1,338 research outputs found
81 fJ/bit energy-to-data ratio of 850 nm vertical-cavity surface-emitting lasers for optical interconnects
No full textThis article may be downloaded for personal use only. Any other use requires prior permission of the author and AIP Publishing. This article appeared in Appl. Phys. Lett. 98, 231106 (2011) and may be found at https://doi.org/10.1063/1.3597799.Extremely energy-efficient oxide-confined high-speed 850 nm vertical-cavity surface-emitting lasers for optical interconnects are presented. Error-free performance at 17 and 25 Gb/s via a 100 m multimode fiber link is demonstrated at record high dissipation-power-efficiencies of up to 69 fJ/bit (<0.1mW/Gbps) and 99 fJ/bit, respectively. These are the most power efficient high-speed directly modulated light sources reported to date. The total energy-to-data ratio is 83 fJ/bit at 25°C and reduces to 81 fJ/bit at 55°C. These results were obtained without adjustment of driving conditions. A high -factor of 12.0GHz/(mA)0.5 and a -factor of 0.41 ns are measured.EC/FP7/224211/EU/VISIT - Vertically Integrated Systems for Information Transfer/VISITDFG, 43659573, SFB 787: Halbleiter - Nanophotonik: Materialien, Modelle, Bauelement
Single-drive high-speed lumped depletion-type modulators toward 10 fJ/bit energy consumption
No full textReduction of modulator energy consumption to 10 fJ/bit is essential for the sustainable development of communication systems. Lumped modulators might be a viable solution if instructed by a complete theory system. Here, we present a complete analytical electro-optic response theory, energy consumption analysis, and eye diagrams on absolute scales for lumped modulators. Consequently the speed limitation is understood and alleviated by single-drive configuration, and comprehensive knowledge into the energy dependence on structural parameters significantly reduces energy consumption. The results show that silicon modulation energy as low as 80.8 and 21.5 fJ/bit can be achieved at 28 Gbd under 50 and 10 Omega impedance drivers, respectively. A 50 Gbd modulation is also shown to be possible. The analytical models can be extended to lumped modulators on other material platforms and offer a promising solution to the current challenges of modulation energy reduction. (C) 2017 Chinese Laser PressNational Natural Science Foundation of China (NSFC) [61120106012]SCI(E)ARTICLE2134-142
Author Arna Bontemps reads to children at the East Winston Branch Library, 1956.
No full textAuthor Arna Bontemps reads to children at the East Winston Branch Library, 1956
SPIN AND CHARGE FLUCTUATIONS IN ITINERANT-ELECTRON MAGNETISM - A VARIATIONAL FUNCTIONAL-INTEGRAL APPROXIMATION
No full textThe single-site spin-fluctuation theory of metallic magnetism developed by Hubbard and Hasegawa is modified and extended to investigate the effects of spin and charge fluctuations on the magnetic properties at finite temperature. The key point of our scheme is to adopt a continuous decomposition of the interacting part of the Hubbard Hamiltonian in the Stratonovich-Hubbard transformation, resulting in the separation of the charge fluctuations from the spin fluctuations. This separation makes a better description for the contrary nature between the two types of fluctuations. In order to remove the ambiguity, a variational principle is introduced to determine the appropriate form of the decomposition of the Hubbard model. The ground-state properties, the Curie temperature, the magnetization, and the paramagnetic susceptibility, etc., at finite temperature are calculated and compared with the Hasegawa theory. The improvement of the Curie temperature is significant, and better behavior of the temperature dependence of the magnetization is obtained.Physics, Condensed MatterSCI(E)6ARTICLE1712795-128044
Two-Photon AMPK and ATP Imaging Reveals Metabolic Recovery in Mouse Rod Photoreceptor Cells
No full textThe title of publisher's version: Two-photon AMPK and ATP imaging reveals the bias between rods and cones in glycolysis utility.In vertebrates, retinal rod and cone photoreceptor cells rely significantly on glycolysis. Lactate released from photoreceptor cells fuels neighboring retinal pigment epithelium cells and Müller glial cells through oxidative phosphorylation. To understand this highly heterogeneous metabolic environment around photoreceptor cells, single-cell analysis is needed. Here, we visualized cellular AMP-activated protein kinase (AMPK) activity and ATP levels in the retina by two-photon microscopy. Transgenic mice expressing a hyBRET-AMPK biosensor were used for measuring the AMPK activity. GO-ATeam2 transgenic mice were used for measuring the ATP level. Temporal metabolic responses were successfully detected in the live retinal explants upon drug perfusion. A glycolysis inhibitor, 2-deoxy-d-glucose (2-DG), activated AMPK and reduced ATP. These effects were clearly stronger in rods than in cones. Notably, rod AMPK and ATP started to recover at 30 min from the onset of 2-DG perfusion. Consistent with these findings, ex vivo electroretinogram recordings showed a transient slowdown in rod dim flash responses during a 60-min 2-DG perfusion, whereas cone responses were not affected. Based on these results, we propose that cones surrounded by highly glycolytic rods become less dependent on glycolysis, and rods also become less dependent on glycolysis within 60 min upon the glycolysis inhibition
