Multi-output AC–DC power supply, 84 W

Mains input supply for a lighting and emergency power product. Universal input, active PFC, two isolated GaN flyback outputs.

Released Gessler GmbH 2025–2026 12 min read Confidential

Role

Hardware engineer, power stage

Ownership

Owned PFC and flyback stages · contributed to PLC subsystem

Scope

Schematic · magnetics spec · layout review · bench bring-up

Constraints

90–264 VAC universal input · 93 % at 84 W · 8.0 mm creepage

Standards

IEC 61000-3-2 Class A · DIN EN 61347-2-7 · CE

Tools

Altium · LTspice · PI Expert · electronic load · thermal camera

Representative model. Not the client's design.

What it is

The supply takes 90–264 VAC universal input and runs an active continuous conduction mode boost PFC stage up to a regulated 380–400 VDC bus. Two isolated GaN flyback converters sit downstream of that bus. The main output is 24 V at 3.5 A — 84 W — at 93 % efficiency.

A separate isolated power line communication subsystem is fed from a second tap on the same bus. The controllers are Power Integrations InnoSwitch5-Pro GaN parts; the flyback transformer is a custom Würth Elektronik component.

I revised and extended the schematic across six hierarchical sheets, completed the multilayer layout to fabrication release, configured the secondary-side control and its I²C startup sequence, and wrote the hardware design documentation.

Architecture

Signal chain

The transformer is the only thing allowed to cross the barrier.

Primary side — energised

Secondary side — isolated

8.0 mm

Mains input

90–264 VAC · 47–63 Hz

EMI filter and bridge rectifier

CM choke · X/Y caps · full bridge

CCM boost PFC

L1 boost inductor · IEC 61000-3-2 Class A

Regulated DC bus

380–400 VDC · C12 470 µF / 450 V

GaN flyback converter

Main rail · 93 % at full load

24 V output

3.5 A · 84 W

Second GaN flyback

Isolated PLC subsystem supply

bus tapped again

The transformer is the only thing allowed to cross the barrier.

Decisions

Kind

Design decision
Problem found
Verification

Problem found U2 · schematic

An I²C address collision

Found: Two identical flyback controllers on one bus. The part has a fixed I²C slave address, so the host could not address them independently.
Did: Resolved it with an analogue multiplexer on the bus, selected by the host before each transaction. Changing controller would have meant requalifying the part and the magnetics.
Result: Kept the qualified design intact, at a cost of two passives and one small IC.

Problem found U2 · design review

Threshold overlap caught in review

Found: During design review, the line overvoltage threshold overlapped the regulated PFC bus voltage under part of the input range. In the field that would have caused nuisance shutdowns, and been extremely awkward to diagnose there.
Did: Retuned the sensing divider before the board was fabricated.
Result: No respin.

Verification

Checked

Primary inductance
Turns ratio
Flux density margin
Core and copper loss

Tool

PI Expert

How I knew the magnetics were right

The custom flyback transformer was specified with Würth application engineers. I then rebuilt the design independently in PI Expert and compared: primary inductance, turns ratio, flux density margin at maximum duty, and the core and copper loss split. The numbers agreed, and where they did not I asked why before releasing.

I would rather check a vendor's design than accept it.

What is not shown here, and why

The schematic, the layout, the magnetics specification and the measured data are Gessler GmbH intellectual property. What I can publish is the architecture, the reasoning, and the decisions — and I have published all three above. The 3D model on this page is built from the architecture description only.

What I cannot do is hand over the designs themselves, and I would not want to work somewhere that wanted me to.

Other boards