Search for: All records

Modern Systems on Chips (SoC) have more than one power domain for efficient power management across multiple functional blocks. In the design of Printed Circuit Board (PCB) pre-layout stage, engineers use physical insight and trial and error methods to place decaps to satisfy the target specifications needed for the SoC for efficient operation in the PCB. This is a time-consuming and ineffective method since there could be numerous possibilities. This could sometimes lead to overdesign or under-design for some power domains; repetitive iterations are required to overcome this challenge. This paper addresses a challenge not tackled at this scale before—specifically, managing multiple power domains in a practical setting. For the first time, we proposed a hybrid algorithm that combines multi-agent reinforcement learning with a genetic algorithm to effectively handle the complexity of decap pre-layout synthesis for multiple power domains. This work optimizes the placement, orientation, value, and number of decaps, offering a solution that accounts for the intricate dependencies and constraints inherent in PCB pre-layout designs, where no initial layout is available to begin the optimization process. This algorithm computes the PDN impedance using an in-house tool making it fast and completely free of other commercial simulation tools. The algorithm has been designed in a way to take inputs directly from a board file or with just ball map and PCB stackup information making it easier to use and automate with the existing frameworks used by engineers. The proposed algorithm is validated through boards with simple and practical SoC cases. In 30 minutes, the proposed algorithm provided a placement configuration that satisfied all 36 power domains of a practical SoC for mobile applications, while considering the practical constraints of the PCB. 

We present a cavity-enhanced solid-state nuclear spin gyroscope based on nitrogen-vacancy centers in diamond. With the two-field interference, we indicate a state-of-the-art rotation sensitivity and pave the way for advancements in high-performance quantum sensing. 

Power converter simulation models are crucial for power integrity simulation in the low- frequency range. In this regard, a behavioral model of multi-phase buck converter that accurately captures non-linear behaviors was proposed in the past. However, more than 20 parameters are needed to model practical multi-phase Buck converters with phase add/drop and PWM/PFM transition features, and determining those parameters is not trivial. In this paper, an automatic parameter extraction method was proposed. The proposed method constructs cost functions that quantify the difference between simulated and measured output voltage waveforms. Using the nonlinear least squares (LSQ) method, the algorithm iteratively minimizes the cost function to find the best estimates of the unknown parameters. SPICE simulations are conducted in each iteration to update the cost function. The proposed method was implemented in a Python program and validated on a practical three-phase buck converter. This program successfully extracted the correct parameters without the user’s intervention. The SPICE model matches measurement results well at various load conditions with the extracted parameters. This method is easy to use since it only requires a couple of oscilloscope-measured waveforms. 

Abstract Quantum sensors based on solid-state defects, in particular nitrogen-vacancy (NV) centers in diamond, enable precise measurement of magnetic fields, temperature, rotation, and electric fields. Cavity quantum electrodynamic (cQED) readout, in which an NV ensemble is hybridized with a microwave mode, can overcome limitations in optical spin detection and has resulted in leading magnetic sensitivities at the pT-level. This approach, however, remains far from the intrinsic spin-projection noise limit due to thermal Johnson-Nyquist noise and spin saturation effects. Here we tackle these challenges by combining recently demonstrated spin refrigeration techniques with comprehensive nonlinear modeling of the cQED sensor operation. We demonstrate that the optically-polarized NV ensemble simultaneously provides magnetic sensitivity and acts as a heat sink for the deleterious thermal microwave noise background, even when actively probed by a microwave field. Optimizing the NV-cQED system, we demonstrate a broadband sensitivity of 576 ± 6 fT/$$\sqrt{{{{\rm{Hz}}}}}$$ $\sqrt{\mathrm{Hz}}$around 15 kHz in ambient conditions. We then discuss the implications of this approach for the design of future magnetometers, including near-projection-limited devices approaching 3 fT/$$\sqrt{{{{\rm{Hz}}}}}$$ $\sqrt{\mathrm{Hz}}$sensitivity enabled by spin refrigeration.