DEV Community: emily3339

How Should a Frameless Torque Motor Be Sized for a Robot Joint?

emily3339 — Fri, 28 Aug 2026 14:45:00 +0000

A frameless torque motor can make a robot joint smaller and more integrated.

It can also create problems if the motor is selected from peak torque alone.

Robot joint sizing requires a complete view of torque, speed, thermal load, mechanical integration and feedback.

The motor is only one part of the actuator.

Start With the Joint Torque Profile

The first question should not be the motor diameter.

It should be how much torque the joint actually needs during operation.

A robot joint rarely operates at one fixed torque.

Torque changes as the arm moves, accelerates, carries a payload and responds to gravity.

The designer therefore needs both peak torque and continuous torque requirements.

Peak torque may occur during rapid acceleration or an unusual operating condition.

Continuous torque is related to the average thermal demand over the operating cycle.

A motor that can produce the required peak torque for a few seconds may still overheat if its continuous capacity is too low.

Consider the Complete Motion Cycle

The most useful motor sizing information comes from the actual motion profile.

This includes expected joint speed, acceleration, load inertia, duty cycle and dwell periods.

Gravity also matters for joints that support a load vertically.

For some axes the motor spends significant time holding torque even when the robot is not moving.

That holding condition generates heat and must be included in thermal calculations.

Ignoring it can lead to a motor that performs well during a short demonstration but becomes thermally limited during continuous production.

Decide Whether the Joint Uses a Gear

Frameless motors are often associated with direct drive robotics but many robot joints still use a reduction gear.

A gear can multiply output torque and allow the motor to operate at a more favorable speed.

It also adds friction, compliance and mechanical complexity.

The selected gear ratio changes the torque and speed required from the motor.

It also changes how reflected inertia appears to the motor.

Motor sizing should therefore be performed together with gearbox selection rather than treating the two components independently.

For joints that truly use direct drive the motor must produce the full joint torque without mechanical multiplication.

That normally requires greater motor torque but removes the backlash and compliance associated with a gearbox.

Diameter and Length Affect Integration

Frameless motors consist primarily of a rotor and stator.

This gives the mechanical designer freedom to integrate the motor directly into the joint housing.

However motor dimensions still influence the complete actuator.

Torque is strongly related to motor geometry.

A larger motor diameter can often provide more torque without requiring a proportionate increase in axial length.

This can be useful in robot joints where keeping the actuator short is important.

The available bore diameter also matters.

Modern robotic joints often need space through the center for cables, hoses, brakes or mechanical shafts.

The motor should therefore be evaluated as part of the joint packaging rather than as an isolated component.

Thermal Design Can Limit Continuous Torque

Compact robot joints are difficult thermal environments.

The motor may be surrounded by gears, bearings, electronics and structural components.

There may be very little natural airflow.

The stator housing often becomes the main thermal path.

Good contact between the stator and the surrounding structure can therefore have a significant effect on continuous torque capability.

High duty applications may require additional cooling.

The designer should also consider how motor heat affects nearby encoders and lubrication.

A joint can remain below the winding temperature limit while still becoming too hot for another component.

Bearings and Encoders Are Part of the Design

A frameless motor normally does not provide its own bearing system.

The joint designer must select bearings capable of supporting radial loads, axial loads and moments generated by the robot structure.

Bearing stiffness can influence positioning accuracy and dynamic response.

Feedback is equally important.

Encoder resolution and accuracy should match the real performance requirement of the joint.

Mounting errors can reduce the benefit of an expensive encoder.

Mechanical alignment should therefore be considered during the earliest stages of joint design.

A frameless torque motor supplier such as HansMotor can be most useful when torque requirements are provided together with the available joint envelope, bore requirement, cooling conditions, speed range and feedback concept.

Validate More Than One Operating Point

A good motor selection should be checked at several conditions.

Maximum acceleration
Maximum operating speed
Continuous production cycle
Static holding condition
Highest expected ambient temperature
Maximum payload
Emergency or abnormal loading where relevant

A motor that passes only the maximum torque calculation is not necessarily suitable for the application.

Final Thoughts

Sizing a frameless torque motor is an actuator design problem rather than a catalog selection exercise.

Peak torque determines whether the joint can handle demanding movements.

Continuous torque determines whether it can operate repeatedly without overheating.

Motor geometry affects the mechanical envelope.Cooling affects usable output.

Bearings and feedback affect the final precision of the joint.

Treating these elements as one system usually produces a better robot joint than selecting each component independently.

Momentary and Latching Push Button Switches Explained

emily3339 — Thu, 27 Aug 2026 09:44:53 +0000

Push button switches may look similar from the outside, but their internal behavior can be very different.

