conformal coating as well as nano coating which can enhance resistance to oxidation, moistures and corrosion by chemicals, solvents or carbon dioxide.In terms of data protection, Apacer adopts the following technologies: CoreSecurity by virtue of unique encryption to prevent SSDs from being shifted to other platforms for reading; CoreProtector to prevent data, hard disks and devices from being written at three protection levels; CoreEraser Technology to provide functions of quick erase, full erase and military-class erase (MIL Erase).To keep track of SSD conditions - such as storage capacity, firmware, S.M.A.R.T values, system workload, block erase count, estimated service life, SSD health, system power hour/cycle/failure - it is usually necessary to install monitoring software in the PC equipped with the target SSD. Instead, Apacer provides SSDWidget, a cloud computing-based monitoring application available for download to handsets or tablets with iOS, for monitoring Apacer's SSDs. Apacer SSDs built-in any connected device will send encrypted information on their operations to servers of the clients' private clouds, and the information can be accessed and displayed using SSDWidget anytime. In addition, such information can be browsed using connected devices setup for the purpose.Apacer has developed uSSD, a microSATA Disk Chip (uSDC) measuring 20mm x 16mm in FPBGA 156 and supporting SATA III. uSDC has been applied to mobile devices, storage on board, slim notebooks and Chromebooks. In addition, Apacer has developed SATA Disk Module which is directly plugged in a 7-pin SATA connector without using a power cable. Power is supplied from the male connector (Not 7-pin) with extra fixed pin of the motherboard SATA port through the female connector of the port to SATA Disk Module.To meet demand for slim SSDs, Apacer has developed a SSD of only 5mm in thickness that features a SATA 6Gb/s interface. Compatible with a 22-pin SATA connector, it weighs only 57g. Compared to Seagate's and Western Digital's 5mm HDD models, Apacer's 5mm SSD has the advantages of 450MB/s sequential read/write speeds and compatibility with SATA connectors.While there are many SSD models for industrial PC, Apacer provides SSD solutions with customized hardware and firmware to meet clients' needs running embedded systems, Lee pointed out.Robert Lee, Director of R&D Dept. at Apacer

Addressing the embedded market's demand for high-performance, energy-saving and connected systems, Microchip has developed solutions basing on the MIPS32 microAptiv core architecture, offering support for high computing power per megahertz, large memory capacity and a wide variety of peripherals. The integrated development framework that combines versatile in-house or third-party drivers, libraries, real-time operating systems and middleware components substantially shortens the time for development and verification of 32bit MCU program codes. The PIC32MZ EC family Microchip's PIC32MZ Embedded Connectivity (EC) family consists of 32bit microcontrollers developed for embedded applications with Internet connectivity. "PIC32MZ is a breakthrough that features large memory capacity and integrated peripherals. It provides class-leading performance such as 330 DMIPS and 3.28 CoreMarks/MHz at 200MHz, enabling more than three times the performance of the previous-generation PIC32MX MCUs, and downsizing storage space/density of program codes by 30% compared with competitors' 32bit MCUs," said CY Lin, technical manager at Microchip. "Boasting industry-leading designs, PIC32MZ incorporates the 12bit analog-to-digital converter (ADC) with a sampling rate of 28M per second, much faster than ordinary MCUs whose ADCs offer maximum sampling rates no higher than a few MHz. PIC32MZ can be relied on to create applications that have been previously beyond imagination." PIC32MZ's remarkable features include: maximum 2MB flash and 512KB SRAM; dual-panel flash with Live Update that allows original firmware burning to be continued in a different flash memory block of a running MCU to which the upgraded firmware is simultaneously downloaded; and SRAM capacity which is four times as large as that of the previous-generation MX family. "PIC32MZ integrates peripherals such as Hi-Speed USB 2.0, 10/100Mbps Ethernet controller, two CAN 2.0b control modules, six UARTs, six SPI/I2S serial interfaces, five I2C communications interfaces, and 4bit Serial Quad Interface (SQI)," Lin said. "PIC32MZ with the built-in hardware crypto engine encrypts data such as AES/3DES/HA, MD5 and HMAC, reinforcing safety of e-commence, such as on-line shopping." The PIC32MZ EC family provides a series of development tools, from entry-level Starter Kits, Plug-in Modules (PIMs) to full Developer Boards. A Starter Kit costs about US$119, and a PIM US$25. A Starter Kit with Multimedia Expansion Board II allows peripherals to directly plugged-in to form embedded system prototypes. And a 168-pin to 132-pin adapter board costs US$59. The PIC32 family and its roadmap Lin also outlined the roadmap for Microchip's PIC32 MCU family, in which "DMIPS performance" and "specification/function" form the horizontal axis and the vertical axis, respectively. PIC32MX1/2 MCU series (66/83DMIPS; far left), which are MCUs with 28-44 pins, provide diversified memory capacities: 16-32KB flash/4-8KB SRAM, or 64-128KB flash/16-32KB SRAM. And they also support peripherals such as USB, I2S, CTMU (charge time measurement unit) and PPS (peripheral pin select). PIC32MX3/4/5/6/7 MCU series (105/131DMIPS) come with 64-100 pins. The entry-level PIC32MX3/4 MCUs provide memory capacities from 32-512KB flash/4-32KB SRAM, as well as built-in USB controllers. The midrange PIC32MX3/4 models provide memory capacities from 64-512KB flash/16-128KB SRAM, and they support peripherals such as USB, I2S, CTMU and PPS. The high-end PIC32MX5/6/7 models provide memory capacities from 64-512KB flash/16-128KB SRAM, and they support USB, Ethernet and two CAN controllers. The newly introduced PIC32MZ EC family (330DMIPS) offers memory capacities, from 1-2MB flash/512KB SRAM, and other functions such as HS USB, Ethernet, two CAN controllers, hardware crypto engine and PPS for various application segments, such as consumer electronics, acoustics, factory/building, household automation, environmental monitoring, electric meter, renewable energy control modules, automobile electronics, household appliances and dashboards. PIC32MZ EC's core architecture, function blocks and developer kits PICMZ EC is based on Imagination's MIPS32 microAptiv core (codename: M4K) with a 5-stage pipeline design and a clock rate of 200MHz. The microAptiv core which renders computing performance of 1.65DMIPS/MHz incorporates a built-in DSP (digital signal processor) and 159 DSP accelerating instructions for switching between 16bit MIPS16 and 32bit MIPS32 instruction sets, demanding less storage space for program codes. Moreover, the interrupt controller substantially reduces latency of an interrupt operation, enabling better power performance. Microchip offers entry-level Starter Kits and PIMs for embedded PIC32MZ EC MCUs. The entry-level Starter Kit with integrated debugging and program development functions features scalable designs such as PIC32 expansion board and specific daughter card, USB power supply, 10/100Mbps Ethernet, bi-directional USB 2.0 OTG controller, 4MB SQI Flash and PIC32MZ EC MCU chip for on-line developer and software download. As a good option for users who are proficient in the Explorer 16 developer boards based on Microchip's 16bit PIC24 and dsPIC or 32bit PIC32 MCU, PIC32MZ PIM can be directly plugged into the Explorer 16 developer board for upgrade of embedded application development by means of PIC32MZ EC MCUs. Multimedia Expansion Board II contains an LCD panel providing versatile features such as high WVGA (800x480), capacitive multi-touch function, VGA (640x480) front-facing lens (300K pixels), Wi-Fi and Bluetooth radio frequency component, 24bit stereo, 3-axis accelerometer and temperature sensor, all of which can be installed on Starter Kit for embedded multimedia applications. Microchip also presents software for decompression of MP3 or AAC, graphical HMI and Internet connectivity. As summarized by Lin, the brand new PIC32MZ EC is a competitive product family characterized by high performance at 200MHz, 330DMIPS and 3.28CoreMarks/MHz, maximum memory capacities including 2MB flash and 512KB SRAM, maximum 48 channels, 12bit ADC with the sampling rate of 28Msps, hardware crypto engine, and chip encapsulation rivaling 64pin QFN/124 VTLA (9x9mm) and cost the buyer just US$6.68 (unit price based on 10K pieces). MPLAB Harmony integrates embedded software and hardware Based on the integrated development environment of MPLAB X IDE, MPLAB Harmony is an integrated interoperable component development framework for Microchip's PIC32 MCUs. MPLAB Harmony coordinates all software solutions including in-house or third-party middleware, drivers, peripheral libraries and RTOS (real time operating system) for interactions. All related third-party software solutions are directly sold by Microchip and they are available at microchipDIRECT, the portal site of Microchip for instant technical support, through which a buyer can be authorized for resale, assistance and warranty. The embedded system developers are meeting the challenges as follows: (1) Complexity: Terminal devices are getting more complicated, which requires large numbers of support items and longer development time. Software accounts for 60% of the entire development time. (2) Rework: When different RTOSs are adopted, new errors may arise in terms of incompatibility between software, and this increases the risk of having to rework program codes. (3) Inconsistency: When the RTOS, middleware and drivers have to be incorporated onto the same platform, conflicts may arise. (4) Test drag: 60% of resources in software development are spent on testing and verifying the