This article introduces an FPGA-based implementation of a spiking neural network with multiple input multiple output (MIMO) neurons using a leaky integrate and fire (LIF) model. A weight update rule is applied in the LIF model to increase or decrease synaptic weight based on active or inactive neurons. Simulation results validate the working functionality of MIMO neurons.

This paper proposes a Differential-Input Body Bias Sense Amplifier (DIBBSA) with an auto-offset mitigation feature suitable for low-voltage SRAMs where the differential bitline signals are applied to the sources as well as to the body of the critical sensing transistors. We simulated and fabricated the proposed DIBBSA architecture with various operational modes in 65-nm CMOS technology to analyze body biasing's effectiveness in mitigating the offset. The standard deviation of offset (sigma(OS)) was measured over 5120 SAs in 10 ICs. The isogate area reduction in sigma(OS) for the proposed DIBBSA-FL and DIBBSA-PD modes resulted in 68.1% and 61.9% compared to conventional Current Latch SA (CLSA) and 24.1% and 18.1% compared to Voltage Latch SA (VLSA) at 0.4 V supply and 25 degrees C, respectively. Carried out measurements on 512 SAs in an IC show the minimum required differential input voltage across the temperature range of 0 degrees C to 75 degrees C at 0.4 V is achieved to be 48% lower compared to CLSA and 28% lower compared to VLSA by both the DIBBSA-FL and DIBBSA-PD modes.

Flip-flops are a key component of digital integrated circuits and substantially affect their power and energy consumption. In this paper, an ultra low-power, contention-free, static single-phase 3-transistors clock load flip-flop is described and referred to as 19-T Ultra Low-power Flip-flop (ULFF). Simulation results in CMOS 65 nm technology show that at nominal conditions and Data Activity (DA) of 10%, the ULFF has 56% and 7% low-power consumption compared to the 18-T Single-phase Clocked Static Flip-flop (18TSPC), and Topologically Compressed Flip-flop (TCFF), respectively. Similarly, ULFF has energy-efficiency comparable to Low-power at Low-data activity Flip-flop (LLFF) at low DAs, and 18TSPC at high DAs.

An array of 30 16-Mbit static random access memories manufactured using 65-nm technology has been designed to characterize the proton and neutron beams used for radiation testing at TRIUMF. Measurements include single-event upset cross sections, beam profiles, collimator design, and shielding effects. The sensitivity of this dosimeter allows for quick measurements of these parameters, which are of interest in radiation test measurements.

Sense amplifier (SA) input-referred offset often dictates the minimum required differential input ( $\Delta V_{\textrm {BL-min}}$ ) and is an important factor in realizing low-voltage static random access memories. This paper presents a HYbrid latch-type Sense Amplifier (HYSA-QZ), where the bitline signals are supplied to multiple internal nodes to significantly reduce $\Delta V_{\textrm {BL-min}}$ . A 65-nm CMOS test chip with arrays of the HYSA-QZ, two intermediate formulations of the HYSA-QZ, conventional current latch SA (CLSA), and conventional voltage latch SA (VLSA) were fabricated. Measurements over 5120 SAs of each type show that the HYSA-QZ implemented with regular- $V_{\mathrm{ T}}$ transistors require 50.0%, and 22.8% lower $\Delta \text{V}_{\textrm {BL-min}}$ with 6.5% (or 4.5%) and 30.7% (or 18.8%) of total gate (or layout) area overhead compared to CLSA and VLSA at 0.4 V, respectively. Iso-gate-area offset improvement was substantiated with Pelgrom’s mismatch model, where the HYSA-QZ with regular- ${\mathrm{ V}}_{\mathrm{ T}}$ transistors showed 46.6% and 7.7% improvements in measured standard deviation of offset distribution compared with CLSA and VLSA, respectively. Measured $\Delta V_{\textrm {BL-min}}$ for the HYSA-QZ remains stable and low over a temperature range from 0 °C to 75 °C at 0.4 V. Moreover, an additional 13.0% reduction in $\Delta V_{\textrm {BL-min}}$ was measured in the HYSA-QZ when using low- $V_{\mathrm{ T}}$ transistors. Finally, the HYSA-QZ operates reliably at $V_{\textrm {DD-min}}$ of 260 mV in 25 °C.

