Low-noise gearmotors for clinical analyzers

An analyzer's noise complaint is a narrowband mesh tone coupled into a panel, not a loud motor. The first gear stage carries the least torque and almost all of the audible sound.

Wedge
Medical and laboratory equipment
Catalog reviewed
2026-08-18
Configurations screened
32 published

A clinical analyzer carries several small indexed drives: a reagent carousel, a cuvette wheel, a probe arm, a mixer, a wash station. Each is a gearmotor turning an inertia through a short move and stopping. Individually the running torques are tens of millinewton-metres. The acoustic problem is not that any of them is loud. It is that each one emits a narrowband tone at its gear mesh frequency, and that tone finds a path into the instrument's sheet-metal panels, which radiate it into a room where a technologist sits for eight hours.

There are two models on this page and both matter. The mechanical load model is the ordinary indexing one: reflected inertia times angular acceleration plus bearing and seal friction. The acoustic model is separate. Each gear stage generates a tone at f = z·n/60, where z is the number of teeth on the pinion and n is that shaft's speed in rpm, with sidebands spaced at the shaft rotation rate from runout and eccentricity. Because the first stage runs at motor speed, its mesh frequency is higher than every later stage by the product of the intervening ratios, and it lands in the part of the spectrum where A-weighting adds nothing and the ear subtracts nothing.

What normally controls the selection

Where the first mesh order lands relative to the A-weighting curve and the panel's modal density, and how stiffly the gearmotor is coupled to the panel. Not the motor. Two configurations with identical published data and identical measured sound power at 1 m in free field can differ by 10 dBA once they are bolted into the instrument, because the transmission path changed. The selection question is therefore which reduction ratio and stage split puts the loudest order somewhere the structure does not amplify, and second, whether the mount can be made compliant without losing positioning stiffness.

Continuous, peak, starting and abnormal

Sizing on one of them is how a mechanism gets a part that works on the bench and fails in the field.

Row Continuous Sizes the thermal case Peak Sizes the torque case Starting Sizes the transient Abnormal Sizes the protection
What the mechanism is doing Almost nothing. A carousel index draws a few tens of millinewton-metres for a fraction of a second every several seconds. Thermally these drives are idle.The acceleration phase of the index, and probe-arm moves that fight a cable harness. Both are short and both are usually well inside the gearhead rating.Every index starts from rest, traverses backlash, and the tooth impact at take-up is a broadband transient sitting on top of the steady mesh tone. On a quiet instrument this click is often the complaint, not the tone.A carousel jams on a mis-seated rack or a probe hits a stuck cap. On the published spur configurations a stall can put up to 8.0 times the gearhead rating into the output stage.

The specifications that get compared are rarely the ones that decide the outcome.

What follows is what actually fails, and which published number sends people the wrong way.

Common failure modes
Failure mode Physical cause What it looks like in the field
A pure tone that appears only after final assemblyStructure-borne transmission. The gearmotor bracket is stiffly coupled to a panel with a mode near the first mesh order.The drive measures quiet on the bench and loud in the instrument. Damping the panel with a hand changes the level. The tone frequency does not move when the panel is damped, only its amplitude.
Sidebands and a rough, warbling characterPinion runout or a bent shaft modulating the mesh. The modulation appears as sidebands spaced at the shaft rate around the mesh line.A spectrum with a mesh peak flanked by evenly spaced smaller peaks. Audibly a beating quality rather than a clean whine. Unit-to-unit variation is large.
Rattle on reversal and at low speedBacklash traversed under torque reversal, or a lightly loaded stage where the teeth separate and re-engage.Broadband clicks at direction changes, worse when the drive is unloaded, better when a small preload is applied.
The instrument gets louder over the first monthsGrease migrating out of the first-stage mesh. The first stage runs fastest, throws grease hardest, and is usually the smallest reservoir.A slow rise in the level of the highest mesh order with no change in the lower orders and no change in current.
Specifications that mislead here
Specification Why it misleads in this application
A single dBA figure for the gearmotorIt is measured in a fixture that is not your instrument, at a distance and in a field that are not your room. It carries no information about which order is loud, and the order is what you can actually change.
Sound pressure at 1 m, free fieldIt describes the airborne path. In a closed instrument the dominant path is structure-borne, through the bracket into the panel. The two are not related by any fixed offset.
Helical gearing as a noise solutionHelical teeth reduce transmission error at the mesh, which lowers the excitation. They do not change where the mesh order lands, and they add axial load that can make a compliant mount worse.
Gearhead backlash in degreesIt bounds the reversal rattle but says nothing about the steady tone, which is generated by teeth that are firmly in contact.

The tradeoffs that matter here

Spur or planetary reduction

Row Spur Planetary
When it is right Low torque, and where a single mesh at each stage makes the spectrum simple enough to diagnose and to design around.Higher output torque, and where the load sharing between planets averages some of the individual tooth error.
What it costs One tooth pair carries the load at each mesh, so transmission error goes straight into the excitation. Lower output torque rating.The spectrum is richer. Mesh orders are set by the planet count and the ring tooth count, not by a simple pinion count, so identifying the offending order takes a real measurement.

