Projects
Each entry documents the requirement, the decision that shaped the design, the hardware that came out of it, and the measured result. Schematics, layouts, CAD, and source are linked where I am free to share them.
Completed work
48 V → 12 V GaN Point-of-Load Converter
A 300 W synchronous buck built around 100 V GaN half-bridges, designed for high density and a clean switching node under load transients.
3U CubeSat Electrical Power System
Solar array MPPT, battery management, and switched load distribution on a single PC/104 board sized for a 90-minute low-Earth orbit.
32-Channel Test Stand Telemetry DAQ
A ruggedized data acquisition board for propulsion and structural test benches — 32 analog channels, hardware timestamping, and a Python host stack.
Planned builds
None of these are built yet. They are the designs I intend to take from requirement to measured hardware next, listed here so the direction is visible rather than implied. Each entry states the target specification and what building it would actually prove. As one gets built it moves up into the projects above, with real data.
1–10 kW Isolated DC-DC Converter
3 kW, 400 V → 48 V, phase-shifted full bridge
Everything gets harder above a kilowatt. Conduction loss stops being a rounding error, magnetics stop fitting on the board, and the thermal path becomes the design driver rather than an afterthought.
What it would demonstrate- Loss budgeting where a 1 % efficiency error is 30 W of heat
- Planar or litz magnetics design at real power density
- Phase-shift modulation and ZVS across the load range
- Forced-air or cold-plate thermal design with measured correlation
LLC Resonant Converter
1 kW, 390 V post-PFC input, half-bridge LLC
The topology that gets used when efficiency is the specification. Zero-voltage switching on the primary and zero-current turn-off on the secondary let it run fast and cool — provided the resonant tank is designed properly.
What it would demonstrate- Resonant tank design and the gain-versus-frequency curve
- Maintaining ZVS from full load down to burst mode
- Start-up, short-circuit, and over-current behaviour, which is where LLC designs usually fail
- Measured efficiency correlated against a first-principles loss model
Three-Phase Inverter
5 kW, 400 V DC bus, SiC half-bridges, SVPWM
DC to AC is where control and power electronics stop being separable. The hardware and the modulation scheme have to be designed against each other, and the errors show up as harmonic distortion rather than smoke.
What it would demonstrate- Space-vector modulation and dead-time compensation
- Current control loop design with real sensor bandwidth limits
- Output filter design against a THD target
- Conducted EMI mitigation on a hard-switched SiC bridge
Active-Clamp Flyback
60 W, 36–72 V input, multiple isolated outputs
The workhorse of auxiliary and housekeeping supplies. Small, cheap, and deceptively difficult — the transformer is the circuit, and leakage inductance decides how much of your efficiency you keep.
What it would demonstrate- Coupled-inductor design including gap, leakage, and winding loss
- Active clamp recovering leakage energy instead of burning it in an RCD
- Cross-regulation between multiple isolated windings
- Conducted EMI from a topology that pulses current on both ports
High-Voltage Bias Supply
1–5 kV output, low ripple, current limited
High voltage changes what the rules are about. Creepage and clearance start driving the layout, insulation coordination becomes a real analysis, and the failure modes are the kind that matter.
What it would demonstrate- Creepage and clearance to IEC 60664 with pollution-degree assumptions stated
- Multiplier or transformer-based topology selection and ripple budgeting
- Partial discharge awareness and conformal coating strategy
- Interlocks, bleed resistors, and a safe-by-design test procedure