Power electronics

PV DC/DC optimiser — 350 kHz, 3×120 Wp

A panel-level boost optimiser for residential PV: per-panel MPPT, three modules in series feeding a string inverter, designed to a BOM target of €3 per 100 Wp. The cost target was the design constraint — every architectural choice, from switching topology to copper weight, was made against it.

Topology

Boost (per-panel MPPT)

Switching frequency

350 kHz

Input window

0 – 40 V, 0 – 20 A

Per-module power

3 × 120 Wp

Switch

Si MOSFET (cost-optimised, no SiC)

Controller

8-bit MCU, P&O MPPT in firmware

PCB

4-layer FR4, 70 µm outer / 35 µm inner

Assembly

SMD only, single-side reflow

BOM target

€3 / 100 Wp

At 350 kHz the inductor stays small enough for a low-profile ferrite drum and the Si switch stays inside its safe operating area without going to GaN — both line items the cost ceiling wouldn't tolerate. The 70 µm outer copper carries the 20 A input rail without a heavy heatsink; 35 µm inner planes give two clean return paths for the high-di/dt loop. P&O MPPT runs on the 8-bit MCU at a few hundred Hz — fine for irradiance dynamics, and cheap.

Most of the engineering value sat in the second-order details: gate-loop inductance held under a few nH, Kelvin source on the MOSFET, snubber tuned to the actual measured ringing rather than a calculated one, and a layout that let the production line place the BOM in one pass.

V2G on-board charger — 11 kW, 3-phase, ZVS / ZCS

Bidirectional on-board charger architecture for vehicle-to-grid: 11 kW three-phase AC interface, soft-switched (zero-voltage and zero-current) to keep efficiency high enough for daily V2G cycling without thermal de-rating. The charger is the link that decides whether the EV can play in the day-ahead and balancing markets at all — every percent of round-trip efficiency lost shows up directly in revenue per cycle.

Power

11 kW (3-phase, bidirectional)

Topology

3-phase active rectifier + isolated DAB

Soft switching

ZVS on primary, ZCS on secondary

Switches

SiC MOSFETs (1200 V class)

Galvanic isolation

Medium-frequency transformer

Direction

G2V & V2G, seamless transition

Use case

Day-ahead arbitrage, FCR, aFRR (light-vehicle aggregated)

The interesting trade is between switching loss (favours soft switching, high f_sw, smaller magnetics) and conduction loss (favours fewer transistors, lower f_sw, bigger copper). For an OBC that needs to be drivable on the road and economic in V2G, soft switching is the only way to land both: 96 %+ peak efficiency without a liquid-cooled brick.

This sits next to my MSc thesis on day-ahead price formation in 2030 with V2G — the same hardware that makes V2G economically interesting only works because you keep the charger thermally relaxed across millions of cycles. Hardware and market view, same problem.

EV drivetrain — University Racing Eindhoven

Part-time during the BSc: drivetrain power electronics for the URE student-team electric formula car. Hands-on with motor controllers, traction inverters, battery interface, and the kind of constraints (mass, packaging, vibration, thermal) that a CAD render hides from you.