Short answer: a frameless torque motor is a motor with everything removable removed — no housing, no shaft, no bearings — supplied as a stator ring and a rotor ring that the robot builder mounts directly into the joint structure. Combined with a precision reducer, encoders and a drive board, it becomes an integrated joint module: the standardised "muscle unit" from which most modern collaborative robots and humanoids are assembled. The architecture saves weight and space in every joint, and its defining constraint is heat: continuous torque is set not by what the motor can produce for a second, but by what the joint can dissipate all day.
Why frameless, and why now
A conventional servo motor carries its own housing, end bells, shaft and bearings; bolting it to a gearbox duplicates much of that structure inside the joint. The frameless format deletes the duplication: the joint housing is the motor housing. Two further advantages follow. The rotor is a ring, so the centre of the joint stays hollow — cables, and sometimes shafts or cooling lines, pass straight through instead of snaking around the outside. And a large-diameter, short-length ("pancake") geometry produces high torque at low speed, which suits robot joints that work through reducers of moderate ratio or, in some designs, almost direct drive. Precision-drive makers such as maxon supply frameless kits alongside their housed motors, and the format has become the default starting point for new joint designs.
Anatomy of an integrated joint module
- Frameless torque motor — stator pressed into the housing, rotor on the rotating flange; sized by torque constant and thermal limit, not by frame horsepower.
- Precision reducer — most often a harmonic drive for its backlash and compactness, sometimes a planetary stage where cost or shock load rules.
- Encoders — typically one on the motor side for commutation and control, and increasingly a second on the output side so the controller can see the joint's true position after the reducer.
- Servo drive — a compact board closing the current and velocity loops locally, commanded over a fieldbus (EtherCAT is the common choice), which keeps the central controller's wiring and computation manageable.
- Structure and bearings — a cross-roller or paired bearing set carrying the joint's moment loads, and the housing that doubles as the heat path.
Sold as one tested unit, the module converts a multi-supplier integration problem into a line item — which is precisely why it has become a product category of its own inside the directory's core components section (160 companies, as of 10 November 2026).
The thermal bargain
Every integrated joint is a negotiation with heat. Copper losses rise with the square of current, the structure is small and sealed, and there is no fan. The practical consequences: continuous torque ratings are a fraction of peak ratings; a joint that performs beautifully in a thirty-second demonstration can derate badly in minute twenty of a work cycle; and two modules with identical peak figures can differ widely in useful work. Engineers evaluating modules should ask for the continuous rating at a stated ambient temperature, the thermal time constant, and test data over a realistic duty cycle — and should treat the reducer's own torque limits (which are mechanical, not thermal) as a separate ceiling.
Why humanoids multiplied the demand
An industrial arm has six or seven joints. A humanoid has powered joints throughout its legs, arms, torso and neck — an order of magnitude more modules per machine, each one weight-critical because every gram in a limb must be carried by the joints below it. The humanoid programmes now publicly pushing toward production — Tesla's Optimus effort and Figure AI among them — therefore translate directly into demand for compact joint modules and their ingredients: frameless motors, harmonic reducers, miniature encoders. It is the clearest example in the industry of a component category being scaled by a single product form, and the reason motor and reducer suppliers feature so prominently in the current supply-chain conversation.
Build or buy the joint?
Robot makers split three ways: buy complete modules, buy frameless kits and reducers and integrate them into their own joint housings, or design the actuator fully in-house. Buying modules is fastest and shifts fatigue-testing and supply risk to the vendor — right for prototypes and modest volumes. Full in-house design pays only when the joint itself is the product's edge, or volumes are large enough to amortise years of iteration. The middle path — own housing and integration, bought motor kits and reducers — is where most serious mid-volume makers land, and it is why component suppliers publish detailed integration guides for their frameless kits: the machining tolerances of your housing become part of their product's performance.
Frequently asked questions
What is a frameless torque motor?
It is an electric motor supplied without its own housing, shaft or bearings: just the stator (the stationary winding ring) and the rotor (the magnet ring). The robot builder mounts the stator directly into the joint's structure and the rotor onto the joint's rotating part, using the machine's own bearings. Removing the duplicated housing saves weight and space and leaves a large hollow centre for cables and shafts to pass through.
What is inside an integrated robot joint module?
Typically four elements packaged as one unit: a frameless torque motor, a precision reducer (often a harmonic drive), one or two encoders for position feedback, and a servo drive board, all inside a housing with the output flange and bearings. The robot maker bolts the module between two limb segments and connects power and a fieldbus cable — the joint arrives as a component, not a project.
What limits the torque of a robot joint?
Heat, mostly. A joint can produce high peak torque briefly, but continuous torque is set by how fast the motor's copper and iron losses can escape through a small, sealed structure with no fan. This is why datasheets show a large gap between peak and continuous ratings, and why realistic duty-cycle testing matters more than the peak figure.
Should a robot maker build joints in-house or buy modules?
Volume and differentiation decide. Buying modules shortens development and de-risks supply for small-to-medium volumes; in-house joints make sense when the joint is the product's core advantage or when unit volumes amortise the engineering. Many makers do both — bought modules on secondary joints, proprietary designs where performance is won or lost.