UVW Commutation Signal Alignment: Calibrating Encoder Output Phase for Brushless Motor Drives

UVW and ABZ: one code disc, two different jobs

An incremental optical encoder with commutation outputs carries two independent signal families on the same code disc. The ABZ family is the metrology: A and B are quadrature square waves repeating every mechanical line pitch, and their count gives position with the encoder's full resolution; Z fires once per revolution to define the datum. The UVW family is the motor's coarse clock: three channels spaced 120 electrical degrees apart, each completing one full period per electrical cycle of the motor—that is, every 360°/pole-pairs of mechanical rotation. On a ten-pole-pair motor, the U channel repeats every 36 mechanical degrees, and the drive only needs to know the rotor angle within one sector (±30 electrical degrees) to apply current to the correct phase pair at power-up.

The two families are physically linked: UVW tracks are exposed or masked on the same disc and read by the same optics, so their phase relation to ABZ is fixed at manufacture. When the encoder is correctly aligned, every UVW edge transition lands at a defined ABZ state—count N after the index, A high and B low, and so on, six sectors per electrical cycle. Alignment quality is exactly this: how closely the measured UVW edges match the expected ABZ positions, expressed in electrical degrees. A drive that trusts misaligned UVW channels applies its phase current at the wrong stator angle, and no amount of downstream servo gain fixes the error at the source. The incremental tracks and their own failure modes are covered signal by signal in the ABZ troubleshooting guide, which pairs naturally with this alignment check.

How commutation angle error becomes torque ripple

A brushless drive produces smooth torque only when the stator current vector leads the rotor's electrical angle correctly. Commutation angle error shifts that vector: in sinusoidal commutation the average torque falls with the cosine of the error while the current between phases no longer converts cleanly into rotation, and in block or six-step commutation the error appears directly as ripple at the electrical frequency and its harmonics. As a working rule, the dominant ripple grows roughly with the sine of the electrical error—five degrees of misalignment is consistent with around nine percent ripple, ten degrees with fifteen to twenty percent, and a worst-case hall-style sector error of thirty electrical degrees can waste half the torque capacity at some rotor positions. The error also shows up as phase-current imbalance, extra heating, and acoustic noise with a signature tied to speed: ripple frequency scales with rotation, which is how it is distinguished from bearing or mechanical defects.

For precision stages the consequences compound. Low-speed jitter from torque ripple limits velocity stability exactly where scanning and metrology applications operate, and it forces the servo loops to be detuned to avoid amplifying the disturbance. Position loops then hunt around the command, encoder interpolation sees irregular motion, and throughput drops because smooth slow moves become impossible. Keeping the total commutation chain—encoder mechanical alignment plus drive electrical offset—inside about ±5 electrical degrees is a reasonable target for demanding motion systems, and the budget should be written into the integration specification rather than discovered at commissioning. How ripple and residual periodic error propagate into servo behavior is treated in the encoder feedback tuning guide, which covers the loop side of the same budget.

The oscilloscope alignment procedure

Preparation first. Power the encoder from a current-limited supply at its rated voltage and terminate the differential outputs per the datasheet; probe U, V, W, A and B with differential probes or a proper differential front end, keeping grounds clean. Use a 100 MHz-class scope, trigger on the U rising edge, and rotate the shaft slowly and repeatably—by hand for static checks, or with the drive in torque mode at a few RPM with the phase leads disconnected from power for dynamic checks.

Method one is the static six-sector check. Rotate the shaft until the U channel rises, freeze the display, and record the A and B logic states and the AB counter reading. Advance through one full electrical cycle and repeat at every UVW edge: six transitions, each with an expected AB state from the drive's sector table. Convert any mismatch into angle directly—each AB count is 360/(counts per revolution) mechanical degrees, multiplied by the pole-pair count to give electrical degrees. The same pass yields the quality checks: UVW duty cycles near fifty percent, edge spacing near 120 electrical degrees, and clean edges with no chatter at the thresholds.

Method two is the back-EMF comparison, the definitive check of encoder-to-motor electrical angle. Spin the motor with an unpowered drive or a second rotator and scope one phase back-EMF against the U channel: for the standard 120-degree sensor placement, the U rising edge should coincide with the phase back-EMF zero crossing. Repeat for the V and W phases against their own EMFs. Any fixed offset here is the commutation angle error; small residuals can be entered into the drive's commutation offset register, but a large or speed-dependent error points at encoder mounting or stator placement and must be fixed mechanically. Quadrature quality of the A and B channels themselves—phase error and amplitude balance—should be verified in the same session using the Lissajous method described in the quadrature signal tuning guide.

• Trigger on U rising and record the AB state and count at all six UVW transitions per electrical cycle
• Convert count error to electrical degrees with the pole-pair factor before judging pass or fail
• Check UVW duty cycle, 120-degree spacing, and edge monotonicity, not just the presence of edges
• Verify back-EMF zero crossings against UVW edges before trusting any drive-side offset correction

Factory calibration and field verification

Alignment is established in two stages that must not be confused. At the encoder manufacturer, the UVW mask and code disc are positioned and verified against the ABZ tracks—often with optical datum equipment—and the resulting commutation relation is checked over temperature and recorded with the unit. At the motor or stage integrator, the drive-side electrical zero is finalized: the classic procedure locks the rotor with a known current vector, reads back the encoder position, and writes the commutation offset register. Both numbers together define the delivered alignment, and both belong in the shipment documentation—an unrecorded offset is a baseline lost, and the incoming acceptance test should specifically check that commutation data accompanies the encoder.

Field verification repeats the factory evidence rather than re-deriving it. After any motor swap, encoder replacement, or coupling rework, run the static six-sector check and compare the electrical-degree error against the as-shipped record; then spin at low speed with a clamp-on current probe and compare phase-current ripple against the commissioning baseline. The drive offset register can absorb a few electrical degrees of drift, but it should be treated as a trim, not a repair: a result that differs materially from the factory record means something mechanical moved, and reseating the encoder or correcting the mounting beats compensating for it. Systems built this way—documented factory alignment, sector-table verification on receipt, and ripple comparison after every intervention—keep commutation quality a specification instead of a surprise, which is the operating assumption in precision motion applications where torque ripple is a measured line item.