Abstract – We describe the chemical engineering knowledge base in terms of five distinct concept domains: i) mathematics and computation, ii) conservation, iii) equilibrium and spontaneity, iv) rates, and v) the structure and property of materials. These concept domains underpin the curriculum and evolve from disparate subject domains presented in the first year into a cohesive whole by graduation. The knowledge base for chemical engineering can thus be expressed in terms of achieving threshold concepts related to each of these domains. This formulation of the knowledge base suggests that it may be examined using concept inventory testing. We provide examples of how such testing can be implemented in order to produce meaningful data on students’ level of concept attainment. We believe that this approach may be of interest to others as a robust and sustainable method for the ongoing assessment of CEAB Graduate Attribute 1.

This paper uses the data from Student Performance Evaluation forms submitted by coop employers to study the level and progression of life-long learning attribute among engineering undergrad students at University of Waterloo. The questions in the survey are mapped to the indicators defining the lifelong learning. The measurements show excellent performance rating, progressing positively as students advancing in their program. We analyze the results and suggest future research to address the shortcomings observed in the survey.

Electrical engineering students at the University of Waterloo require opportunities to develop hardware workflow skills. For example, creating, populating, and debugging printed circuit boards. Students may be fortunate enough to gain these skills on co-op work terms, but there is no structured on-campus opportunity to learn these topics. These workflows cannot be added to the existing curriculum without displacing other valuable learning opportunities. An optional “Embedded μC PCB Activity” is being developed to fill this gap. The Kolb learning cycle and project-based learning principles have been applied in its creation, and the Kirkpatrick Model of Training Evaluation is used as a framework to assess student reaction and learning. The activity was first trialled and found unsuccessful in 2017 but is now prepared to undergo another trial due to iterative improvement.

Variability in offset voltage, bitcell transistor conductance, and leakage currents can lead to marginal and intermittent failures in low-voltage SRAMs. In this paper, we develop a model of these marginal faults that includes such sense amplifier and bitcell variability. Using simulations and measurement data from a 65 nm test chip, we investigate the likelihood of these failures and propose how to stimulate their occurrence during testing.

Abstract – To address the new process of graduate attributes (GAs) assessment as required by the Canadian Engineering Accreditation Board (CEAB), the University of Waterloo (UW) employed six Graduate Attributes Lecturers (GALs) and four Accreditation Assistants (AAs) with a key role of leading the outcomes assessment process in each of the engineering departments. The GALs work collaboratively with each other and their departments to come up with a process of outcomes assessment. The collaboration methods and techniques used by the GALs in developing shared indicators for the common GAs are proving to be highly effective, and have led to significant progress. One of these methods is a structured brainstorming sessions for developing measurable performance indicators for the common GAs. The following paper describes in detail the collaboration methods and techniques used by the GALs and AAs to develop shared indicators for the professional skills GAs. The paper also discusses the factors that proved to be successful in the whole process as well as the challenges faced by the team.

–At the University of Waterloo, 1B electrical and computer engineering students participate in a series of hands-on, open-ended design activities. In particular, a wind-your-own motor activity has been trialed on four occasions. The activity is widely recognized as being fun, but are the students designing or kludging? The Kirkpatrick Model of Training Evaluation was used as a framework to assess student perceptions of the activity and to guide focus group discussions. The activity was iteratively improved to maintain a positive reaction while increasing student learning.

Abstract – As the Canadian Engineering Accreditation Board moves toward outcomes-based assessment, engineering institutions are employing a variety of assessment tools and methods. Surveys are usually thought of as indirect measures. However, survey tools can easily provide direct measures as well, and can therefore simultaneously present direct and indirect assessments to students. This paper examines the benefits and possibilities of exploiting such simultaneous direct and indirect assessment.

Devices based on piezoelectric materials have traditionally been modeled in PDE simulation software. These simulations are expensive to create and run. In this paper it is shown that lumped-parameter models of such devices can provide good fidelity with low computational cost. Modelica models of supporting components, along with a system-level model of a linear piezoelectric stepper motor are presented. The simulation results show good agreement with published experimental results. Future research is proposed based on the components and model.

