Cross Flow Fan Noise: An Engineering Guide to Sources, Specs & Real Reduction
Your product passed the lab. Then field units came back: tenants above a restaurant complain about a low-frequency hum from the floor convectors at 2 a.m., or a hospital’s procurement team rejects your fan coil unit because the display case next to the bed “whistles.” The datasheet said 38 dB(A). So who is wrong?
Nobody is lying — but noise is the most misread specification on a cross flow fan datasheet, and the most expensive one to get wrong late in a project. Acoustic rework after tooling freeze can cost an OEM months. This guide is written for design engineers and product managers who need to diagnose where cross flow fan noise actually comes from, read supplier noise figures with the right skepticism, and apply reduction techniques that hold up in production — not just in the prototype.
What Datasheet dB(A) Actually Measures — and What It Hides
Almost every cross flow fan datasheet, including the LONGWELL catalog we’ll draw data from, lists noise as sound pressure level in dB(A) measured at a fixed distance (typically 1 meter in an anechoic or semi-anechoic chamber — LONGWELL, for example, runs a dedicated B&K noise lab for exactly this purpose).
Two traps hide inside that single number:
1. Pressure vs. power. Sound pressure (Lp) is what a microphone hears at one point; sound power (Lw) is what the fan radiates in total. Two fans with identical Lp can differ by several dB in Lw because of directivity. If your end-product acoustic spec is written against Lw (common in European HVAC procurement and in ErP-adjacent documentation), a datasheet Lp value is not directly comparable.
2. Distance and environment. Sound pressure drops roughly 6 dB per doubling of distance for a free-field point source — but a cross flow fan mounted inside a cabinet, behind a grille, or in a trench convector is anything but free-field. Cavity resonance, ducted inlets, and reflective mounting surfaces can add 3–8 dB to the “same” fan measured in open air.
The practical rule for OEM work: treat datasheet noise as a comparison metric between fans under identical test conditions, and schedule an application-level measurement of your own assembly early. Never sign an end-product noise guarantee based on component-level figures alone.
Where Cross Flow Fan Noise Actually Comes From
Cross flow (tangential) fans have a distinctive acoustic signature because of their unusual geometry: an impeller that is long relative to its diameter, with blades passing through an asymmetric volute-free arc. Their noise splits into three families:
| Noise family | Physical origin | Spectral character | Typical fix in OEM design |
|---|---|---|---|
| Aerodynamic (dominant) | Blade passage over the rear guide arc, vortex shedding at impeller edges, turbulence interaction with the outlet grille | Broadband whoosh + tonal “blade passing” components | Larger, slower impeller; grille bar spacing detuned from blade passing frequency; inlet flow conditioning |
| Mechanical | Bearing wear or contamination, residual imbalance, rotor-to-housing rub, motor shaft misalignment | Narrowband hum tied to rotation frequency (1× RPM) and bearing defect frequencies | G6.3 dynamic balancing, sealed ball bearings, tolerance control on housing |
| Electromagnetic | Magnetostrictive hum, PWM commutation whine in BLDC/EC drives, coil hum at line frequency in AC shaded-pole motors | High-frequency whine (PWM) or 100/120 Hz hum (AC 50/60 Hz) | Drive frequency optimization, damping washers, EC motor with quiet commutation logic |
For most cross flow fans running at normal HVAC speeds, aerodynamic noise dominates. That is actually good news: aerodynamic noise responds strongly to speed, and speed is the one variable you control in software.
Real Numbers: How Diameter and Length Move the Noise Figure
To make this concrete, here are catalog figures from the LONGWELL LWCD DC cross flow series (3-phase BLDC motor, 24 V, ball bearing, aluminum alloy impeller). All values are manufacturer sound-pressure data under catalog conditions — read them as relative comparisons, which is what they’re good for:
| Series (rotor Ø) | Example model | Speed (RPM) | Air flow (m³/h) | Static pressure (Pa) | Noise dB(A) |
|---|---|---|---|---|---|
| LWCD-30 (Ø30 mm) | LWCD-3090MN-07 | 2,400 | 27 | 12 | 26 |
| LWCD-30 | LWCD-30420MN-07 | 2,400 | 125 | 17 | 34 |
| LWCD-45 (Ø45 mm) | LWCD-4590MN-06 | 2,000 | 50 | 21 | 37 |
| LWCD-60 (Ø60 mm) | LWCD-60180MN-07 | 2,000 | 152 | 29 | 38 |
| LWCD-60 | LWCD-60420MN-07 | 2,000 | 355 | 35 | 43 |
| LWCD-100 (Ø100 mm) | LWCD-100380MN-06 | 1,500 | 800 | 106 | 55 |
| LWCD-120 (Ø120 mm) | LWCD-120380MN-06 | 1,350 | 900 | 68 | 56 |
| LWCD-120 | LWCD-120780MN-06 | 1,350 | 1,900 | 140 | 60 |
Three engineering lessons fall out of this table:
- Longer isn’t louder per unit of air. Going from the 180 mm to the 420 mm impeller in the same Ø60 mm series nearly doubles airflow (152 → 355 m³/h) for only 5 dB(A). Length scales airflow almost linearly while noise grows slowly — this is precisely why cross flow fans win in long, thin air paths like air curtains and trench convectors.
- Bigger and slower beats smaller and faster. The Ø120 mm series delivers 900 m³/h at 56 dB(A) turning at 1,350 RPM; a small-diameter fan pushed to deliver similar airflow would need dramatically higher tip speed and would pay for it acoustically.
- Pressure duty matters. The LWCD-100 series runs at 106–229 Pa — real ducted duty — and sits at 55–61 dB(A). Noise figures quoted for low-pressure free-delivery conditions will not survive a high-resistance system; always request the P-Q curve and noise at your actual operating point. (If you need a refresher on curve reading, see [LINK-1].)
