• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.25 no.11
• /
• pp.1627-1634
• /
• 2021

This paper introduces a low-power compact ADC circuit for analog Convolutional filter for low-power neural network accelerator SOC. While convolutional neural network accelerators can speed up the learning and inference process, they have drawback of consuming excessive power and occupying large chip area due to large number of multiply-and-accumulate operators when implemented in complex digital circuits. To overcome these drawbacks, we implemented an analog convolutional filter that consists of an analog multiply-and-accumulate arithmetic circuit along with an ADC. This paper is focused on the design optimization of a low-power 8bit SAR ADC for the analog convolutional filter accelerator We demonstrate how to minimize the capacitor-array DAC, an important component of SAR ADC, which is three times smaller than the conventional circuit. The proposed ADC has been fabricated in CMOS 65nm process. It achieves an overall size of 1355.7㎛², power consumption of 2.6㎼ at a frequency of 100MHz, SNDR of 44.19 dB, and ENOB of 7.04bit.

• ETRI Journal
• /
• v.30 no.6
• /
• pp.826-832
• /
• 2008

The metal-ferroelectric-metal (MFM) capacitor in the ferroelectric random access memory (FeRAM) embedded RFID chip is used in both the memory cell region and the peripheral analog and digital circuit area for capacitance parameter control. The capacitance value of the MFM capacitor is about 30 times larger than that of conventional capacitors, such as the poly-insulator-poly (PIP) capacitor and the metal-insulator-metal (MIM) capacitor. An MFM capacitor directly stacked over the analog and memory circuit region can share the layout area with the circuit region; thus, the chip size can be reduced by about 60%. The energy transformation efficiency using the MFM scheme is higher than that of the PIP scheme in RFID chips. The radio frequency operational signal properties using circuits with MFM capacitors are almost the same as or better than with PIP, MIM, and MOS capacitors. For the default value specification requirement, the default set cell is designed with an additional dummy cell.

• 제어로봇시스템학회:학술대회논문집
• /
• 2003.10a
• /
• pp.1066-1070
• /
• 2003

Recently technologies have created new principle and theory but the PID control system remains its popularity as the PID controller contains simple structure, including maintenance and parameter adjustment being so simple. Thus, this paper proposes auto tune PID by fuzzy logic controller based on FPGA which to achieve real time and small size circuit board. The digital PID controller design to consist of analog to digital converter which use chip TDA8763AM/3 (10 bit high-speed low power ADC), digital to analog converter which use two chip DAC08 (8 bit digital to analog converters) and fuzzy logic tune digital PID processor embedded on chip FPGA XC2S50-5tq-144. The digital PID processor was designed by fundamental PID equation which architectures including multiplier, adder, subtracter and some other logic gate. The fuzzy logic tune digital PID was designed by look up table (LUT) method which data storage into ROM refer from trial and error process. The digital PID processor verified behavior by the application program ModelSimXE. The result of simulation when input is units step and vary controller gain ($K_p$, $K_i$ and $K_d$) are similarity with theory of PID and maximum execution time is 150 ns/action at frequency are 30 MHz. The fuzzy logic tune digital PID controller based on FPGA was verified by control model of level control system which can control level into model are correctly and rapidly. Finally, this design use small size circuit board and very faster than computer and microcontroller.

• Proceedings of the IEEK Conference
• /
• 2004.06b
• /
• pp.533-536
• /
• 2004

In this paper, CMOS A/D converter with 10bit 32MSPS at 3.3V is designed for HPNA 2.0. In order to obtain the resolution of 10bit and the character of high-speed operation, we present multi-stage type architecture. That consist of sample and hold(S&H), 4bit flash ADC and 4bit Multiplier D/A Converter (MADC) also the Overflow and Underflow for timing error correct of Digital Correct ion Logic (DCL). The proposed ADC is based on 0.35um 3-poly 5-metal N-well CMOS technology. and it consumes 130mW at 3.3V power supply.

