Hackaday · a turbine disc balanced to about 1 mg
Balance the Rotor
This rotor carries a hidden heavy spot. Spin it and you get what a balancing rig gives you: how hard it shakes, and where in the turn the shaking peaks, measured from a once-per-turn marker. Two spins and a little vector arithmetic tell you where to put a correction weight. Get it under 1 mg.
weight
20 mg
Holes every 15°, measured from the marker the Hall sensor sees. Weights you add stay until you remove them.
not measured yet
accelerometer (line) and the once-per-turn marker (ticks), three turns
each run as an arrow: length = how hard it shakes, angle = where the peak falls after the marker
run log
1. Spin as found
The heavy spot makes the rotor pull outward once a turn, so the bearing shakes at the running speed. The rig reports that shaking as an arrow: a size and an angle. The angle is not where the heavy spot is; the mounting adds its own unknown delay.
2. Spin with a trial weight
Add a known weight anywhere and spin again. The arrow moves. The change between the two arrows is what your trial weight did, delay and all, so it tells you how this machine turns milligrams into shaking.
3. Cancel the first arrow
Scale and turn the trial weight until its effect is exactly opposite the first arrow. That is the correction. Take the trial off, put the correction on, and spin to check. If the angle falls between holes, split it across the two neighbours.
This is the textbook single-plane influence-coefficient method, simulated: the machine's response is linear, with a random gain and delay for each rotor, plus a little measurement noise. The real rig in the story used an ADXL335 accelerometer and a Hall sensor on an ESP32, reported about 1 mg of sensitivity at 4,000 rpm, and the balancing itself was done slowly, by hand. Source: Maya Posch, Balancing A Turbine Rotor To 1 Milligram With A DIY Dynamic Balancer, Hackaday, 25 January 2026, on a video by AlfMart CNC Garage. Made by Claude, 3 October 2026.