Consumer electronic appliances
The desire of consumers and business professionals to be connected anytime and anywhere is driving the sale of handheld micro-notebooks, set-top boxes, Internet appliances, and client computing devices. Often referred to as consumer electronic appliances (CEAs), such products feature different technology requirements from the personal computer industry that has dominated the marketplace this past decade. Specifically, CEAs call for:
Total system power consumption less than 4W
Support for a variety of processor architectures
Very small form factors weighing less than 1.5 kg
The ability to assign and partition the right amount of processing power for the job
Low heat dissipation
Market research firm International Data Corp. predicts that U.S. consumers will purchase almost as many information appliancs as personal computers by 2002.
One system-oriented silicon solution designed to address these requirements is an integrated 2D accelerated graphics and display sub-system from Media Q Inc. The company's MQ-200 product is pitched as the first device to use embedded DRAM technology for the CEA marketplace. Its 2 Mbytes of embedded DRAM eliminates the requirement for dedicated display memory, resulting in a 12-×12-mm BGA package. Not only is the MQ-200 one of the smallest of any embedded systems display devices available, it is one of the first to integrate the frame-buffer directly into the architecture. These design features help the MQ-200 maximize bandwidth and performance with minimal power consumption.
In addition, the MQ-200 supports multiple CPU interfaces, including the Hitachi SH-7750, Intel StrongARM SA-11xx series, the NEC VR-41xx MIPS family, and Toshiba's TR-39xx MIPS processors. It also has a PCI interface, giving it the ability to work with processors like Transmeta's Crusoe and AMD's Elan series.
To ensure graphics operations are not adversely affected by display refresh requirements, the MQ-200 features a 128-bit graphics engine; two on-board graphics controllers share the 1.6 G bytes/sec on-chip memory bus for as much as 6×the graphics performance of currently available products.
Simultaneous display capability, furthermore, lets users independently program the primary LCD and optional CRT displays at different resolutions, refresh rates, and pixel depths. Microsoft Pocket PowerPointusers, for example, are given the option to display presentation slides on the CRT while notes are shown on the LCD.
Mobile PC Companion's independent display controller and dual-display system allow direct interface to a broad range of liquid crystal displays and CRT monitor displays.
Finally, low-power design techniques extend battery life. Built-in intelligence, for example, ensures minimum power dissipation for all active operations and prevents power consumption in inactive modules. Five power-saving modes and alternate window capability for standby modes are available, as is configuration software, to further reduce power consumption.
"Media Q is one of the few companies focused on the needs of the consumer electronic appliance market," states Seiichi Hirai, general manager, Semiconductor Group, Toshiba Corp. "Earlier silicon offerings for handheld devices have attempted to satisfy this market's difficult cost, high performance, portability, and power requirements. By using embedded DRAM, innovative power management, and a high-performance graphics engine, the MQ-200 allows a new generation of richly featured devices to come to market."
More and more of today's information is created, edited, and transmitted in the digital domain. Ironically, the majority of video displays are still based on analog technology more than a century old—the cathode ray tube, or CRT. This, needless to say, severely limits the full potential of digital video.
DLP integrates a projection lamp and an electronic video signal from a source such as a VCR or computer. The processed light, reflected by the DMD, produces an all-digital picture.
One notable exception has been the introduction, beginning in 1996, of ultra-portable and micro-portable projectors. Based on Digital Light Processing™(DLP) technology from Texas Instruments, these projectors weigh as little as 1.3 kg, yet offer brightness as high as 1,300 ANSI lumens. Now manufactured by 3M, NEC , Sharp, Pulse, Philips, Mitsubishi Electronics, and others, such projection systems were the talk of Infocomm 2000, held last June in Anaheim, California.
As explained by TI engineer Lars Yoder, DLP is an optical system driven by digital electronics. Its enabling component, called a Digital Micromirror Device™(DMD), acts as a semiconductor light switch. The DMD is an array consisting of thousands of microscopic mirrors fabricated on hinges atop a static random access memory (SCRAM). The hinges allow each mirror to tilt between an "on" or "off" state. When the mirrors are not operating, they sit in a "parked" state at 0 degrees.
To operate, DLP integrates a projection lamp and a digital or an analog video signal from a VCR, computer, or similar source. Analog signals are converted to digital in the DLP's or the original equipment manufacturer's front-end processing. Once in digital format, the video or graphic signal passes to the DMD where each pixel of information is mapped directly to its own mirror in a 1:1 ratio.
