A small mechanism on the chest of a finished garment had to look alive. A second wearable had to flex a soft structure over and over while staying comfortable for hours. A constant-speed motor reads instantly as a machine, which kills the illusion before any of the practical constraints even arrive. And those constraints were tight: fit behind a shallow shell, stay light high on the garment, keep vibration out of the chest, run near-silent for hours, switch on and off dependably, with no bulky box and no exposed wiring.
People read life from the shape of a motion, and a constant-speed motor has no shape.
Human eyes are very good at telling living motion from mechanical motion, and they do it from timing far more than from appearance. Mechanical motion tends to be uniform: the same speed, the same cycle, no hesitation. Living motion accelerates and decelerates, moves faster one way than the other, and pauses. A heartbeat is a clear example. The contraction is quick, there is a brief hold, the release is slower, and then there is a real rest before the next beat. At rest, systole takes roughly one third of the cycle and diastole roughly two thirds, about 0.8 seconds in all at 75 beats per minute. A motor turning a cam at a steady rate repeats the same motion with none of that asymmetry, and the eye reads it as a machine at once.
There are two ways to get a shaped motion. One is to build the shape into the mechanism, with a specially profiled cam or a gear train, which adds parts, mass and noise. The other is to use an actuator whose position can be commanded, such as a small servo, and have firmware move it along whatever profile is wanted, moment by moment. The shape then lives in software, where changing it costs nothing. A compact linkage between the servo and the moving surface supplies the mechanical advantage, trading travel for force, so a small actuator can move a larger structure without a gearbox.
Servos bring a hidden problem. A small servo holds its position by constantly correcting against any force on it, which means current flowing and often an audible hum, even when it looks perfectly still. Stopping the control signal does not always stop this, because many servos keep holding the last position they were given. The dependable way to make one silent is to remove its power during the rest, with a transistor switch such as a MOSFET on its supply.
There is no model in any of this. It is motion design, power electronics and firmware timing, and the result is judged by a human eye.
- Fast contractionthe quick stroke, light brightening
- Brief holdthe punctuation the eye looks for
- Slower releaselight easing on the same curve
- A real pauseservo supply gated off, silent
- Next beatabout 0.8 seconds per cycle at 75 beats per minute
The brief was to make it move and look alive, and the reflex answer is a micro-motor and a cam. What makes a beat convincing is that it is asymmetric and punctuated: a fast contraction, a brief hold, a slower release, then a real pause. No constant-speed rotation produces that shape. So the difficult part of the job moves out of the mechanism and into the controller.
Wearable mechanisms fail on look, sound and feel long before they fail mechanically.
Most small mechanisms are judged on whether they move: enough torque, enough travel, enough cycles. A wearable that has to look alive is judged on something else entirely. A motion that is technically correct but uniform reads as fake. A faint hum in a quiet room is louder than any specification sheet suggests. Vibration that is trivial on a bench is obvious through fabric against a chest. These are failures of perception, and they show up on a person, never on a test stand.
The second family comes from treating the phone as part of the mechanism. If the motion profile is generated in an app and streamed to the device, every dropped connection is a stopped heartbeat. Wearables live in pockets, crowds and radio interference, so a design that depends on a continuous link will visibly fail in normal use.
The reflex, and the fix.
Start from the motor and gear it down
It is the obvious mechatronics answer, and there is a part in a catalogue for it.
Sizing from the motor brings a gearbox, mass high on the garment and gear whine, and none of it produces the asymmetry that makes the motion read as alive.
Start from the motion, shape it in firmware
A firmware waveform engine, mechanical advantage from a compact linkage, and power-gating hardware on the servo supply.
A lifelike beat from a small actuator, near-silent for hours, with the gearbox and its mass designed out.
Begin with the motion people should see. The gearbox, the mass and the whine all disappear once the motor stops being the starting point.
Turning a rotary actuator into an organic pulse
The beat lives in firmware as an asymmetric profile. The light is tied to the same beat state, brightening through the contraction and easing through the release, so movement and light are one event and never two that merely happen to be synchronised.
Designing for silence, runtime and reliability
The honest way to silence the servo is to switch off its supply with a MOSFET through the rest phase. Cutting the PWM signal is not enough, because many small servos keep holding position, drawing current and humming after the signal stops.
A fail-safe that keeps the beat on the board
The waveform runs on the device, so a dropped phone signal never stops the heart.
Tuning until it looks right
The beat was dialled in by eye against the real piece at the target tempo, on a bench rig, before any part was finalised or any mould was cast. A fixed rotation speed was never the reference.
What each choice cost, and what it bought.
AStarting from the motion
The reflex is to choose a motor and gear it down until it has enough torque. Starting from the motion to be seen, a fast stroke, a hold, a slower release, a pause, made the actuator choice follow from it. A position-commanded servo shaped by firmware could produce that profile, and a motor at constant speed could not. The cost was a firmware waveform engine to build. The gain was that the gearbox, its mass and its whine were designed out before they were ever designed in.
BMechanical advantage from a linkage
A small actuator moving the surface directly would not have had the force, and a bigger actuator brings mass and noise. A compact linkage trades travel for force, so a small servo can drive the visible motion. The linkage needed careful geometry to fit behind a shallow shell. It kept the moving parts small and light high on the garment, where every gram is felt.
CPower-gating through the rest
Cutting the control signal is the obvious way to quiet a servo, and on many small servos it does not work, because they keep holding position and humming. Switching the servo's supply off with a MOSFET through the rest phase guarantees silence and saves battery. It costs a little extra hardware and a careful wake so the servo picks up cleanly on the next beat. It is the reason the piece runs near-silent for hours.
DKeeping the beat on the board
The phone could have driven the waveform, which would have made changes easy to push. The waveform lives on the device instead, with the light tied to the same beat state, so movement and light are a single event. Changing the beat now means changing firmware. In return, a dropped phone signal never stops the heart.
A single headline number hides where a system fails. This work was scored on the dimensions that actually decide whether it holds in production, measured on real, held-out cases rather than the demo path.
A geared motor turning at a constant rate. It sounds like a machine, adds mass high on the garment, and keeps humming through the pause because only the signal was cut.
A lifelike beat made by a firmware motion profile and mechanical advantage, with no need for a bigger motor. Near-silent for hours through power-gating, the light breathing on the same curve, and a fail-safe that keeps the beat going if the phone signal drops. Delivered end to end from mechanism through motor, driver, board, firmware and power, with wiring diagrams, bill of materials and CAD.
What it owns, and what it hands to a person.
The timing is grounded in physiology. At rest, systole takes roughly one third of the cycle and diastole roughly two thirds, about 0.8 seconds per beat at 75 beats per minute. The illusion holds at rest tempos and was tuned by eye against the finished piece, so a change to the shell or the mounting means a retune on the bench, not a parameter change.
When the deliverable is a perception and not a measurement, specify the perception first and let it choose the hardware. The asymmetry that made this read as alive was free in firmware and expensive in mechanism.
People judge whether a motion is alive from its timing: asymmetry, holds and pauses. A uniform cycle reads as a machine however well it is built.
When the shape of a motion lives in firmware, changing it costs nothing. When it lives in the mechanism, every change is a new part.
Stopping a servo's control signal does not always stop it drawing current or humming. Switching off its supply during rest is the dependable route to silence.
If a wearable's core behaviour depends on a phone link, it fails whenever the link does. Keep the essential behaviour on the device.
Where the goal is a perception, tune against the real finished object, because a change to its shell or mounting changes the result.