compatibility of components. If errors are detected when it has already entered the verification process, the time needed to fix them will be 10 times as much as that needed to fix them at the initial design stage. (5) Fragmented support: it is difficult to obtain all software components from a single supplier's website. (6) Obsolescence: The future is unpredictable, and when considering a software component, the developer must evaluate future economic risks. MPLAB Harmony, Microchip's fully integrated development framework, has the following advantages: (1) reduced R&D schedule and expenditure; (2) all in-house and third-party source codes tested, debugged and verified by Microchip without re-work and a user's program codes highly reusable in the Harmony framework; (3) modularized framework effectively enabling stack and integration of software components; (4) pre-tested software components with possible mistakes excluded before verification; (5) Microchip's single-source MPLAB Harmony program codes covering third-party software components; (6) faster time to market because of reduced lead-time. Steven Lin presented the architecture of MPLAB Harmony as follows: the lowest layer of peripheral libraries (PLIB) consists of PIC32MX and PIC32MZ; the second layer of device drivers consists of Interniche's TCP/IP and freeRTOS or Wittenstein's OpenRTOS; the third layer of Middleware consists of Interniche's TCP/IP and freeRTOS, Wittenstein's OpenRTOS and Micrium μC-OS/III, and wolfSSL's CyaSSL Embedded SSL. The fourth layer of common system services consists of USB device, USB host, TCP/IP, file system, Wi-Fi 802.11g, cryptography libraries, GDDX/GRC graphics, MP3/AAC decoder, peripheral libraries and math/DSP libraries built in MPLAB Harmony. Furthermore, other components such as Bluetooth audio stack, smartphone accessories, Class B safety and mTouch sensing will be incorporated in the future. At present, the users can freely access the fundamental framework and most libraries of MPLAB Harmony, which has been available for download from Microchip's website since November 18, 2013.The users have options to purchase third-party developer tools and drivers/libraries (the premium edition). Lin concluded that MPLAB Harmony as a development framework integrating most cross-industry software components provides one-stop shopping services that facilitate porting and reduce lead-time. Steven Lin, Technical Manager, Microchip

Unlike those of ordinary PCs, industrial PCs' functions and features vary depending on the operating environments and requirements, such as for high performance, 24/7 operation , wide-range working temperatures or voltages, dust-proof and noiselessness. A specialist in industrial computing devices, Advantech offers various ARK models of high-performance, low-power-consuming, fanless, industrial Box PCs and extensible Box PCs whose performance is comparable to desktop PCs. These extensible models can be mounted on walls or the backs of display screens, and are designed to meet the various environmental requirements for industrial, commercial and IoT needs.Intelligent Systems Drive Diverse Industrial ApplicationsIn a discussion on how industrial, fanless, intelligent systems enable high-efficiency management through seamless integration, Advantech senior manager Sandy Chen introduced the company's ARK fanless intelligent systems and related solutions. The current trend shows that different industries require different embedded industrial intelligent systems to serve specific purposes, such as for high stability, low power-consumption, noiselessness, extensibility, wide-range working temperatures or voltages, high computing performance and dust and shock resistance.Fan cooled systems used in conventional PCs result in dust accumulation, which eventually reduces fan speeds, heat dissipation, and system stability. In addition, fans are noisy and prone to incurring unexpected additional maintenance costs. So to meet all these requirements, Advantech developed a complete lineup of fanless solutions.Highly Reliable Fanless Systems Run SilentlyAdvantech's ARK fanless intelligent system series offers a range of unibody Box PCs. The series is designed to work under a wide voltage range from 9 ~ 36V. Its advanced, fanless design is dustproof and noiseless and its heat-dissipation structure and chassis efficiently conduct heat to the external environment. With industrial grade or wide-temperature grade components, the series can stably operate within -20 to 60 degrees Celsius.The ARK series features a ruggedized chassis which is IP40 dustproof, 5G shockproof, 50G fall-proof, and with prolonged MTBF (Mean Time Between Failures). The series has garnered various safety certifications, including CE, FCC, CCC, UL, BSMI and CB, and can therefore be marketed in countries that observe strict safety standards.Five Types of Fanless Intelligent Systems Meet Specific Industrial Needs Advantech's ARK series consists of five types of systems