One of the most important distinctions is whether the switch is momentary or latching. Choosing the wrong type can create unexpected behavior in the finished product, even when the footprint and electrical configuration appear correct.

Understanding where the switching state should be stored is the simplest way to make the right choice.

What Is a Momentary Push Button Switch

A momentary switch remains active only while the actuator is being pressed.

When the user releases the button, an internal spring returns the mechanism to its original position.

This behavior is common in:

reset buttons
menu controls
keypad inputs
test equipment
handheld electronics
control panels

In many modern products, the momentary switch simply sends a signal to a microcontroller. Firmware then determines what happens next.

For example, pressing a power button may tell the controller to turn a system on, even though the physical button immediately returns to its original position.

What Is a Latching Push Button Switch

A latching switch mechanically maintains its state after the user releases the button.

Press once and the switch changes state.

Press again and it returns.

This is why latching designs are also commonly described as push on push off or self locking switches.

A compact latching push button switch can be useful when the state needs to remain mechanically maintained rather than stored by software.

Typical applications include:

audio equipment
test instruments
control systems
consumer electronics
mode selection
hardware power control
The Main Difference Is Where the State Lives

This is the most useful way to compare the two technologies.

With a momentary switch, the physical button does not remember anything. The system usually remembers the state electronically.

With a latching switch, the mechanism itself maintains the state.

Consider a device with an electronic power button.

A momentary switch sends a short signal to the controller. The controller then decides whether to keep the system powered.

A mechanical latching switch can maintain the contact state even without a processor making that decision.

Neither approach is universally better.

The correct choice depends on the architecture of the product.

When Momentary Switching Is Better

A momentary switch is usually preferable when software already controls the system.

It provides several advantages.

The firmware can assign multiple functions to one button. A short press might perform one action while a long press performs another.

Behavior can also be changed through software without redesigning the switch mechanism.

Momentary tactile switches are therefore common in compact electronic interfaces. Engineers comparing tactile switch options can choose different sizes, mounting styles, operating forces and actuator heights depending on the design.

Momentary switching also works well when the external control needs to remain visually unchanged regardless of the system state.

When Latching Switching Is Better

Latching switches are useful when maintaining a physical state has value.

This can include equipment where users expect a clear mechanical on and off condition.

They can also be useful when electronic state retention would unnecessarily increase circuit complexity.

One important advantage is that the mechanical position can remain unchanged even when the device loses power.

However, designers should not assume every latching switch is suitable for switching mains power directly.

Electrical ratings, contact configuration and applicable safety requirements must still be checked for the specific component.

Do Not Identify the Switch by Appearance Alone

A common sourcing problem occurs when momentary and latching variants use nearly identical housings.

A buyer may receive a part that fits the PCB perfectly but operates differently.

This can be difficult to detect during basic incoming inspection.

The safer approach is to control the complete manufacturer part number in the BOM and include functional behavior in the inspection procedure.

A useful test is simple:

Press the switch.
Release it.
Check whether the electrical state remains changed.
Press it again.
Confirm whether the state returns.

A press and hold test cannot distinguish all momentary and latching variants.

Consider Travel and Enclosure Geometry

Latching mechanisms often need enough actuator travel to complete the locking and releasing sequence.

The external button or plastic cap must not prevent this movement.

This becomes important when a switch sits behind a custom enclosure.

Tolerance from several parts can accumulate:

PCB position
switch height
plastic button height
enclosure wall thickness
mounting hardware

An apparently small dimensional error can prevent reliable latching.

Prototype testing should therefore use the actual enclosure whenever possible.

Contact Configuration Still Matters

Mechanical behavior and electrical configuration are separate specifications.

A latching switch might use SPST, SPDT, DPDT or another contact arrangement depending on the design.

For example, a DPDT switch can change two circuits simultaneously.

Engineers should verify the contact diagram instead of assuming the pin configuration based only on the number of terminals.

Consider Lifecycle

A button used several times every hour requires a different lifecycle target from a configuration control changed only occasionally.

Estimate how frequently the user will operate the switch over the expected service life.

Then compare that requirement with the manufacturer's published mechanical and electrical life under relevant test conditions.

Electrical life can also depend heavily on the load being switched.

Final Thoughts

The choice between momentary and latching push button switches should begin with one question:

Should the mechanical switch maintain the state, or should the electronics maintain it?

Once that decision is clear, engineers can evaluate footprint, actuator travel, contact configuration, electrical rating and lifecycle.

Treating mechanical behavior as an explicit design requirement prevents one of the easiest switch selection mistakes to make.

Han's Motor

emily3339 — Fri, 14 Aug 2026 11:35:39 +0000

A gearbox is not only a mechanical part. It is also a stack of problems your control code spends its life fighting. Backlash at every reversal. Compliance you tune around. Couple the motor straight to the load and most of those problems leave, and a few new ones arrive to take their place.