DLP pixels are more uniform and more closely spaced than polycrystalline silicon pixels, offering a higher fill factor and better perceived resolution.
If the signal is 640×480 pixels, notes Yoder, the central 640×480 mirrors on the device will be active. The mirrors outside of this area are turned to the off position.
Electrically addressing the memory cell below each mirror with the binary bit plane signal electrostatically tilts each mirror on the DMD array to the on or off position at speeds greater than 1,000 times per second. Pulse width modulation determines how long each mirror tilts in either direction.
At this point, DLP becomes a simple optical system. Light from the projection lamp is directed at the DMD. Mirrors in the on position reflect light through a projection lens onto a screen to form a digital, square-pixel projected image.
The inherent digital nature of DLP enables noise-free, precise image quality with digital gray-scale and color reproduction. Compared to liquid crystal display (LCD) technology, DLP is more efficient because it is based on the reflective DMD and does not require polarized light. Finally, close spacing of the micromirrors causes video images to be projected as seamless pictures with higher perceived resolution.
As the final link in the chain that makes possible a completely digital video information structure, DLP technology appears uniquely positioned for HDTV as well as the eventual convergence of TVs and PCs.
Speech recognition and synthesis—built into products ranging from telephony devices to command and control applications for household appliances to security systems using voice password protection—promise many benefits to society at large. For specific populations such as the elderly or sight impaired, these products prove even more useful.
One example of a product making high-quality, low-cost speech recognition possible is an 8-bit microcontroller developed by Sensory Inc. Part of the company's Interactive Speech™product line, the RSC-300/364 is a single-chip solution that combines the flexibility of a microcontroller with advanced speech technology. End products can use one or all of the RSC-300/364 features in a single application, including speech recognition, speech and music synthesis, speaker verification, and voice record/playback.
Speech recognition. A neural network performs speaker-independent or speaker-dependent speech recognition. The former requires on-chip or off-chip ROM to store the words to be recognized; the latter requires external memory (e.g., SRAM, optional Serial EEPROM, or Flash) to store speech recognition information.
Speech and music synthesis. The RSC-300/ 364 uses a MIDI-like system to generate music. Speech synthesis requires on-chip or off-chip ROM to store audio sounds.
Speaker verification. After a speaker trains the chip on a specific word, the chip is able to identify the original speaker by that word.
Record/playback. Depending on the quantity and quality of playback desired, the RSC-300/364 can perform audio record and playback at various compression levels.
Chief components include the 8-bit variable-length-instruction microcontroller, on-chip A/D and D/A converters, an input amplifier, PWM speaker output, plus RAM (2.5 kBytes) and ROM (64 kBytes, RSC-364 only). The product also offers the ability to address off-chip RAM or ROM.
During operation, an external microphone passes an audio signal to the preamplifier and A/D converter to change the incoming speech signal into digital data. Output audio signals are derived from a D/A converter or PWM. Typical operating current is 10 mA at 14.32 MHz and 3V.
Creating applications using the RSC-300/364 requires the development of electronic circuitry, software code, and speech/music data files. One recent example: a voice-activated, hand-held phone book and auto dialer produced by Humanity-Oriented Technology Corp. of Taiwan.
Based on the RSC-300/364, "Voicedex" combines all the functions of a traditional electronic telephone book, such as the storage and retrieval of names and phone numbers, with speech recognition technology as the user interface. Software code for the RSC-300/364 can be developed by Sensory or by external programmers using the RSC-300/364 Development Kit.
recognition DSP alliance
Toshiba, a supplier of telephone answering device chips in Japan, and Sensory Inc. announced earlier this year a joint agreement to produce speech recognition-capable DSPs for telephony products. The new Toshiba/Sensory alliance will produce DSP chips featuring voice activated dialing for totally hands-free operation of digital telephone answering machines, speakerphones, and other telephony devices.
Toshiba has decided to introduce voice recognition into speakerphones and telephone answering devices to improve functionality with increased ease of use. "We want to make speech recognition available to our consumers at a reasonable cost," says Yasushi Hirata, manager of Toshiba's Telecom Systems LSI Department. For Sensory, the Toshiba partnership represents a move into the software licensing business. Sensory has taken the core technology out of its low-cost ICs, and has strengthened the capabilities to run on DSPs such as Toshiba's.
"We believe speech recognition is the next direction for human machine interfaces and it is exciting to have both IC and software solutions available," adds Erik Soule, Sensory's director of marketing.