that meet specific industrial purposes: palm size models, screen-integrated wall-mounted models, highly extensible I/O models, high-performance models, and surveillance/in-vehicle models.The ARK-1100 palm-size (133.8 x 43.1 x 94.2 mm) mini models run on power consumption of only 5 ~ 10W, support up to quad-core processors, work within -20 to 60 degrees Celsius, and support 2.5-inch Hard Disk Drives (HDD)/SSDs or mSATA storage devices. The ARK-1100, with 12V DC power input, VESA/DIN-Rail sliding and wall mounting specifications, as well as support for miniPCIe + SIM, wireless modules and integrated antennas, is particularly suitable for IoT applications or for tiny spaces requiring extremely low-power consumption.The ARK-1500 series are slim (231 x 46.3 x 133 mm) screen-integrated wall-mounted models supporting multiple screens (HDMI/VGA/LVDS ports) and their VESA/DIN-Rail and wall-mount specifications allow them to be directly hung on the back of the screens. They have a wide range of working temperatures from -20 to 55 degrees Celsius. The series supports 12V DC power input, miniPCIe + SIM, and removable or fixed 2.5-inch hard disk drives/SSDs. In addition, the ARK-1500 has a fixed, lockable DC/VGA/COM/DIO/HDMI/LVDS port structure. The series is suitable for use in combination with a panel for panel PCs, and in dashboards and digital signage.The ARK-2000 series consists of highly extensible I/O systems in a 265 x 69 x 133 mm form factor. They support Intel® processors (from AtomTM to CoreTM i7), working temperatures from -20 to 60 degrees Celsius, a wide voltage range (12V DC or 12 - 24V DC), 2.5-inch HDDs/SSDs or mSATA or CFast storage devices, extensible MIOe modules for easier customized design, and miniPCIe + SIM module extension. The ARK-2000 is ideal for applications that require a lot of flexible extensions.The ARK-35 series high-performance models come in a 250 x 110/90 x 230 mm form factor and support high-performance CPUs (CoreTM i series with a maximum of four cores), and extension solutions (two PCI, PCIe x1, PCIe x4, MIOe module, two sets of miniPCIe). They work from -10 to 60 degrees Celsius with SSDs, and from -10 to 50 degrees Celsius with HDDs. They offer a wide range of operable voltages (12V DC or 9 - 34V DC), and two sets of 2.5-inch removable HDDs/SSDs or two sets of mSATA or CFast storage devices. The ARK-35 series is suitable for applications requiring high computing performance and card extension capability.The ARK-V in-vehicle series models and ARK-S surveillance models, both available in dimensions of 265 x 75 x 133 mm, support various processors (from AtomTM to CoreTM i), a wide working temperature range from -20 to 60 degrees Celsius, 2.5-inch HDDs/SSDs, and mSATA and CFast storage devices. The ARK-V supports ignition power, isolated serial ports and isolated DIO, on-board GPS and G-sensor functions, while the ARK-S features PoE and VMS (Power View).High Performance Leads the CompetitionIn terms of CPU performance, while competitors' products may support only 1.33GHz - 1.86GHz AtomTM dual-core processors, Advantech's ARK-112x series supports quad-core processors whose performance is 3.1 - 3.6 times the competitors' as measured in SiSoftware Sandra Aggregate Arithmetic Performance testing, with an extra 1.25 times with Turbo Boost enabled.In terms of graphics performance, new quad-core, low power processors are superior to the previous AtomTM series by 4.8 ~ 7 times, as measured by 3DMark 2006, 1280 x 1024 resolution testing. To survive the harsh environments where industrial PCs usually operate, Advantech products are designed to achieve higher performance and stable, fanless, heat dissipation.Seamless IntegrationIn addition to the H/W or mechanical design on ARK fanless intelligent systems, Advantech offers "360-degree" intelligent embedded designs (solutions seamlessly integrating hardware, firmware, and software), where "360" refers to triple security, six times more than common energy-saving levels, and zero-distance management.The triple security refers to white-list protection, backup/recovery, and sentinel hardware key. White-list protection makes use of built-in McAfee security software to designate permitted executable software applications, with all unauthorized applications disabled, which prevents viruses or Trojan horses. Using Acronis-developed software tools that also come with all Advantech systems, the time for automatic backup/recovery can be set without affecting regular operations. The sentinel hardware key allows execution of specific software functions through precisely designated passwords.The capability of 6 x energy saving is enabled by energy-saving management, highly-efficient time management, and total cost reductions. Energy-saving management saves power through setting upper or lower load levels or on/off times to let systems run in energy-saving modes when there is little or no other work. Highly-efficient time management updates content by remote control without the