loading...
Locks the rotor at nine speeds from 5% to 95% of the no-load speed and measures the mean torque with the motor exactly as set above. The straight line is the ideal brushed-motor formula; the gap between them is the commutation penalty of the inductance.
running...
The science
What a BLDC motor is. A brushless DC motor has permanent magnets on the rotor and three coil groups (phases A, B, C) on the stator. There are no brushes: an inverter with six transistors switches the current from phase to phase so that the stator field always runs about 90 electrical degrees ahead of the rotor magnets. Three Hall sensors tell the controller which of six 60-degree sectors the rotor is in. This model is the common 12-slot / 14-magnet outrunner (7 pole pairs, so one mechanical turn = 7 electrical turns = 42 commutations).
Equations. The back-EMF of phase x is proportional to speed and to a flux shape function f (a trapezoid with 120-degree flat tops, or a sine):
e_x = Ke w f(theta_e - 2 pi x / 3), theta_e = 7 theta_m
L di_x/dt = V_x - V_n - R i_x - e_x, V_n = (V_a + V_b + V_c - e_a - e_b - e_c) / 3
T = Ke ( f_a i_a + f_b i_b + f_c i_c ), J dw/dt = T - T_load(w)
In six-step drive two phases conduct (one to the + rail with duty D, one to 0 V) and the third floats. On the flat tops of the EMF this gives T = Kt I with Kt = 2 Ke, so the motor behaves like a brushed DC motor:
w = (D Vdc - 2 R I) / (2 Ke), w0 = D Vdc / (2 Ke), Kv = 60 / (4 pi Ke) rpm/V, T_stall = Kt D Vdc / (2 R)
Commutation and torque ripple. When the sector changes, the outgoing phase current cannot stop at once because of the inductance L: it freewheels through a diode while the incoming current builds up, so for a time of about t_c ~ L I / V the torque has a dip or spike. The relative loss grows with L and speed: the current has to rise through the time constant L/R, and when L/R is not short compared with one sector (60 electrical degrees, 1/(42 x turns per second) seconds) the current never reaches its steady value. That is why the locked-speed torque in the sweep sags below the ideal straight line, more and more at high speed. Advancing the Hall timing by adv degrees starts the incoming current earlier and partly compensates for this lag (too much advance puts current on the wrong part of the EMF and lowers the torque per ampere). The floating phase back-EMF crosses zero exactly 30 degrees (plus the advance) after the commutation point; sensorless controllers use that crossing instead of Hall sensors.
Independent checks in the page. (1) steady-state speed against the closed form above, with the commutation penalty removed by Richardson extrapolation L to 0; (2) energy closure: electrical input = copper loss + stored magnetic energy + kinetic energy + work on the load; (3) brute-force count of sector changes per turn against 6 p = 42; (4) the zero-crossing angle of the floating-phase EMF; (5) no-load speed per volt against the nominal Kv; (6) locked-speed torque at three speeds against the straight torque-speed line; (7) the equilibrium speed, where the ideal line meets the load curve, against a time-domain run. Checks 1, 6 and 7 use a fast, low-inductance copy of the motor and Richardson extrapolation to remove the commutation penalty; the live motor above is not changed.
Honest simplifications. PWM is averaged (the high-side switch is represented by a mean voltage D Vdc, no switching ripple); the switches and diodes are ideal (no dead time, no voltage drops); each phase has the same self-inductance and there is no mutual inductance or saliency; the EMF shapes are idealised and Ke is constant (no saturation, no demagnetisation, no temperature drift); there is no cogging torque and no iron loss; the mechanical load is a simple sum of Coulomb, viscous, constant and propeller (quadratic) torques. Real motor values differ: this is a teaching model, not design data.