Overview
I designed and built a distributed embedded-control architecture for high-power modular loads using Advanced Energy rack-mount, remotely managed, hot-swappable power infrastructure. The power layer used 4 kW PSU blades and supported approximately 12 kW of capacity per rack unit.
The goal was not simply to deliver power. The platform was built to observe and control the electrical system while relating that behavior to the physical environment around the installation. Electrical state, thermal response, hardware faults, and environmental behavior were treated as interacting parts of one system.
System architecture
Rack-scale power layer. Advanced Energy hot-swappable PSU infrastructure provided remotely managed high-density power delivery, with approximately 12 kW per rack unit using 4 kW power-supply blades.
Embedded control layer. A microcontroller-based distribution controller interfaced with the power system over PMBus for telemetry and control while supervising downstream high-current DC distribution to individual high-power modules.
Equipment monitoring and fault response. The control layer monitored voltage, current, load behavior, thermal behavior, and module health. It was designed to detect abnormal operating conditions, adjust electrical operating points, isolate faults, and preserve service to unaffected modules.
Environmental sensing layer. A separate BLE mesh sensing network provided distributed environmental telemetry around the installation. Those measurements could be correlated with electrical load, thermal behavior, airflow, and system operating state.
Observability and control
The architecture separated observability into two interacting domains. Equipment health was derived from electrical and power-system telemetry. Environment health was derived from distributed sensing of the physical environment.
The control objective was to correlate those domains rather than treat power delivery and environmental performance as separate systems. A change in load, efficiency, temperature, airflow, or hardware condition could therefore be evaluated in the context of both the equipment and the environment it was driving.
Commercialization
The platform was developed to a market-ready white-box state and progressed into discussions with potential buyers, integrators, and regulated end users. I also began discussions with New York State around deployment in licensed cultivation facilities, where existing rules referred to permitted numbers of “lights” but did not clearly map onto a modular architecture composed of independently placed light pods, shared distribution hardware, and centralized power infrastructure.
The project ultimately stopped for nontechnical reasons. Investor support fell away, and the remaining acquisition opportunities I encountered required relocation to the West Coast. Rather than continue financing and operating the venture independently while making that relocation gamble, I chose to step away and return my attention to other parts of my life and engineering work.
Validation environment
Controlled-environment horticulture was selected as the original application because inefficiencies and control changes become visible across several downstream variables at once, including heat load, HVAC demand, humidity, airflow, watering demand, and biological response.
That made the application useful as a system-level validation environment: changes in electrical and thermal behavior could be compared against measurable consequences in the physical environment rather than evaluated only at the power-delivery layer.