How to Read a Joint Module Datasheet

1. The Three Specs That Decide Everything

Joint module datasheets list dozens of parameters, but selection ultimately reduces to three decisions:

Decision	Spec to read	What it answers
Will it move my load?	Torque-speed curve	Continuous vs peak capability at the speed I need
Will it hit my tolerance?	Precision grade (backlash, repeatability, encoder)	How close to the commanded pose the joint really gets
Will my controller talk to it?	Communication protocol	Fieldbus support, cycle time, ecosystem fit

Everything else — size, weight, IP rating, temperature — is a constraint, not a decision. Start with these three.

2. Reading the Torque Curve

The torque-speed curve is a performance map, not a single number. The classic shape for an integrated joint has two regions:

• Constant-torque region (low speed): the module can deliver its full rated torque up to the rated speed. Robot joints live almost entirely in this region — which is why rated speeds of 39.6 RPM (PP-series harmonic) and 50 RPM (planetary) are normal.
• Constant-power region (above rated speed): torque falls as speed rises. If your application needs fast repositioning, check the curve here, not the rated number.

Two numbers deserve special attention:

• Continuous torque: sustainable at thermal equilibrium (stated ambient + duty cycle). This is your design load. The PP-80, for example, is rated 35 Nm continuous / 70 Nm peak.
• Peak torque: short-burst capability (roughly 2× continuous on the PP-series) for acceleration spikes and impacts — never a steady-state number.

Rule of thumb: size steady-state loads against continuous torque with 20–30% headroom, and keep peak torque for transients. Our torque calculation guide walks through the static and dynamic math behind this.

For the thermal side of continuous torque — why it shrinks in hot environments and what derating means in practice — see Integrated Driver Heat Dissipation.

3. Understanding Precision Grade

"Precision grade" is not one spec — it is a bundle of three, and buyers routinely confuse them:

Spec	Unit	What it measures	PP-series values
Backlash	arcmin	Lost motion when direction reverses	<1 arcmin (PP-52L…PP-80); <0.5 arcmin (PP-90…PP-170P)
Repeatability	°	Spread of returned poses to the same command	±0.003° small models; ±0.002° large models
Encoder resolution	bits	Finest position change the feedback can report	17-bit dual absolute (motor + output)

• Backlash is the mechanical precision grade — the harmonic drive's reason for existing. Sub-arc-minute backlash means no dead zone when a weld path or a pick-and-place motion reverses direction. See our backlash deep-dive for how it is measured.
• Repeatability answers "will it return to the same pose?" — the spec that matters for cycle consistency. It is not the same as absolute accuracy (which needs calibration).
• Encoder resolution is the electrical precision grade. Dual absolute encoders (motor + output) give true output position without homing — see absolute vs incremental for why this matters.

When a buyer says "precision grade," pin down which of the three they mean — the answer changes the product you quote.

4. Communication Protocol: EtherCAT vs CANopen vs CAN

The fieldbus is the third decision, and it is usually locked by your controller ecosystem before the joint is chosen. The landscape for integrated joint modules:

Protocol	Topology / wiring	Cycle & determinism	Best for
EtherCAT	Ring, standard Ethernet cable	≤1 ms distributed-clock synchronized motion across dozens of axes	Multi-axis industrial arms and lines; large controller ecosystem (TwinCAT, CODESYS, etc.)
CANopen (CiA 301/402)	Daisy-chain/bus, 2-wire	Millisecond-class, good for smaller axis counts	Simple, proven, low-cost multi-drop systems
CAN 2.0 (plain)	Bus, 2-wire	Event/command driven	Budget builds, humanoid/education platforms (planetary series)

• Choose EtherCAT when the arm is part of an industrial line: cycle-synchronous motion, distributed clocks, and the ability to drive joints + grippers + sensors on one network. The PP-series harmonic modules speak EtherCAT and CANopen.
• Choose CANopen when the system is smaller, the budget tighter, or the existing controller is CAN-based. Setup guidance is in our CANopen protocol guide and EtherCAT setup guide.
• Plain CAN 2.0 fits the planetary PP-series and humanoid/education projects where simplicity and cost lead.

Check three protocol details before committing: CiA 402 profile support (motion states), whether absolute position is transmitted on startup (no homing), and whether the vendor supplies setup files for your controller.

5. The Three Together: A Selection Walkthrough

Take a 6-axis arm, 1.0 m reach, 5 kg payload, ±0.1 mm positioning at the tool, running on an existing EtherCAT line:

1. Torque curve: wrist-to-shoulder static + dynamic analysis gives peaks of 45/42/25/11/8/3.8 Nm → maps to PP-90 (base), PP-80 (elbow), PP-70/PP-52 (wrist) with 25%+ headroom. (Full walkthrough in the torque guide.)
2. Precision grade: ±0.1 mm at 400 mm tool radius ≈ ±0.014° — the PP-series ±0.003° repeatability and <1 arcmin backlash clear it with a wide margin.
3. Protocol: EtherCAT for the existing line; all selected modules are EtherCAT-native with CiA 402 support.

Result: a joint map chosen in three decisions instead of thirty parameters.

6. The Datasheet Checklist

• ☐ Continuous torque at MY ambient and duty cycle (not just the headline number)
• ☐ Peak torque ≥ my worst acceleration spike
• ☐ Backlash, repeatability and encoder resolution vs my tolerance budget
• ☐ Protocol matches my controller; CiA 402 profile + no-homing absolute position
• ☐ Vendor supplies setup files, wiring diagram and per-serial test data