A direct drive motor connects the rotor or forcer straight to the load, with no gearbox, belt, or lead screw in between. Here is what changes once that transmission is gone, from the point of view of the code that runs the axis.

Backlash stops being a software problem

Gearing gives you backlash, the lost motion you feel at every direction change. In firmware you pay for it with reversal compensation and settling time you can never fully tune out. Remove the gearbox and the gear backlash is gone at the source, so that compensation code goes with it.

Be precise about the claim. Direct drive removes the backlash the gearing introduced. It does not remove compliance elsewhere in the axis, in a coupling or in the structure, so a stiff mount still matters. What you get back is a mechanical path that does not lie to your position loop at reversal.

You inherit the full load inertia

This is the trade. With a gearbox, the load inertia reflected back to the motor is divided by the square of the gear ratio. A 10:1 stage divides it by 100. Take the gearbox out and that division is gone. The motor, and your position loop, see the full load inertia directly.

That is why direct drive motors are larger and cost more at a given torque, and why the load-to-motor inertia ratio now sets your loop bandwidth and stability margin. You are not tuning against a comfortably reduced inertia anymore. You are tuning against the real thing.

Commutation moves to the front

With no gearbox in the way, smoothness of force or torque comes down to the current loop and to knowing the rotor position within one electrical cycle. In field-oriented control, torque tracks the q-axis current. That holds for a surface-magnet PMSM. An interior-magnet motor adds a reluctance term, so treat it as a close approximation, not a law.

# Simplified. A real drive runs this in the current loop at tens of kHz.

# 1. Commutation needs the rotor angle within one electrical cycle,
#    referenced to the correct alignment offset.
theta_e = pole_pairs * encoder.angle() + commutation_offset

# 2. In field-oriented control, torque tracks the q-axis current.
#    True for a surface-magnet PMSM. Interior-magnet motors add a
#    reluctance term, so this is an approximation, not a law.
i_q = torque_command / torque_constant
i_d = 0  # no field weakening in this example

# 3. With a gearbox you would divide load inertia by the gear ratio
#    squared. Direct drive removes that division, so the position loop
#    is tuned against the full load inertia.

Two feedback jobs live here, and developers mix them up constantly. One is coarse rotor position used to commutate the motor, often from hall sensors. The other is the fine axis-position feedback that tells you where the machine actually is. They are different devices with different wiring and different resolution, and confusing them is a classic reason a motor will not start smoothly.

A debugging heuristic worth keeping

When an axis runs rough, a periodic torque variation that tracks the electrical angle is a typical clue that commutation, not the mechanics, is the source. Treat it as a clue, not a proof. Bearing friction, guideway drag, and load variation each produce their own periodic signatures, and you have to rule those out before you blame the feedback or the commutation offset.

Feedback becomes the ceiling on precision

With no gearbox to average out feedback error, the encoder is part of the axis, not a bolt-on. Your precision ceiling is whatever the feedback device can resolve and hold. Two decisions land straight in your firmware.

First, incremental or absolute. An incremental encoder sends counting pulses and needs a homing move after every power cycle. An absolute encoder assigns each position a unique code and reports true position the moment power returns, with no homing. On a direct drive axis that cannot afford a homing sequence at startup, absolute feedback is usually the default, and your boot code is simpler for it.

Second, resolution is not accuracy. A high step count per revolution tells you how finely the device subdivides, not how well it knows the true angle. Match the encoder to the axis and its error budget rather than to a headline number.

Time-grating encoders are one option worth knowing here. They are contactless and field-based, and they read angle from a phase shift in the time domain rather than by counting finely etched lines, so the precision comes from time-domain electronics. They are absolute, and on a direct drive axis they integrate as one feedback set with the motor. The honest boundary matters too. This is not the pick at the top metrology tier, where wafer-grade stages still run optical. It earns its place on large, hollow-shaft, or contaminated axes, where coolant and swarf would force an optical scale into a sealed, clean gap.

When direct drive is the wrong call

Direct drive is not free. The motor is larger, costs more at a given torque, and usually needs cooling. If your drawing tolerates backlash and you do not need the stiffness or the settling time, a geared axis is cheaper and simpler to build. The deciding question is usually blunt. Does a specification on your drawing fail if there is backlash? If nothing fails, you may not need a direct drive axis at all.

We build these. HansMotor has designed and built direct drive motors, linear motors, torque motors, and time-grating encoders in-house in Shenzhen since 2005, holds more than 230 patents, and is part of the Han's Laser group (SZSE: 002008). We also contributed to China's GB/T 34115-2017 standard for permanent-magnet linear motors. For the longer breakdown of when direct drive earns its cost, HansMotor has a full write-up.

Written by the HansMotor Engineering Team.