Multiple scattering theory is a versatile two-and three-dimensional method for characterizing the acoustic wave transmission through many scatterers. It provides analytical solutions to wave propagation in scattering structures, and its computational complexity grows logarithmically with the number of scatterers. In this paper we show how the 2D method can be adapted to include the effects of time-varying material parameters. Specifically, a new T-matrix is defined to include the effects of frequency modulation that occurs in time-varying phononic crystals. Solutions were verified against finite difference time domain (FDTD) simulations and showed excellent agreement. This new method enables fast characterization of time-varying phononic crystals without the need to resort to lengthy FDTD simulations. Also, the method of combining T-matrices to form the T-supermatrix remains unchanged provided that the new matrix definitions are used. The method is quite compatible with existing implementations of multiple scattering theory and could be readily extended to three-dimensional multiple scattering theory.

Phononic crystal effects are typically narrowband and highly dependent on the size and spacing of the scatterers within the crystal-properties that are fixed and unchangeable. We propose that it may be possible to dynamically alter the behaviour of phononic crystals by varying the material properties in time. In this paper we seek to extend the existing static phononic crystal theory so as to handle time-varying material parameters and also to provide insight into the factors that govern their behaviour. Beginning with the transmission matrix method, we develop a new method to determine the acoustic wave transmission through a time-varying corrugated tube waveguide, which is analogous to a one-dimensional (1D) phononic crystal. These expressions are further developed for periodic material parameter variation signals. Our method permits the development of a closed-form solution to the acoustic wave transmission through a 1D time-varying phononic crystal. It shows excellent agreement with finite-difference time- domain simulations and with execution times several orders of magnitude faster. Preliminary results show that the band-gap properties can be significantly altered using material parameter time variation as the driving mechanism.

The past 10 years have seen remarkable developments in microwave, optical and acoustic methods that make use of the unique wave transmission properties of a class of ordered composites known as metamaterials. These materials differ from regular materials in that wave propagation is dramatically affected by the size and arrangement of their small-scale structure, and not just by the choice of constituent materials. Their characteristic feature sizes can be much less than, or on the same order as a wavelength. They have a variety of potential applications ranging from super-lenses, to enabling subwavelength imaging resolution, to the creation of a Harry Potter-like ‘invisibility cloak’. In this review we provide an overview of some of these achievements, much of which has been proposed and demonstrated with optical and electromagnetic waves. However, more recently these ideas have been extended to acoustic waves, offering the opportunity of controlling ultrasound wave propagation in a unique manner.

Ternary content-addressable memories (TCAMs) can perform high-speed and deterministic table lookups. However, its main drawback is high power consumption. A significant portion of the TCAM power is consumed by match-line sense amplifiers (MLSAs) for match detection. This paper presents a 20-kilobit TCAM featuring two MLSAs with positive-feedback techniques. The proposed circuits have been fabricated on a test chip in 0.18- mum CMOS technology. Energy measurement results of the two MLSAs show reductions of 56% and 48%, respectively, over the conventional current-race MLSA.

The acoustic properties of a photonic crystal is controlled by changing its material properties as a function of time, which affects the acoustic wave scattering within the crystal. An 1D transmission matrix method (TMM) is used to modulate the incident waves within a time-varying photonic crystal at every scattering interface. Acoustic wave propagation within a corrugated tube waveguide is analogous to acoustic wave propagation with a 1D photonic crystal consisting of alternating layers of differing media. The time varying-TMM (TV-TMM) simulation execution times are four to five orders of magnitude faster than the FDTD simulations used for comparison and provide exact solutions at the frequencies of interest. The shapes of the band-gap and band-gap edges are altered, which arises the possibility of changing the band-gap characteristics using material parameter variation as the controlling mechanism.

Phononic crystals demonstrate remarkable effects, but are static and unchangeable. We wish to use dynamic materials as the scatterers within phononic crystals as a means of controlling their properties. We have extended existing static phononic crystal theory to handle time-varying material parameters. The new method provides a closed-form solution to acoustic wave transmission through 1D dynamic phononic crystals. Band gap properties can be significantly altered using time-varying material parameters as the driving mechanism.