Ten machines from the same batch: some run quietly, some hum, some click as they turn. When that happens, my troubleshooting order is fixed at three steps: classify the fault by sound signature first, then check assembly stress, then work through moving parts, cabling and resonant cavities. Eight cases out of ten are located in the first two steps. Guessing at a fully assembled machine is the slowest route, because the same noise can come from an unbalanced motor, from a screw pulling the housing out of shape, or from a cable rubbing against the edge of a hole.
Why designers should care: noise is a return and a complaint, and almost every cause traces back to a tolerance, a preload or a cable route that was never fixed on the drawing.
The order: sound classification, assembly stress, moving parts and cabling, resonant cavities. Continuous friction points to interference or flash, periodic clicking points to rotating parts and hard stops, and humming at one particular speed points to housing resonance. The three field tests are: loosen the torque and listen for a change, press the housing and listen for the noise to stop, and remove one layer at a time and listen for it to disappear.
Step 1: Classify by Sound Before Taking Anything Apart
The core answer: spend ten minutes classifying the noise and save half a day. There are three families.
Continuous friction, a rustle or squeak that persists whenever the machine is moving, points to interference, flash or an over-pressed fit. Two surfaces that should have 0.1 to 0.2 millimetres of clearance have closed up after assembly, and they rub continuously.
Periodic clicking, once per revolution or once per open-and-close, follows the motion, and points to gear meshing, snap fits, hard-stop bumps or an eccentric bearing seat. On one of our small appliances the click tracked the motor and turned out to be the gearbox output shaft running with more than 0.5 millimetres of one-sided clearance in the housing bore.
Humming under specific conditions, only at a certain speed, load or attitude and gone as soon as the speed changes, is almost always about housing stiffness, cavity volume and support point positions.
After classifying, do two things: record a clip with a phone pressed against the machine and slow it down to hear the rhythm, then press the palm of your hand against different areas of the housing in turn. Wherever the noise drops noticeably is the main radiating surface, and that is where you look at supports and ribs. In the field this beats any instrument.
Step 2: Assembly Stress - Loosen the Torque Half a Turn First
The core answer: after classifying, do not touch the motor, touch the screws. Assembly stress is the most easily overlooked and most easily verified family.
The typical symptom is a machine that is quiet when assembled and then, overnight or after a few degrees of temperature change, clicks once by itself or develops a light plastic rubbing sound. A plastic part has been forced into a deformed position and the stress is releasing slowly. The cause is usually insufficient housing flatness: tightening four screws twists the part.
Snap fits behave the same way. On ABS parts we hold single-sided snap interference inside an empirical band, because too much of it creates preload that speaks up as soon as the machine warms. Where materials with different shrinkage meet, such as a PC/ABS part snapped onto a pure ABS part, the male and female halves of the same rib shrink differently, and the temperature rise pushes them against each other.
The test: loosen each of the four suspect screws half a turn and power up again. If the noise drops noticeably or disappears, the root cause is assembly stress, and the direction is flatness, boss height tolerance and snap interference. If nothing changes, go straight to the next step and stop polishing the housing.
Boss height tolerance deserves its own look. Four bosses on one housing panel differing by 0.2 millimetres mean only the two tall ones carry load once tightened, and the panel is bent into a slight arc. The deformation is invisible, and the symptom only appears as a click when you press the housing by hand.
Step 3: Moving Parts, Cabling and Resonant Cavities - Rebuild One Layer at a Time
The core answer: our internal method is layered rebuild. Strip the machine down to the core module, listen once, then listen again after each layer goes back on: bracket, harness, top cover, bottom cover. Whichever layer makes the noise worse contains the root cause.
Cabling is where we trip most often. A cable that is not secured knocks against the housing as soon as the machine vibrates, a rhythmic slap that follows the vibration frequency. Routing channels, cable clips and grommets together solve most of it. Clips spaced too far apart still let the middle section swing, so we place fixing points near the grommet and near the connectors at each end.
Motor and fan isolation pads are the next item. A pad that is too hard transmits vibration straight into the housing; too soft and the module moves further, which needs swing clearance so it does not strike anything inside the cavity. There is no universal value: try several durometer grades against module weight and rotational frequency and judge by ear and hand.
Resonant cavities are the most direct diagnosis of all. Nudge the speed slightly and the noise disappears: the original speed was hitting the natural frequency of a panel or cavity. The usual fix is to raise the stiffness of the flat area. Ribs should be 0.5 to 0.6 times the main wall thickness, or a large flat panel can be changed to a curved surface. Ribbing is more economical than thickening the whole wall, which brings sink marks and a new cosmetic problem.
Root-Cause Criteria
Noise that changes with speed or load points to the excitation side: motor, fan balance, gear meshing. Noise that changes when screws are loosened points to assembly stress. Noise that stops when the housing is pressed points to housing resonance or insufficient stiffness in the radiating surface. Noise that disappears when a particular layer is removed points to that layer's fits or cable fixing. A single click after standing or a temperature change points to stress release or a shrinkage mismatch between materials.
What Designers Can Take Away
Classify before you dismantle. Ten minutes of listening and pressing saves half a day of guessing.
Loosen the torque first. Half a turn on four screws tells you whether the root cause is assembly stress.
Rebuild layer by layer. The layer that makes the noise worse contains the fault, and cabling is the usual suspect.
Fix four items on the drawing. Snap interference, housing flatness, boss height tolerance and cable routing. A noise complaint costs far more than the two extra hours on the drawing.
FAQ: Structural Noise Questions
Q1: The machine was quiet when assembled, then clicked once overnight. Why?
A: Most likely stress release: a plastic part forced into a deformed position relaxes back. Check housing flatness and boss height tolerance first, then whether snap interference is excessive. Where two materials with clearly different shrinkage are mated, a temperature change triggers it too. The verification is to loosen the screws and leave it the same amount of time.
Q2: How do you tell motor noise from housing resonance?
A: Shift the speed slightly. If the noise changes or disappears, it is resonance; if it follows along, it is the excitation source. Pressing different areas of the housing with the palm is the other method: wherever the noise drops is the main radiating surface.
Q3: Do you need professional equipment to troubleshoot noise?
A: Structural noise is located in the field with a phone recording, layered rebuild and hand probing. Frequency analysis is for excitation-side faults such as motor electromagnetic noise or bearings, and those go to the supplier. On the structural side the point is isolating variables.
Q4: How do you avoid noise problems at the design stage?
A: Fix four items on the drawing and in the 3D: interference, flatness, boss height tolerance and cable routing. Route cables during modelling with fixing points defined rather than pulling them at the assembly bench, and rib large flat panels at 0.5 to 0.6 times the main wall thickness.
Hertz Industrial Design (Dongguan, China) works across product appearance design, structural design and CMF, and more than 320 of its designs have reached mass production. Website: www.hezidesign.com.
Originally published on the Hertz Industrial Design website (hezidesign.com), Design Experience column.
Top comments (0)