• Proceedings of the IEEK Conference
• /
• 2005.11a
• /
• pp.497-500
• /
• 2005

This paper presents a design of a hybrid sensor chip which integrates an A/D converter module and a phase measurement module for measuring power line phase. Recursive sliding DFT based phase measurement module is designed using time shared multiplier which can reduce the size of SoC implementation. A/D converter is based on the sigma delta modulation in order to minimize the implementation space of the analog part and designed to obtain 8-bit resolution. Computer simulations and FPGA implementation are performed to verify hybrid sensor chip design. The hybrid sensor chip for 4-channel power line phase measurement is fabricated by using 0.35 micrometer CMOS process.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.33B no.3
• /
• pp.186-192
• /
• 1996

A modified learning rule is introduced for the implementation of feedforward artificial neural networks with on-chip learning circuitry using standard analog CMOS technology. Learning rule, is modified form the EBP (error back propagation) rule which is one of the well-known learning rules for the feedforward rtificial neural nets(FANNs). The employed MEBP ( modified EBP) rule is well - suited for the hardware implementation of FANNs with on-chip learning rule. As a ynapse circuit, a four-quadrant vector-product linear multiplier is employed, whose input/output signals are given with voltage units. Two $2{\times}2{\times}1$ FANNs are implemented with the learning circuitry. The implemented FANN circuits have been simulatied with learning test patterns using the PSPICE circuit simulator and their results show correct learning functions.

• Proceedings of the IEEK Conference
• /
• 2002.07b
• /
• pp.1030-1033
• /
• 2002

This paper, a new simple controller operates in continuous conduction mode (CCM) for Boost power factor collection converter is introduced. The duty ratios are obtained by comparisons of a sensed signal from inductor current and a negative ramp carrier waveform in each switching period. By using the proposed controller, input voltage sensing, error amplifier in the current feedback loop, and analog multiplier／divider are not required, then, the control circuit implementation is very simple. To verify the proposed controller, the circuit simulation for Boost power factor correction converter was applied. For the results, the input current waveform was shaped to be closely sinusoidal, implying low THD.

• Journal of the Institute of Convergence Signal Processing
• /
• v.14 no.4
• /
• pp.249-254
• /
• 2013

A fast multiplier and ADC are essential to process the analog signals in real time. The double-base number system(DBNS) is known as an efficient method for this purpose. The DBNS uses the numbers 2 and 3 as the base numbers simultaneously. The system has an advantage of fast multiplication, less chip area, and low power consumption compared to the binary multiplier. However, the inherent errors of the log number's intrinsic tolerance in DBNS are accumulated in a FIR digital filter, so the signal-to-noise ratio(SNR) has a tendency to be degraded. In this paper, the nonlinear encoder of ADC is designed to compensate the accumulated errors of DBNS by analysing the error distributions of various filter coefficients. The new ADC does not sacrifice its own advantages because the encoder circuits are modified only. The experiments were done with an FIR filters those were designed to have -70dB of SNR in stop band. The proposed nonlinear ADC encoder could drop the SNR to -45dB in stop band, in contrast to -35dB with the linear encoder.

• Journal of the Korean Magnetics Society
• /
• v.12 no.6
• /
• pp.206-211
• /
• 2002

We have constructed a new kind of magnetic motion capture sensor based on the pulse magnetic field method. 3-orthogonal magnetic pulse fields were generated in turns only one period of sinusoidal waveform using 3-orthogonal magnetic dipole coils, ring counter and analog multiplier. These pulse magnetic fields were measured with 3-orthogonal search coils, of which induced voltages by the x-, y-, and l-dipole sources using S/H amplifier at the time position of maximum induced voltage. Using the developed motion capture sensor, we can measure position of sensor with uncertainty of ${\pm}$0.5％ in the measuring range from ${\pm}$0.5 m to ${\pm}$1.5 m.

• The Bulletin of The Korean Astronomical Society
• /
• v.37 no.2
• /
• pp.209.2-209.2
• /
• 2012

UFFO Burst Alert & Trigger telescope (UBAT) is one of major instruments of UFFO-Pathfinder. The UBAT aims at 10 arcmin resolution localization of Gamma Ray Bursts with X-ray coded mask technique. It has $400mm{\times}400mm$ coded mask aperture, hopper, shielding and detector module with effective area of $191cm^2$. The detector module consists of an assembly of 36 64-ch MAPMTs and $25mm{\times}25mm$ pixellated YSO crystal array, and associated analog and digital electronics of about 2500 channels. We performed a vibration test using a dummy MAPMT with the detector module structure to measure the indused stress applied onto the MAPMT. We designed a sub-structure on the detector module to avoid the resonance that would otherwise deforms the detector module structure. A finite element analysis confirms the reduction of the load acceleration down to 12g. The experimental results are to be reported. Consequently, it proves that the MAPMT arrays of the flight UBAT detector module structure would survive in the space launch environment.