Brushed coreless or brushless

Row Coreless brushed Brushless
When it is right Low duty, and where the absence of slot cogging matters. A coreless rotor has no iron teeth, so it contributes no cogging torque ripple to excite the train.Continuous or high-duty drives, and where brush noise growth over life is unacceptable.
What it costs Brush and commutator noise is broadband and rises as the brushes wear. It is the component of motor noise that changes over life.Commutation torque ripple at the electrical frequency and its harmonics adds discrete orders of its own, and the driver's PWM frequency can appear acoustically through magnetostriction.

How the drive is mounted

Row Rigid bracket to the chassis Elastomer isolators Isolate the panel instead
When it is right Positioning stiffness is the priority and the panel modes are known to be clear of the mesh orders.The tone is above the isolator's resonance by a factor of at least three, which for a mesh order above 1 kHz is easy to arrange.Several drives share one radiating surface, so treating the panel fixes all of them at once.
What it costs The stiffest possible structure-borne path. Any tone the drive makes is delivered into the panel with no loss.Compliance in the positioning path. On an indexed axis this shows up as settling time and as a static deflection under the index torque.Damping treatment adds mass and cost to a large part, and has to survive cleaning and service access.

Worked example

Find where the noise energy is for a small spur-geared carousel drive, and decide which stage to spend money on.

Configuration
MM-C1226-S00523-060A, 12 mm coreless, spur
Reduction ratio
52.25:1 in four stages
Motor speed at the rated point
6 800 rpm
Output speed at the rated point
130.1 rpm
Panel
1.0 mm steel, 200 mm × 300 mm
Calculation
Step Expression Result
Per-stage ratio52.25^(1/4)2.689 per stage
First-stage mesh frequency6 800 rpm ÷ 60 × 12 teeth1 360 Hz
Second-stage mesh frequency2 529 rpm ÷ 60 × 12 teeth506 Hz
Third-stage mesh frequency941 rpm ÷ 60 × 12 teeth188 Hz
Fourth-stage mesh frequency350 rpm ÷ 60 × 12 teeth70 Hz
Sideband spacing, first stage and output6 800 ÷ 60 and 130.1 ÷ 60113 Hz and 2.17 Hz
A-weighting applied to each order+0.7 dB at 1 360 Hz, −3.2 dB at 506 Hz, −12 dB at 188 Hz, −24 dB at 70 HzFirst stage is about 25 dB more audible than the fourth for equal radiated power
Panel flexural rigidityD = 200 × 10⁹ × (0.001)³ ÷ (12 × (1 − 0.3²))18.3 N·m
Panel fundamentalf₁₁ = (π/2)·√(D/ρh)·(1/0.2² + 1/0.3²)86.6 Hz
Panel modes below the first mesh orderN = A·f ÷ (1.8·c_L·h), with c_L = 5 291 m/sAbout 9 modes below 1 360 Hz

Result

The first stage is the acoustic problem and it carries the least torque of any stage in the train. Its mesh order sits at 1 360 Hz where A-weighting adds 0.7 dB, while the output stage sits at 70 Hz where A-weighting removes 24 dB. For equal radiated acoustic power the first stage is roughly 25 dB more audible than the last. The panel has about nine modes below 1 360 Hz, so tuning the drive speed to miss them is not realistic: a 15 percent speed change moves the tone 200 Hz and lands it on a different mode. The money goes into the first stage and into the mounting: tighter pinion runout, finer pitch or a helical first stage, a grease that stays in the first mesh, and an isolator whose resonance is below 400 Hz so that 1 360 Hz is attenuated.

What would change it

Real tooth counts move every frequency proportionally. A 10-tooth first pinion puts the first mesh at 1 133 Hz, still in the same region. Running the carousel slower moves all orders down together and helps, because the A-weighting curve falls steeply below 500 Hz, but it costs throughput. Changing from four stages to three at the same ratio raises the per-stage ratio and lowers the second and third orders while leaving the first where it is. A different panel changes the modal map completely, which is why this measurement has to be repeated on the real instrument.

Evidence, in the order it is worth collecting.

Each step names the measurement, not the intention.

  1. Measure the spectrum, not the level

    Narrowband sound pressure with the drive running at the production operating point, in the assembled instrument, with the orders identified against the calculated mesh frequencies.

  2. Separate airborne from structure-borne

    Run the drive on soft mounts outside the panel, then rigidly mounted, and compare. An accelerometer on the panel next to the bracket resolves it in one measurement.

  3. Establish unit-to-unit variation

    The same measurement on at least ten units. Mesh amplitude varies with runout and assembly, and a single sample tells you nothing about the production distribution.

  4. Measure after run-in and after aging

    Repeat at the end of the life test. Grease migration and brush wear both change the spectrum, and both get worse.

  5. Set an acceptance limit per order

    An overall dBA limit is not testable at incoming inspection. A limit on the first mesh order amplitude is.

Send the noisy unit and its operating profile.

Send a unit that exhibits the complaint and the speed it runs at. You get back an order analysis identifying which stage is responsible, and whether the path is airborne or structural.

Send the noisy unit and its operating profile