manager having to travel to the work site. Total cost reductions are achieved through smart power management and stable data transfer during information exchange (for example, when a public bus passes by a stop, it needs stable data transfer). This reduces total equipment cost.Zero-distance management means remote monitoring, notification, and control. When a traffic light system, for example, breaks down, the maintenance staff - under conventional management - is always dispatched and asked to repair the system as soon as possible, without first ascertaining the cause of the problem (such as software or component failure, vandalism, or weather conditions). Although the repair will be finished eventually, a lot of social resources may be wasted during the repair process, and harm will be done to the product image. It is a different story with remote monitoring. The firmware provides real-time information on system conditions (such as CPU voltage, frequency, temperature), allowing management personnel to monitor the conditions of systems in different locations. The remote notification function can be set to issue warnings whenever there are abnormal conditions, or directly inform the maintenance staff through text messages, and remote control also allows for real-time diagnosis and operation.Advantech's ARK fanless industrial Box PCs are quality products created through excellent design and comprehensive testing. Therefore they can be totally relied on for their stability and high performance. They are comprehensive embedded intelligent systems integrating hardware, firmware, and software. Their high efficiency, power savings, easy management, and stable performance translate into low maintenance costs. Each of the ARK devices takes into consideration extensibility through integration of wireless modules, and they are ideal IoT solutions. These fanless, intelligent systems offer high management efficiency and accelerate time-to-value to meet various industrial or computing requirements.Sandy Chen, Senior Manager of Advantech

With the fast expanding reach of the Internet and the approaching of the era of cloud computing, the HTML5 open platform enables quick development of cross-platform applications, easing concerns that there may be different user experience on different connected devices. Mozilla, known for its Firefox web browser, is also eyeing the mobile operation system (OS) market. It has launched Firefox OS, which is based on HTML5 and Open Web resources. It is a truly open source platform that frees software and hardware designers from the restrictions that would otherwise be imposed on them by closed and semi-closed platforms.The advantages of open platformDuring a recent forum, Mozilla Taiwan Marketing Director Michael Hung explained why Firefox OS is necessary and where its advantages lie.In terms of mobile computing, iOS and Android combine for an overwhelming share of more than 90% of the smartphone market. Not much room has been left for others, not even for the software giant Microsoft, whose Windows Phone can only be fighting for third place and a small fraction of the market. Why, then, a small organization like Mozilla is eager to have a share of the "Red Ocean" by launching its Firefox OS? Hung noted that it has something to do with the nature of Mozilla. Historically speaking, Netscape had more than 80% of the browser market in the early days of the Internet, until Microsoft launched its Internet Explorer (IE) in 1996 and the tables were beginning to turn. In 2002, Microsoft's IE had over 80% of the market.With a drastically shrunk market share, Netscape was taken over by AOL. It then released its browser source codes. It was at this time that the Mozilla Foundation was first conceived. The formal establishment of the foundation in 2003 came with a goal to return to the browser market, and the Firefox 1.0 was officially unveiled towards the end of 2004. The foundation's volunteers even used their own money to run an advertising campaign in New York Times to announce to the world that there was now an alternative to IE.Many users thought at the time that IE's dominance might not be a bad thing; after all it was a mature product. But its dominance meant there was no pressure from competitions, and its upgrades came very slowly, hindering the development of the Internet. For example, there was a gap of five years between IE6 and IE7.Mozilla on the other hand did not aim at profits, and its Firefox underwent constant revisions (currently it is updated every six weeks) to meet fresh needs with new functions and components. Firefox's market share then grew gradually, and by 2009 it had obtained more than 30% share. Mozilla has succeeded in turning the browser market into an open market where no single standard can dominate it.Fighting the giantsSomething similar is happening in the mobile platform market: iOS and Android have more than 90% of the market. That's why Mozilla decided to jump in to offer a truly