Frameless vs. Framed Motors in Humanoid Robot Joints: A Packaging and Thermal Analysis

When developing a custom humanoid robot, one of the first major architectural decisions for the actuation system is whether to use commercially available framed servomotors or to integrate frameless torque motor kits directly into a custom joint housing.

While framed motors are easier to test on a bench, they bring severe penalties in weight, volume, and thermal performance when stacked into a dense humanoid structure—such as a bipedal leg or a multi-DOF arm. This is why almost all funded humanoid programs eventually transition to frameless hollow-shaft humanoid actuators.

In this deep-dive, we analyze the specific mechanical, thermal, and electrical trade-offs that drive this transition, and what engineering teams must prepare for when designing with frameless motors.

The Parasitic Mass Penalty of Framed Motors

A commercial framed motor is designed as a standalone product. It comes with its own external housing, its own front and rear bearings, its own output shaft, and typically a generic mounting flange. When you mount a framed motor to a harmonic or cycloidal reducer, you are forced to use a mechanical coupling and a structural adapter plate.

In a humanoid robot joint, every millimeter of axial length and every gram of unsprung mass matters. The redundant components add what we call "parasitic mass"—weight that does not contribute to torque generation or structural integrity of the limb.

By moving to a frameless motor design, the robotics team can design a single, unified CNC housing that supports the stator directly, uses the rotor to drive the wave generator or sun gear, and shares a common bearing architecture. This eliminates the coupling, the motor shell, and the adapter flanges, often reducing the overall joint axial length by 30-40% and mass by up to 35%. When multiplied across 30 to 40 joints in a humanoid, this saves tens of kilograms of mass, allowing the robot to carry a larger battery payload and run for hours rather than minutes.

Thermal Dissipation and Continuous Torque Limits

Humanoid robots require high continuous torque to hold poses against gravity (e.g., a squatted stance). Heat generation in the stator coils (I²R losses) is the primary limiting factor for continuous torque.

The Problem with Framed Motor Cooling

Framed motors rely on their generic cylindrical shell to dissipate heat. In a robot arm or leg, this shell is often buried under plastic cosmetic covers or tucked inside a structural limb, leading to poor convective cooling. As the internal temperature rises, the motor undergoes rapid thermal derating. If pushed too hard, the winding enamel melts (typically around 155°C or 180°C depending on the insulation class), causing a catastrophic short circuit.

The Frameless Thermal Advantage

By utilizing a frameless stator pressed or potted directly into a custom actuator housing, the entire robot limb becomes a heat sink.

• Thermal Potting: Using high-thermal-conductivity epoxy potting compounds (e.g., >1.5 W/m·K) to secure the stator eliminates insulating air gaps. The epoxy transfers heat directly from the copper windings to the aluminum outer shell.
• Interference Fit (Shrink Fit): Heating the aluminum housing and dropping in the room-temperature stator creates a flawless mechanical and thermal bond once cooled.
• Fins and Liquid Cooling: The custom housing can feature external cooling fins exposed to ambient airflow, or even integrated liquid cooling channels for ultra-high-performance bipedal legs.

This direct, engineered thermal path significantly increases the continuous torque limit of the exact same electromagnetic active parts by up to 40% compared to a closed-frame equivalent.

Cable Routing and the Hollow-Shaft Mandate

Humanoid robots suffer from notoriously complex cable management. Routing high-current power lines, CAN/EtherCAT buses, and absolute encoder signals past continuously spinning joints is a massive reliability risk.

Framed motors typically have a solid shaft, forcing cables to route around the outside of the joint. This requires "service loops" (extra cable slack) that increase the snag hazard, expand the robot's outer envelope, and lead to rapid fatigue failure from repetitive bending.

Frameless torque motors are inherently designed as large-diameter rings. This hollow-shaft architecture allows the robot's main wiring harness to pass directly through the center of the joint axis of rotation.

• Twist vs. Bend: Cables passing through the center of rotation undergo torsional twisting rather than bending. Torsional strain is vastly less damaging to copper strands, extending harness life from thousands of cycles to millions.
• Slip Rings: A large hollow bore allows the integration of capsule slip rings for infinite rotation joints (e.g., a continuous panning neck or wrist).