A low noise, low power dynamic amplifier with common mode detect and a low power, low noise comparator for pipelined SAR-ADC
No full textThis thesis presents a high gain, low noise and low power dynamic residue amplifier and a low power, low noise dynamic comparator designed in TSMC 28nm process for a two step Pipelined SAR-ADC. The cascoded integrator dynamic residue amplifier (CIDRA) achieves a gain of 30dB with THD of 47dB (11 mV pp input). The input referred noise across tem- perature and process corner is 55 µV and it operates at a frequency of 500MHz while the energy consumption is 390 fJ. The low power and low noise pseudo-latch preamp dynamic comparator (PLPDC) shows a delay of 250pSec for a differential input of 16 pV and consumes 91 fJ (current is 91 µA for 100 MHz clock) of energy. The input referred offset is 4 mV (?).Microelectronics & Computer EngineeringElectrical Engineering, Mathematics and Computer Scienc
P86 The Screening and Diagnosis of Cystic Fibrosis-Related Diabetes in the United Kingdom
No full textCryogenic Comparator Characterization and Modeling for a Cryo-CMOS 7b 1-GSa/s SAR ADC
No full textThis paper reports the experimental characterization and modelling of a stand-Alone StrongARM comparator at both room temperature (RT) and cryogenic temperature (4.2 K). The observed 6-dB improvement in the comparator input noise at 4.2 K is attributed to the reduction of the thermal noise and to the suppressed shot noise in the MOS transistors becoming dominant at cryogenic temperature. The proposed model is employed in the design of a loop-unrolled 2\times time-interleaved 1-GSa/s 7b SAR ADC for spin-qubit readout. As predicted by the comparator model, the ADC is noise-limited at RT to a SNDR of 38.2 dB at Nyquist input, while this improves to 41.1 dB at 4.2 K, now limited by distortion, thus resulting in the state-of-The-Art FoMw for cryo-CMOS ADC of 20.9 fJ/conv-step. Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.QCD/Sebastiano LabElectrical Engineering, Mathematics and Computer ScienceElectronicsQuantum Circuit Architectures and Technolog
A 10-b 330nW Third-Order Predictive SAR ADC Dedicated to Neural Recording Brain Implants
No full textThis paper reports on a predictive analog-todigital converter (ADC). The proposed ADC employs a linear predictive filter to prepare a prediction for the current sample based on the values of the previous digital codes. This leads to significant reduction in the mean bit cycle of the converter. It is shown in this work that this idea is significantly more effective for the digitization of biological signals (e.g., intra-cortical neural signals). Compared with other similar techniques available in the literature, the proposed predictive ADC is significantly more successful for small signal-to-noise ratios. The proposed algorithm results in 48% and 37% reduction in the converter’s mean bit cycle compared with the conventional and LSB-first structures, respectively. Designed and post-layout simulated in a 90-nm standard CMOS technology and operated at 200 kS/s with a supply voltage of 0.4 V, the 10-bit predictive ADC consumes 330 nW. The circuit occupies a core area of 0.025 mm2, achieves an ENOB of 9.42 bits, a figure-of-merit of 2.4 fJ/conv.-step, and an SFDR of 65.8 dB. The DNL and INL of the circuit are within 0.45 LSB and 0.56 LSB, respectively.Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.Bio-Electronic
Energy-Efficient SNN Implementation Using RRAM-Based Computation In-Memory (CIM)
No full textSpiking Neural Networks (SNNs) can drastically improve the energy efficiency of neuromorphic computing through network sparsity and event-driven execution. Thus, SNNs have the potential to support practical cognitive tasks on resource constrained platforms, such as edge devices. To realize this, SNN requires energy-efficient hardware which can run applications with a limited energy budget. However, the conventional CMOS implementations cannot achieve this goal due to the various architectural and technological challenges. In this work, we address these issues by developing an energy-efficient and accurate SNN hardware based on Computation In-Memory (CIM) architecture using Resistive Random Access Memory (RRAM) devices. The developed SNN architecture is based on unsupervised Spike Time Dependent Plasticity (STDP) learning algorithm with online learning capability. Simulation results show that the proposed architecture is energy-efficient with a consumption of ≈20 fJ per spike, while maintaining state-of-the-art inference accuracy of 95% when evaluated using the MNIST dataset.Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.Computer EngineeringQuantum & Computer Engineerin
- …
