Task 4.1: Dynamic Multi-Domain Multi-Hierarchy Power Management

Supporting the diverse supply needs for different blocks:

Functional (V, I, P, tunability, accuracy, dynamic speed, etc.)
Efficient
– Fast

– Integrable

Achievements

Task: 4.2 2.5D/3D Technologies for Sensor Integration

Objective

Design and demonstrate a 3D heterogeneous integration technology specific for chip-scale and high-performance multi-spectral sensor system accounting thermal, electrical, fabrication and assembly constraints.

Comparison to SOTA Package Architectures

Electrical Feedthrough Within Thermal Bridge

Technical Approach – Glass Substrate

Take Aways

Demonstrated single-layer glass package with initial design rules
Need Cu interconnect for sub-0.1 dB loss in vertical transitions
Need low-temperature bonding to enable low-loss dielectric

Task: 4.3 ML-Assisted AMS Simulation of Sensor

Objective

State-of-the-art and Limitations

No robust RF simulators for computationally-efficient package-aware DSE and optimization

Package simulators

Simulation with traditional FEM-based methods (e.g., Ansys) is computationally prohibitive
Existing SOTA analytical models are limited
Oversimplified single signal-ground pair
Key parasitics are neglected
Transverse modes are assumed

Accelerated AMS design/simulation

DSE with SPICE methods is computationally prohibitive
Existing SOTA ML frameworks are limited
No sizing-based DSE
PVT variations are not simultaneously considered

Artifacts Generated from the Task

Utilized for predicting impact of packaging on the overall power efficiency of the CogniSense sensor
Started a collaboration with Tingyi Gu to accelerate the design and simulation of an integrated MIMO LiDAR system