AC vs. DC/EC: The Noise Behavior Difference Buyers Miss
Motor family changes not just efficiency but the noise character. LONGWELL’s AC cross flow fans (LWCA series, shaded-pole motor, 115/230 V) and DC/EC fans (LWCD/LWCE, 3-phase BLDC) behave differently in three ways:
| Aspect | AC shaded-pole (LWCA) | DC BLDC / EC (LWCD / LWCE) |
|---|---|---|
| Speed behavior | Fixed by line frequency & taps (2/3-speed winding options); cannot run below design speed | Continuously variable 0–100% via 0–10 V or PWM control |
| Partial-load noise | Unchanged — runs at full speed regardless of thermal demand | Drops substantially at reduced duty; partial-load operation typically yields a 6–10 dB(A) reduction versus full speed |
| tonal signature | 100/120 Hz line-frequency hum + aerodynamic broadband | Soft-start ramp, quiet commutation logic; FG tach output lets the system prove actual speed |
| Best fit | Cost-driven appliances with fixed duty (ovens, fireplaces, basic heaters) | Noise-sensitive, variable-load products: fan coils, display refrigeration, medical, elevators |
That partial-load row is the single most underrated noise strategy available. A fan coil unit that spends 70% of its life at 40–60% speed doesn’t just save energy with an EC fan — it sits several decibels below its worst-case figure most of the time. This is why noise-sensitive segments (medical devices, elevator cabins, display cases) have migrated almost entirely to DC/EC platforms.
Application Noise Targets: What “Quiet” Means in Context
Before optimizing anything, anchor your target to the end environment. Typical design targets used in OEM acoustic discussions:
| Environment | Approx. ambient/design target | Implication for fan selection |
|---|---|---|
| Bedroom / hotel guest room (night) | 25–30 dB(A) background | Fan contribution should target ≤30 dB(A) at 1 m → Ø30–60 mm series at reduced speed |
| Elevator cabin | ~35–40 dB(A) with ventilation running | Dedicated low-profile elevator fans, soft-start EC drive |
| Office / retail with FCU | 35–45 dB(A) NC range | Ø60–100 mm at partial speed; grille design matters |
| Commercial kitchen / industrial | 55+ dB(A) acceptable | Noise rarely the binding constraint; prioritize pressure duty |
An elevator or hotel application asking for 30 dB(A) is not solvable by “a quieter fan model” alone — it demands the whole toolkit below.
Seven Engineering Levers to Cut Cross Flow Fan Noise
1. Slow the fan down with real speed control. Aerodynamic noise scales roughly with the fifth to sixth power of tip speed. Cutting speed 20% can take out 5–7 dB(A) — far more than any downstream silencer for the same effort. This is only practical on 0–10 V/PWM-controlled DC/EC platforms; on AC shaded-pole motors your only lever is a 2/3-speed winding tap. See [LINK-2] for the control-interface decision guide.
2. Choose a larger, slower fan instead of a smaller, faster one. The table above is your argument to whoever owns the BOM: moving from Ø45 to Ø60 mm, run slower, delivers the same airflow several dB quieter and usually with better efficiency.
3. Detune the grille. The outlet grille is the fan’s aerodynamic signature generator. Bar spacing that coincides with blade spacing produces coherent tonal radiation; uneven or swept bar patterns, or spacing chosen so grille-passing tones don’t sum with blade-passing tones, typically buy 2–4 dB of tonal reduction at zero cost delta.
4. Specify balancing and bearings precisely. LONGWELL’s cross flow impellers are balanced to G6.3 per JB/T 9101-1999 and run on sealed double-ball bearings; the LWCD series bearing life is rated 30,000–40,000 h (L10). If your supplier can’t state a balancing grade or an L10 life, mechanical noise growth over life is uncontrolled — and it will show up in warranty-year field units, not samples.
5. Condition the inlet. Cross flow fans are sensitive to inlet distortion. Struts, wiring harnesses, or foam edges within one impeller diameter of the inlet create discrete tonal noise. Keep the inlet path clean, and use a bell-mouth or radiused inlet edge where geometry allows.
6. Isolate mechanically. Rigid mounting transmits 1× RPM vibration into sheet-metal panels that then radiate as a loudspeaker. Rubber grommets or compliant sleeve mounts cost cents and typically remove several dB of structure-borne hum — this is the classic fix for the “2 a.m. convector hum.”
7. Buy noise data at your operating point. The most powerful lever is contractual: require supplier measurement (B&K-lab-grade, semi-anechoic) of sound pressure and sound power at your actual P-Q operating point, at full and reduced speed. Suppliers with in-house anechoic labs can do this within a sample-order budget.
[IMAGE PLACEHOLDER] IMAGE-1: Annotated cutaway diagram of a cross flow fan showing the three noise source families (aerodynamic arc, bearing, motor) with callouts. Alt text: "Diagram of cross flow fan noise sources: aerodynamic blade-passing noise, bearing noise, and motor/electromagnetic noise."
How to Write Noise Into Your OEM RFQ
A specification that survives production should include:
- Metric and distance: sound pressure dB(A) at 1 m and sound power Lw if your end market needs it.
- Test condition: free-field or in-assembly; chamber class.
- Speed points: noise at 100%, 75%, and 50% commanded speed for controlled fans.
- Tonal limit: a penalty or cap on discrete tones (e.g., tone-to-broadband ratio), because tonal noise is perceived as more annoying than broadband at equal dB(A).
- Life stability: noise re-measurement at 2,000–5,000 h aging to catch bearing degradation.
- