An encoder on a servo motor that failed twice in a year. First time, the customer replaced the encoder. Second time, the encoder board was replaced again. The encoder was bolted to the motor shaft with a rigid coupling to a short extension shaft. The customer thought the encoder was defective. It wasn’t — the coupling was the defect.

Encoders have a specified maximum shaft misalignment. Most incremental encoders allow 0.1-0.2 mm of radial misalignment and a few tenths of a degree of angular error. That sounds generous until you bolt a rigid coupling between the encoder and a shaft that runs 20 mm from a motor bearing. If the encoder mounting face isn’t perfectly concentric with the motor shaft (and it rarely is), the rigid coupling forces the encoder’s internal bearing to take up the misalignment. The encoder bearing is small, lightly loaded, and not designed for radial force. It wears out. The encoder reads erratically. The board fails.

The mounting was the problem

The encoder sat on a stamped steel bracket, bolted to the motor flange. The bracket wasn’t machined, so its bore was out of concentricity by about 0.3 mm. The rigid coupling transferred that error straight into the encoder shaft. Every rotation, the encoder bearing was loaded sideways. It sounded fine for a few months. Then the signals started glitching.

The fix wasn’t a new encoder. It was a bellows coupling — a stainless bellows that takes up 0.5 mm of radial misalignment, 1 degree of angular error, and still has near-zero backlash. The bellows flexes; the encoder bearing doesn’t. The same encoder has run for three years since. The bellows coupling costs $25 more than the rigid one. The encoder it saved costs $300.

What to look for in a coupling for encoders

  • Bellows or spiral-cut type. Both take misalignment without loading the bearing.
  • Clamping hubs, not set screws. A set screw crushes the shaft and causes tiny eccentricity — enough to kill an encoder bearing over time.
  • Zero backlash. For position feedback, any torsional play shows up as servo oscillation at low speed.
  • Low windup torque. The coupling must not fight the encoder shaft.

If you can’t change the coupling, at least align the encoder bore to the motor shaft with a dial indicator before tightening. Get the runout under 0.05 mm. That alone extends encoder life on rigid couplings by years.

The signal check

Before blaming the encoder board, look at the A/B signals on a scope. If the signals show jitter that tracks the shaft rotation (once per rev), it’s mechanical — misalignment, coupling, or a worn bearing. If the jitter is random and present at standstill, it’s electrical — noise on the cable, a bad ground, or a marginal supply. The customer’s scope showed once-per-rev jitter. That pointed at the coupling. The first two replacements missed the diagnosis because nobody scoped the signals.

Encoders die from shaft misalignment, not from electrical failure. A rigid coupling on a stamped bracket guarantees it. Use a bellows coupling with clamping hubs, align to under 0.05 mm, and scope the signals before you buy another encoder. The encoder wasn’t defective twice — the coupling was.