All projects Project 01 / 04

Independent R&D · Hardware

Electromagnetic Accelerator

A long-running engineering build that evolved through three physical versions — from a high-voltage test setup to a functional two-stage system informed by electromagnetic simulation, mechanical CAD, embedded timing, and measured results.

ANSYS Maxwell
Fusion 360
Arduino
C/C++
3D printing
Physical testing

Role: Designer & builder
Timeline: Jun 2022 – present
Type: Individual project
Focus: Embedded electromechanics

Stage-control source SomeHeresy/Electromagnetic-Accelerator

Fig. 01The V3 bench: Arduino stage control, the capacitor bank, and both coil stages.

01 — Overview

Three versions, one expanding test loop.

The project began as an experiment in electromagnetic acceleration and grew into a system-level build connecting circuit behavior, coil geometry, timing logic, mechanical repeatability, and measured physical performance.

Measured velocity, two stages: ~11m/s
Projectile mass: 54.64g
Sequential coils: 2stages

02 — The challenge

Control the sequence, not just the circuit.

A multi-stage accelerator only works when each coil energizes at the useful part of the projectile’s motion. The central challenge was coordinating sequential electromagnetic stages while turning a loose experimental setup into a mechanically repeatable system.

Model the magnetic system and evaluate coil-timing behavior.
Create stage-control logic that could translate simulated timing into physical activation.
Design supports and launcher geometry that kept the two-stage assembly aligned.
Measure the result and preserve failed tests as engineering evidence.

03 — Engineering process

Simulation, CAD, firmware, then the bench.

Electromagnetic simulation

I used ANSYS Maxwell 2D to study the electromagnetic field and support coil-timing optimization. The simulation became a design input for the embedded sequence rather than a standalone visualization.

Mechanical design and fabrication

I modeled support components in Fusion 360 and fabricated parts through 3D printing, improving the physical alignment and repeatability of the launcher.

Embedded control

Arduino stage-control logic coordinated sequential coil activation. The final documented build used four 400 V, 330 µF capacitors and a two-stage arrangement.

04 — Iteration record

Three versions, four years.

Versions 1 and 2 were physical prototyping and experimental iteration. Version 3 is where CAD, electromagnetic simulation, and precise embedded timing entered the build. The failed tests stay in the record — they are where most of the learning actually happened.

Version 1

Early prototype

2022 · 550 V · single stage

The first question was the simplest one: does the loop work at all? Charge a bank, fire a single coil, and measure whatever comes out the other end.

The Version 1 prototype on a green cutting mat: a high-voltage boost board, capacitors, and a single coil wound on a black tube — Fig. 02The Version 1 prototype and test setup.

Speed detector readout from a Version 1 test — Fig. 03The speed detector showing a measured result.

Version 1 under test with the speed measurement device — Fig. 04Running the setup through the speed measurement device.

V1 — Full assemblyYouTube (opens in a new tab)

V1 — Experimental documentationYouTube (opens in a new tab)

Version 2

Two-stage preparation

Architecture · sensing experiments

V2 existed to prepare the system architecture for a second stage — and to find out how to sense the projectile well enough to time it. The photoelectric sensing approach was tested here and later abandoned, which is what pushed the timing problem toward simulation in V3.

Version 2 prototype circuit setup — Fig. 05The Version 2 prototype circuit.

The ferrous projectile used in Versions 2 and 3 — Fig. 06The 54.64 g slug used in both V2 and V3.

V2 — Test fires at multiple voltagesYouTube (opens in a new tab)

V2 — Close-up footageYouTube (opens in a new tab)

V2 — Photoelectric sensor test, later abandonedYouTube (opens in a new tab)

Version 3

Functional two-stage design

Simulation-informed timing · ~11 m/s

The working build. Two stages sit close together and fire in sequence under Arduino control, with the timing derived from ANSYS Maxwell rather than from a sensor. Fusion 360 supports and 3D printing gave the assembly enough mechanical repeatability to test the same thing twice.

The Version 3 launcher: a coil wound on a spool around a clear barrel, wired through a diode to the stage-control circuit — Fig. 07The launcher: two stages coupled and timed from simulated data.

ANSYS Maxwell field model of the accelerator — Fig. 08ANSYS Maxwell field model.

ANSYS Maxwell circuit model driving the coils — Fig. 09ANSYS Maxwell circuit model.

Fusion 360 model of the coil support — Fig. 10Fusion 360 coil support.

V3 — Setup showcaseYouTube (opens in a new tab)

V3 — Test firesYouTube (opens in a new tab)

V3 — Failure example: what went wrongYouTube (opens in a new tab)

Videos play on this page. Full channel: @SomeHeresyGaming (opens in a new tab)

05 — Outcome

A working system — and a better engineering method.

The documented two-stage build reached approximately 11 m/s with a 54.64 g projectile. More importantly, the project taught me to connect simulation assumptions to embedded timing, physical fabrication, repeatable testing, and failure analysis instead of treating those disciplines as separate tasks.

The next step was explaining the number, not just measuring it — which became the Coilgun Timing Simulator.

Next · Project 02 Coilgun Timing Simulator

All projects Read my résumé Get in touch