Advanced Electromagnetic Design: Kt vs. Kv

When selecting frameless motor kits, the robot design team must carefully balance the Torque Constant (Kt) and the Velocity Constant (Kv). For humanoid robots, high torque density at low speeds is highly desirable. Engineers must look at the motor's slot-fill factor. Hand-wound stators can achieve higher fill factors but at a much higher labor cost, while automated needle winding is highly scalable for pilot production but slightly reduces the copper density. The pole/slot combination is also critical. A high pole count (e.g., 21 slots / 14 poles) increases torque density but drastically increases the electrical frequency the motor controller must supply at high speeds, which can push standard inverters to their switching frequency limits.

Engineering Challenges of Frameless Integration

While the benefits are overwhelming, transitioning to frameless motors shifts the burden of precision manufacturing from the motor vendor to the robot design team (or their ODM partner).

1. Air Gap Concentricity: The air gap between the stator and rotor in a high-torque frameless motor is typically between 0.3mm and 0.5mm. Your custom housing and bearing seats must maintain strict concentricity (often <0.02mm) to prevent the rotor from rubbing against the stator under heavy radial loads.
2. Cogging Torque and Slot/Pole Ratios: Custom integrations must carefully select the pole/slot combination. High pole-count motors provide higher torque density but can suffer from severe cogging torque, which ruins backdrivability unless compensated by advanced software or skewed stators.
3. Encoder Calibration: In a framed motor, the encoder is factory-calibrated to the rotor's magnetic poles. In a frameless build, the assembly team must install the magnetic ring or optical disc and perform a rigorous phase-alignment calibration to ensure efficient commutation. Dual-encoder setups (one on the rotor, one on the output shaft) complicate this further, demanding rigorous firmware initialization routines.

Quality Assurance and Testing in Pilot Builds

When an ODM factory assembles a custom frameless joint, the assembly must pass rigorous electrical testing before the mechanical build is even completed.

• Hi-Pot (High Potential) Testing: Ensures that pressing the stator into the aluminum housing did not scratch the wire enamel and short the phase windings to the ground (the housing).
• Surge Testing: Verifies there are no turn-to-turn shorts within the coils.
• Back-EMF Measurement: Spinning the rotor with an external fixture to measure the generated voltage, ensuring the magnets haven't been demagnetized during shipping or assembly.

Frameless vs. Framed: Side-by-Side Performance Comparison

The following table summarizes quantitative trade-offs that drive the frameless vs. framed decision for custom humanoid actuator programs. Values are representative ranges for motors in the 50-100 mm stator OD class commonly used in humanoid hip, knee, and shoulder joints.

Key insight: Frameless motors deliver 20-40% higher continuous torque at the same package temperature because eliminating the intermediate motor housing cuts the thermal resistance path in half. This is not a marginal improvement — it directly determines whether a hip joint can walk continuously or must stop to cool down.

Decision Framework: When to Use Each Motor Type

Use this five-question framework to decide between frameless and framed motors for each joint position in your humanoid robot:

1. Is axial length the binding constraint? If your joint envelope is shorter than 40 mm from mounting face to cable exit, a framed motor almost certainly will not fit. Use frameless.

2. Does the joint need continuous operation under load? Walking gaits, load-carrying arms, and repetitive manipulation cycles all require high continuous torque. Frameless motors' superior thermal path makes them the default choice for any joint that operates above 50% duty cycle.

3. Is the team prepared to own the integration? Frameless motors require precision housing machining (concentricity under 0.02 mm), stator press-fit tooling, encoder phase calibration, and hi-pot testing. If the team lacks this capability and has no ODM partner, framed motors reduce integration risk at the cost of performance.

4. What is the program volume? For fewer than 10 units of a demonstration robot, framed motors with adapter plates may be faster to implement. For 20+ units heading toward pilot production, the investment in frameless tooling and fixtures pays for itself in size, weight, and thermal performance.

5. Does the joint require cable-through or hollow-shaft routing? Frameless motors inherently support hollow-shaft architectures for cable-through routing. Framed motors with solid output shafts cannot accommodate this without significant modification.

Conclusion

Transitioning to frameless motors requires a higher initial engineering investment in precision CNC machining, bearing alignment, and thermal potting. However, for a serious humanoid robot ODM program, it is the only viable path to achieving the power density, thermal reliability, and clean cable routing required for a commercial humanoid product.

Partnering with a specialized ODM factory allows robotics teams to bypass the painful trial-and-error phase of frameless integration. By relying on a partner equipped with CMM metrology, dynamometer test benches, and precise stator insertion tooling, teams can move straight from 3D CAD to validated, high-performance pilot assemblies.