Yoshiaki Tadokoro

A compressed digital output CMOS image sensor with analog 2-D DCT processors and ADC/quantizer

Progress in CMOS-based image sensors is creating opportunities for a low-cost, low-power one-chip video camera with digitizing, signal processing and image compression. Such a smart camera head acquires compressed digital moving pictures directly into portable multimedia computers. Video encoders using a moving picture coding standard such as MPEG and H.26x are not always suitable for integration of image encoding on the image sensor, because of the complexity and the power dissipation. On-sensor image compression such as a CCD image sensor for lossless image compression and a CMOS image sensor with pixel-level interframe coding are reported. A one-chip digital camera with on-sensor video compression is shown in the block diagram. The chip contains a 128/spl times/128-pixel sensor, 8-channel parallel read-out circuits, an analog 2-dimensional discrete cosine transform (2D DCT) processor and a variable quantization-level ADC (ADC/Q).

Simple Voltage-to-Time Converter With High Linearity

A simple voltage-to-time converter having a high linearity of time/voltage relationship is described. This circuit consists of one monostable multivibrator with a constant current source. The circuit does not have a ramp generator and a voltage comparator. The linearity error of the time/voltage relationship is less than ??0.05 percent for a voltage variation from 0.5 to 14 V.

Multiwave computing circuits using integrated opto-electronic devices

VLSI technology provides the means for implementing parallel systems with large numbers of computing elements. However, the high cost of communication in area, time, and power, relative to that of logic and storage, constrains the design and engineering of large-scale tightly-coupled systems. The problem severity increases as die size increases, and is at its worst in wafer-scale integration (WSI) and multichip module (MCM) packaging. Multiwave optical computing, where discrete wavelengths are employed as multiplexable information carriers, presents an interesting solution to the communication crisis in next-generation integrated systems. A computer architecture using multiwavelength opto-electronic integrated circuits (multiwave OEICs) provides the wavelength space as an extra dimension of freedom for parallel processing. A key feature is that several independent computations can be performed in a single optical circuit using wavelength space, as if it were several computing circuits operating in parallel.<<ETX>>

Simple Magnetic Delay unit with High Accuracy and Reliability For Digital Filters

The delay unit for digital filters can be realized with hardware in either of two ways, i.e., the analog and digital types. If accuracies between 0.1 and 1 percent are required, then it is known that the analog methods simplify the construction of the digital filters. This paper presents the delay unit for the digital filters using a single core which circumvents the problem of core matching. The magnetic delay unit has essentially no practical limitation on the long delay time. Therefore, the digital filters for very low frequencies can be easily realized. The characteristics of the core material to be required for such a unit are discussed and a method of practical realization is shown. In the practical unit, the delay time of 1 hour is realized without trouble. As to the relation between the input and output signals, an accuracy of within approximately 0.05 percent can be maintained when the whole supply voltage is changed by ±4 percent. Under the conditions of the temperature change from 20 to 50°C, there appears no drift caused by the change of temperature. For large variations of time during 24 hours, the drift is not observed experimentally, so that the above mentioned accuracy can be maintained. These facts show that the delay unit is fairly stable and reliable. As an example of a digital filter to be realized for the noise rejection, the simple first-order low-pass filter has been constructed by the method of the bilinear transformation.

Simple real-time Fourier analyzer and its application to analysis of magnetic torque curves

A method for the simple construction of the discrete Fourier analyzer operating in real time is described. The simplicity is realized by using two basic analog elements, i.e., magnetic analog memory elements and Fourier coefficient calculators using operational amplifiers. Practically, the real-time Fourier analyzer for the calculation of the Fourier coefficients with a scheme of 12 ordinates is constructed. The zero drift of this analyzer is ± 0.1 percent/hour for the full scale. The minimum amplitude of analyzable input signal is about 0.02 in ratio to the maximum amplitude. The maximum sampling frequency is approximately 3 kHz and the minimum sampling frequency has no practical limitation. As an example for its application, the magnetic torque curves are analyzed in on-line real time. The analyzed results almost agree with the results calculated by a general-purpose computer in off-line. This real-time Fourier analyzer is simple and suitable especially for the analysis of the signals changing slowly.

A New Magnetic-Coupled Monostable Multivibrator with Variable an Wide Pulsewidt

It is well known that magnetic-coupled monostable multivibrators have isolated and high-power outputs. These multivibrators which have been presented so far employ the operation of the major hysteresis loop. In these multivibrators, however, the output pulsewidth cannot be controlled over wide ranges and is severely influenced by variations in the source voltage. These problems can be solved by using the operation of the minor hysteresis loop and an R-C network to determine the pulsewidth. In this circuit, the output pulsewidth and the magnitude of output voltage can be controlled independently. The pulsewidth is expressed as T0 = CR In 2 until the magnetic core is saturated and easily adjusted by making either the resistor or capacitor variable. The magnitude of output voltage can be varied by the winding ratio and by the source voltage. A maximum-to-minimum pulsewidth ratio can be extended to about 1000. This value is 100 greater than that of the circuit employing the operation of the major hysteresis loop. The linearity error of the R-T0 characteristics is less than ± 0.1 percent. Its load characteristics are much better than those of conventional transistor multivibrators. The stability for the change of the source voltage is reduced to 0.06 percent/V over the range 4 to 12 V.