Motor Dynamic Balancing: What the Rotor Magnet Contributes and What to Specify

ISO 1940-1 states the allowance as a balance quality grade G, and permissible residual specific unbalance follows from grade and speed: e_per = G / ω, with ω the angular velocity in rad/s. In the form most people use at the bench, e_per in micrometres is about G × 9550 / n, where n is rpm. Take a servo rotor at G2.5 running 30,000 rpm: 2.5 × 9550 / 30,000 gives roughly 0.8 µm of permissible mass eccentricity.

Multiply by rotor mass for the permissible residual unbalance in g·mm. Two things follow from the shape of that expression. The allowance falls in inverse proportion to speed, so the same grade at 30,000 rpm permits a tenth of what it permits at 3,000 rpm. And it is a whole-rotor allowance: shaft, lamination stack, magnet, adhesive and retention sleeve all draw on the same 0.8 µm.

Typical practice puts general electric motors at G6.3, medium and larger machines and many servo rotors at G2.5, and high-speed precision spindles at G1.0 or G0.4. Those are conventions the motor designer sets, not something a magnet supplier gets to impose.

Six contributors turn up on a PM rotor, and the magnet owns three or four of them depending on how the rotor is built. Worth separating them before anyone argues about who caused the reading on the machine, because the correction for each is different.

A one-piece sintered ring starts with less built-in asymmetry than a segmented ring, for the plain reason that it has no bond lines. Segmented rings are arc segments bonded to a hub, and every bond line is a mechanical and thermal interface. Segment-to-segment mass scatter and adhesive thickness variation both add asymmetry the one-piece ring does not have, and both land around the circumference in whatever pattern the assembly fixture happened to produce.

None of that is an argument against segmented construction. Arc segments are the cheapest manufactured form for a given remanence, and on long rotors or large diameters there is often no alternative. The balance consequence just needs pricing in at design stage instead of turning up at first article.

On a short, high-speed rotor with a tight grade, my own preference is to pay the piece-price premium for a one-piece ring and take the time back on correction and scrap. Construction also sets the airgap waveform, which we go through in magnetic circuit design for PM motors, so balance is rarely the only input to that decision.

The magnet's share of the balance budget is carried by wall thickness variation, coaxiality and cylindricity, and all three have numbers on a standard ring specification. Radial-oriented multipole rings run a standard envelope of OD 20 to 75 mm, ID 15 to 68 mm, wall 2 to 7 mm and height 5 to 50 mm, held to ±0.05 mm as standard, with coaxiality 0.1 and cylindricity 0.04.

Grades run N30 to N48 in this construction, a narrower band than the full sintered NdFeB grades catalogue, with temperature classes to 35EH and straight or skewed multipole magnetization. Tighter than standard is a conversation about process capability and price, and it should happen before the drawing is released.

One number gets misapplied often enough to be worth calling out. The ±0.02 mm figure belongs to one-piece sintered Halbach rings, where critical features are produced net-shape to that tolerance. It is not a radial ring tolerance. Radial rings are ±0.05 mm, and carrying the Halbach number onto a radial ring drawing produces a quotation package nobody can honour.

Below Ø20 mm OD the part moves to a multipole-oriented ring process, which reaches 8 poles from Ø6.3 mm and goes down to a Ø4.4 mm ring at 4 poles. At those diameters the mounting and adhesive contributions dominate the magnet's own asymmetry and the discussion moves to fixturing, which we cover in micro multipole magnets for coreless motors.

If the ring is bonded to a hub or shaft before it ships, concentricity and bond line stop being the motor builder's problem and move into the magnet assemblies scope. That is the version we prefer when the balance grade is tight, because the party controlling the fixture is then the party holding the tolerance.

Coaxial permanent magnetic drives built in our production network are held to GB Grade 2.5 balance with 0.02 mm accuracy, a drive-assembly figure that should not be written against a loose magnet. Quality documentation runs to ISO 9001:2015 and IATF 16949:2016 through the certified plant in our production network, with PPAP Level 3 packages available.

Correction in general use means removing material by drilling or grinding on a balance ring or the lamination stack, adding balance weights, or both, with two-plane balancing once the rotor is long enough that one plane cannot cancel the couple. All of it is ordinary work and none of it needs to touch the magnet.

Taking material off the magnet is a different matter. The coating is the corrosion barrier and a ground or drilled face breaks it. Local heating from a wheel can push that spot past its working point. Neither risk is theoretical on a part that then sits in a sealed motor for ten years. Where a rotor genuinely needs correction material out at magnet radius, put it on the drawing as a designed feature in the hub or a balance ring, decided in design review with the magnet supplier in the room.

The practical reason to be explicit: a balance operator with the rotor in the fixture and a drill in hand takes material wherever it is easiest to reach. If the drawing does not say where correction is allowed, sooner or later it gets taken somewhere nobody intended.

Send the operating point and the rotor build and we can size the magnet's share of the balance budget while the drawing is still open, rather than after the first balance report. Engineering response is within 1 business day, pricing within 2 business days. What we need: