Every robot joint is a feedback loop. The encoder is the sensor that closes it — and in a collaborative or humanoid robot, the encoder choice determines accuracy, safety, power consumption and how the whole arm feels. Here is what changes when you move from industrial servos to robot joints, and why multi-turn absolute feedback has become the default.
The joint is not a servomotor
An integrated robot joint packages a torque motor, harmonic reducer, encoder, driver and brake into one compact module. The reducer multiplies torque and rotates many turns relative to the output — which means the encoder must track far more than one revolution to know the absolute output position.
That is the core argument for multi-turn absolute encoders: they report true position across the full gear ratio at power-up, with no homing sequence and no lost-position risk after a crash or power cycle.
Why single-turn is not enough
- Gear ratio: with reducers from 50:1 to 160:1, the motor side rotates dozens of turns per output revolution. Single-turn feedback cannot encode that.
- Power-off safety: collaborative robots must recover position safely after an emergency stop. A multi-turn absolute encoder restores position instantly.
- Backdrivability: humanoids and exoskeletons are backdriven by gravity and contact. True absolute position prevents drift that accumulates on every cycle.
Inductive sensing fits the joint environment
Inside a joint, the encoder sits centimeters from motor windings, brake coils and current-carrying cables. That environment is hostile to optical encoders (dust, oil and condensation can block the light path) and demanding for magnetic encoders (strong magnetic fields from the motor can corrupt readings).
Inductive encoders are immune to magnetic interference, have no optics to contaminate, and tolerate the −40 to +85 °C range that joints see under load. This is exactly the feedback technology inside our RM10 rotary encoder — and why the RM series pairs naturally with joint modules such as the Ti5 Robot line we distribute.
Resolution, latency and protocol
For high-bandwidth current loops, position updates must arrive with microsecond-level latency. Practical joint designs use:
- 17-bit resolution (131,072 steps/turn) as a strong baseline;
- BISS-C for fast, CRC-protected bidirectional communication on high-dynamic joints;
- RS485 (Tamagawa protocol) where the drive ecosystem already speaks it.
Our RM series carries RS485, BISS-C and SSI on one connector definition, so a single qualified encoder can serve multiple joint platforms — and a protocol change does not force a cable redesign. See BISS-C vs SSI vs RS485 for the full comparison.
Size, weight and power
In a humanoid, every gram counts. Inductive sensing needs no glass disk, no LED and no read-head optics, which keeps the encoder thin and light. Battery-backed or mechanically geared multi-turn options add a little mass; electronic multi-turn (counting via power management) keeps weight down at the cost of a small standby current — choose per application.
What to ask your encoder supplier
- Does the multi-turn option survive a full power loss without battery?
- What is the ESD immunity and vibration rating for joint-mounted operation?
- Can the form factor be adapted to your hollow shaft, flange and connector layout?
- What is the lead time for samples and volume?
Designing a joint or upgrading an existing arm? Send us your torque, size and protocol targets — our engineers reply within 24 hours with a recommended RM10 configuration.