open platform.Hung pointed out that iOs is a closed platform, while all those in the Android camp must follow Google's specifications. The availability of incompatible platforms means developers have to create different apps for different OSs.The iOS platform requires apps to be written in Objective C, and all SDK (software development kits) must be developed specifically for Apple's hardware platform. The same is true for the Android camp, whose engineers use Java or some languages to develop the apps. We can see that program developers must learn all necessary languages and hardware system specifications when developing new apps. Although some third-party software tools in the market claim that they can enable cross-platform transplant, developers must pay for the programs, as well as access to the libraries. And they have to do debugging and performance tuning on their own. They have to spend a lot of time and resources, which means extra costs of app development.At present, the standards of mobile platforms are set by the OS providers themselves. This means an app could be made useless overnight if the OS provider changes the standards. But the Open Web Platform is one where no individual has the absolute authority. Everything must be collectively discussed and decided. The W3C (World Wide Web Consortium) makes the standards for everyone to follow. The open-source technology gives the industries room to improve.The Firefox OS architectureWhile the desktop OS segment has been steered towards Open Web designs, the mobile OS segment has remained in a world of closed platforms. Mozilla, in a bid to pry open this closed world, devised the Boot to Gecko (B2G) project in 2011, and rebranded it Firefox OS in 2012. Open Web standards and architecture are incorporated into Firefox OS, allowing developers to use HTML5 to develop various cross-platform native apps. They do not have to rewrite their apps to fit different hardware architecture, and this accelerates their products' time to market.Despite being a latecomer to the OS market, the Mozilla open source model has received much approval from vendors. The initial version of Firefox OS was released in April 2013. It consists of a Linux (Gonk) kernel, with Gecko as its rendering engine. The next layers are: HTML5-compatible Web APIs and Web Activities; GAIA user interfaces (based on HTML5, CSS and JavaScript); and HTML5 applications.The booting up of Firefox OS starts with the bootloader, through the Linux kernel init, to the Gecko process and finally to GAIA system apps. Since the OS is based on HTML5 and Open Web resources, developers only have to learn HTML, JavaScript and CSS; no other languages are necessary.It is worth noting that the Firefox OS marketplace is also open. Vendors can not only leverage the app resources readily available in the Firefox marketplace, but also set up their own marketplaces (they set their prices and maintain their own software) for use by Firefox handsets. This is quite attractive, compared to the other two OS platforms.Currently, Firefox OS's major R&D work is handled by Mozilla Taiwan, which was set up two years ago and currently has a team of more than 100 people. A comprehensive list of HTML5 APIs and Web APIs is available on the Mozilla Foundation's website, and the current development status for each API is shown. For example, "D" refers to desktops, "A" refers to Android (Firefox also supports the Andriod platform), and "B" refers to B2G (i.e. the Firefox OS platform). Different colors are used to indicate the progress.Furthermore, handset platforms are very different from desktop ones (e.g. handsets offer calling, vibration, screen rotation, power management, and location functions). New APIs developed for handsets must be approved by the W3C committee before becoming new standards. The W3C decision-making process is transparent, with all of the committee's meeting minutes being made public weekly. Developers can freely obtain the source codes, and if there are bugs, they can find solutions through discussions within the Mozilla community.The future of Firefox OSFirefox OS was demonstrated at Mobile World Congress in February 2013, and picked up momentum fast. The first Firefox OS handset was introduced into the market in Spain in July. Other commercialization projects have also been picking momentum.So far, Firefox OS already has all necessary app support, and it allows customized software to meet specific market demands, such as those for YouTube, Facebook and Twitter. As of 2013, Firefox OS-based devices were available in 14 countries in Europe and Central and South Americas. They initially targeted entry-level smartphone segments, or emerging markets where there was a transition from feature phones to smartphones, avoiding markets already dominated by iOS and Android.Alcatel, LG, ZTE and Foxconn have already announced moves to become hardware partners of Firefox OS. They will launch Firefox handsets and tablets. At CES 2014, Panasonic also announced plans to introduce Firefox OS-based smart TVs. VIA's APC now also supports Firefox OS.Although Firefox OS is still a young product with fewer apps than its two major competitors, it has very strong prospects as its APIs, software and hardware ecosystems improve. The focus for 2014 is to raise the performance of Firefox OS, enhance its functions and expand its popularity among vendors.Mozilla Taiwan Marketing Director Michael Hung

