ENGR 320 · Jan – Apr 2026 · Research & Presentation Lead
Three-phase BLDC inrunner
A brushless motor built from printed parts and hand-wound copper, on a $100 budget. The first build barely turned. This is what was wrong with it, how the team worked it out, and what changed before it ran.
2.02 Ω, calculated (3 × 0.672 Ω in series)CalculatedPhase resistance, RₚₕWorkingPer coil: 200 turns × π × 12.7 mm × 84.2 Ω/km ≈ 0.672 Ω. Per phase: 3 × 0.672 Ω = 2.016 Ω ≈ 2.02 Ω.InputsThree coils per phase in series (nine teeth, three phases). Every turn taken as a 12.7 mm-diameter circle; 24 AWG copper at 84.2 Ω/km (20°C). From the team’s presentation.LimitsAn estimate from wire length: layer build-up, end turns, leads and temperature are ignored, so the real figure is likely higher. No phase resistance was measured.
PETG HF (Tg 66°C per datasheet; was PLA / PLA+, Tg ≈ 55°C)
Control MCU
Arduino Uno · PWM via Servo.h
ESC
Hobbywing Skywalker 30A V2 (rated 30 A)
BOM total
$94.92 CAD (budget: $100)
Date
Jan 2026 – Apr 2026
Course
ENGR 320 · motor design project, UBC Okanagan
01
Overview
What
A three-phase inrunner with a 9-slot stator and an 8-pole rotor carrying 16 magnets in N-N-S-S pairs, wound by hand in 24 AWG enamelled copper. The stator, rotor and housing were drawn in CAD and 3D-printed; the magnets, copper wire, steel bolts and ESC were bought.
Why
It was the ENGR 320 motor design project, built to a $100 CAD budget.
My part
I served as Research & Presentation Lead on our five-person team, investigating the motor’s design and performance issues and presenting our engineering findings. The team revised the stator teeth and winding configuration, and the updated motor spun up.
Result
The revised motor spins up cleanly, shown in the video below. The bill of materials came to $94.92.
24 AWG enamelled copper, about 200 turns per pole, wound by hand.
Drive
Hobbywing Skywalker 30A V2 ESC (sensorless, rated for 30 A), with the PWM generated by an Arduino Uno (Servo.h). The motor’s operating current was not recorded.
Manufacture
Stator, rotor and housing 3D-printed; magnets, copper, bolts and ESC bought within the budget.
03
Initial design
The original design was printed throughout in PLA and PLA+, the stator teeth included, with 50 turns per pole, and its coils were wound in phase order around the stator, ABCABCABC.
Under power it produced almost no torque, and instead of rotating it oscillated. Those looked like one problem. They were two.
Fault
Magnetic reluctance
01
Symptom
Almost no torque under power.
Diagnosis
With printed PLA teeth (relative permeability ≈ 1, magnetically no different from air) almost the whole flux path was high-reluctance, so the flux, and with it the torque, was far lower than needed.
Correction
Replaced the PLA teeth with steel bolts, which give the flux a low-reluctance path. The same rebuild raised the winding from 50 to about 200 turns per pole, so the two effects were never separated, and no torque was measured before or after.
Risk
Thermal limit
02
Symptom
Not a failure: run time had to be limited to protect the stator.
Diagnosis
Resistive (I²R) heating in the windings meant continuous runs had to be kept short, so the PLA+ stator (Tg ≈ 55°C) would not soften and warp into the rotor. No temperature was measured.
Correction
Replaced the printed wire retainers with steel bolts and reprinted the base, rotor and magnet holders in PETG HF (Tg 66°C on the manufacturer’s datasheet), for more thermal margin than PLA+. Operating temperatures were not recorded.
Fault
Winding sequence
03
Symptom
Oscillated instead of rotating.
Diagnosis
With 9 teeth and 8 rotor poles, neighbouring teeth sit 160° apart electrically, so the original ABCABCABC winding put each phase’s three coils 120° apart and their torque contributions cancelled. That matches the oscillation the team saw.
Correction
Rewound as AaABbBCcC: each phase on three neighbouring teeth, with the middle coil wound the opposite way (lower case), so the three coils add instead of cancelling. After the rewind it spins up cleanly.
05
Engineering changes
First build compared with the revised motor
Part
First build
Revised
Basis
Stator teeth
Printed PLA · μᵣ ≈ 1 · 50 turns
Steel bolts · μᵣ ≈ 200 · ~200 turns
Nominal, from material dataRelative permeability, μᵣValuesPLA ≈ 1 (no better than air). Steel bolts ≈ 200, a round nominal figure: references put structural carbon steel on the order of 100 to well over 1,000, depending on grade and flux density.SourceTypical material figures used to reason about the magnetic circuit. These bolts and this print were not characterised, and no flux was measured.
Base, rotor, holders
PLA / PLA+ · Tg ≈ 55°C
PETG HF · Tg 66°C
Nominal, from material dataGlass transition, TgValuesPLA+ ≈ 55°C, the team’s approximate figure (eSUN publishes a heat-distortion temperature of about 54°C, not a Tg). PETG HF 66°C, from Bambu Lab’s datasheet (DSC); its 70°C figure is the Vicat softening point.SourceMaterial data, not measurements of these prints. No winding or housing temperature was recorded; continuous run time was limited instead.
Winding order
ABCABCABC
AaABbBCcC
Observed, not measuredClean spin-up on the revised windingSeenOn the bench after the rewind, and in the demo video on this page.Not takenNo speed, current or torque reading is recorded for the run.
Physical prototype
PETG HF base and rotor · Steel-bolt stator teeth · 24 AWG windings
06
Outcome and limitations
The revised motor, shown in the photograph, has steel-bolt stator teeth with about 200 turns per pole, a PETG HF rotor and base, and the regrouped winding. It spins up cleanly; the video below shows it running.Observed, not measuredClean spin-up on the revised windingSeenOn the bench after the rewind, and in the demo video on this page.Not takenNo speed, current or torque reading is recorded for the run.
The project notes do not separate how much of the improvement came from the steel teeth, the higher turn count and the rewind, and no torque or current was measured before or after.
Motor demo
Spin-up on the AaABbBCcC winding · Hobbywing Skywalker 30A V2 ESC