The prospects of smart applications, which are needed to serve the purposes of cloud-based networks of sensing devices, IoT (Internet of Things) and terminal devices, have triggered strong interests in Big Data and inspired imaginations of next-generation cognitive computing and artificial intelligence. The Foxconn Group has teamed up with US-based Mozilla to build cross-platform smart applications using HTML5 Web-based Firefox OS. The smart applications allow clients to build highly optimized devices integrating support for eight screens, the Internet and the cloud."Embedded sensing and IoT smart devices, plus HTML5 cross-platform Web-based OS technology, help industry players optimize hardware and software resources for integrated applications," said Foxconn iDSBG (Innovation Digital System Business Group) senior director William Liang at the Embedded Technologies and Applications Forum that took place in Taipei on January 22, 2014.Embedded smart systems and high-speed wireless networks bring about innovative applicationsSmart devices are made up of embedded systems in combination with sensors and Internet-based communication capability, Liang indicated. Smart applications cover three main areas. (1) IoT and sensing devices that cover physiological sensing, smart home, surveillance, automotive electronics, transportation, smart grids and atmospheric detection. (2) Smart terminal devices that provide information, daily-life assistance, home care, smart assistance, interactive operation, decision-making management and dashboard applications. (3) Cloud computing: Big Data collecting and intelligent analysis. IoT, the Internet and high-speed wireless networks are the backbone of daily smart applications, and 4G LTE is crucial to further development of such smart applications.According to Liang, smart devices come in two broad categories. (1) Embedded smart devices: environmental sensing/control devices, and wearable devices, such as smart bracelets, smart watches, smart glasses, smart toys, smart cars, smart home appliances, robots and surveillance devices. (2) Smart terminal devices: smart access and control platforms (the eight screens). The first and second of the eight screens are the handset and tablet respectively, while the third and fourth are the notebook and desktop PC. The fifth is the connected TV, the sixth the connected and smart TV, the seventh the electronic white board and the eighth the outdoor digital signage.According to forecasts by IDC, EmbeddedWorld and BusinessInsider, the global M2M market value will come close to US$7 billion in 2014, while global smart device shipments reached 1.738 billion units in 2013 and accumulated shipments will reach nine billion units by 2018.Smart applications: Integration of 8 screens, Internet, cloudBig Data is formed by tracing, gathering, tabulating and analyzing Small Data from sensor networks and IoT through the cloud. Cloud computing-based Big Data and smart applications can provide more accurate information on user behavior than conventional statistical methods, and such information can facilitate the development of customized services and new business opportunities, Liang said. Coming after cloud computing will be cognitive computing and artificial intelligence (AI). Cognitive computing can be applied to environmental estimation and prediction, and it can be used to remind users of routine work and offer recommendations. While cognitive computing is expected to be widely in use in the next three to five years, AI will eventually become the most important application following cloud and cognitive computing. Along with increasing 4G LTE bandwidth, cloud computing-based AI data processing will be increasingly feasible.Judging from the industry's development from the pre-2000 PC era, through the age of the Internet to the present stage of cloud computing, it is definitely trending towards AI applications. The future of Big Data, cognitive computing and AI will be materialized through the integration of the eight screens. And eight major segments of applications accompanying the trend are identified as follows: work, education, entertainment, home/social networking, security, health care, property transactions, and eco-friendly vehicles.For microprocessors and MCUs (microcontroller units) used in smart devices, entry-level models (for sensing nodes) include ARM Cortex-Mo, 8051; mid-range models (such as those for wearable devices) include ARM Cortex-R, Intel Quark and MIPS; and high-end models include ARM Cortex-A, x86/64 and NIPS for terminal devices and IA64, PowerPC, x86/64, ARMv8 and GP/GPU for cloud computing. Software platforms for smart applications include TinyOS, other RTOS, Linux for sensing devices, Android, iOS, Web-based OS, Windows for terminal devices and Linux for cloud computing.In terms of complexity of smart devices' computing work, there are three levels. (1) Low-level complexity: collecting sensing data, data pre-processing and communication among devices. (2) Medium-level complexity - voice recognition, facial detection, multimedia and UI (user interface) processing. (3) High-level complexity (cloud-based intelligent analysis) - semantic analysis, image recognition, sensing data analysis and prediction, user behavior prediction, commercial demand analysis, scientific computing and advanced AI application.The high performance of smart terminal devices' integrated computing relies on optimization through tight integration of hardware and software. Such integration includes: optimization of graphics and UI functions based on GPU; HSA of CPU and GP/GPU; and collaborative data processing where low-power sensing devices and high-performance smart terminal devices screen and pre-process data prior to cloud-based Big Data processing, computing and analysis.Development of operating systems for smart devicesWith abundant software support, Android is currently the most competitive operating system for high-end embedded smart devices. Its ecosystem is expanding fast, facilitated by a myriad of Android apps, Liang indicated. Of those "eight screens" currently handsets and tablets are mostly touch controlled in a single-person multi-tasking mode, with their applications focusing on communication, information sharing and personalized work. Notebooks and desktop PCs mostly work on keyboard input in a multi-person multi-tasking mode and they are mostly for work purposes. Connected TVs, portable TVs, electronic white boards and digital signage work through remote control or body movement sensing. They support single- or multi-tasking by multiple persons and they are chiefly meant for sharing, multimedia and entertainment purposes. Obviously, different OS are developed for different types of terminal devices. But Foxconn, eyeing a cross-platform for all "eight screens," has adopted HTML5 standardized by W3C (World Wide Web Consortium) to facilitate cloud-based Big Data services. HTML5 technology is expected to be used to develop next-generation operating systems and applications.Of the smart device operating systems, Linux and Windows belong to OS1.0 with emphasis on hardware resource management, while iOS and Android are the key players in the OS2.0 era focusing on software resource management. The Web-based Firefox OS running on HTML5 technology can be considered OS3.0 with emphasis on cloud resource management. There are more than four billion web pages around the world, which can be seen as a large volume of potential Web apps.HTML5 is gifted with the capability of cloud integration, Liang pointed out. HTML5 API specifications cover MultiTouch, WebTelephony, WebSMS, WebBluetooth, WebGL, Camera, GeoLocation, BatteryAPI, WebVibration, WebNFC, WebUSB, WebContact, FullScreen AP, and Setting AP. While HTML5 technology already provides sufficient support for smart terminal devices and applications, additional related Web API service technologies are still being developed.Foxconn, Mozilla team up to promote Firefox OSThe advantages of Android OS come mainly from its hierarchical design that facilitates collaboration among related industries, and its abundant software support that allows for easy development and replaceability of applications. However, Android has a disadvantage in inefficient execution of applications. Firefox OS is a completely open Web-based OS jointly developed by many experts around the world borrowing from Linux's open-source concept. HTML5 is expected to facilitate tight integration of Firefox OS and hardware to optimize functionality.Having signed a cooperation agreement on June 3, 2013, Foxconn and Mozilla on January 7, 2014 jointly announced Firefox OS Contribution Program to provide software resources and trial tablets for developers of Firefox OS-based applications in a bid to accelerate Firefox OS' penetration in the tablet market and the establishment of an ecosystem.Foxconn thinks that HTML5 is a technology for developing a next-generation OS for the integration of the "eight screens, the Internet and cloud." The cooperation with Mozilla has allowed Foxconn to obtain Firefox OS core technology to produce optimized system products with highly integrated hardware and software. Foxconn is establishing the world's largest HTML5 and cloud-based application R&D center in the Kaohsiung Software Park in southern Taiwan. It is looking to attract developers of HTML5-based cross-platform applications and create related business opportunities.Foxconn iDSBG (Innovation Digital System Business Group) senior director William Liang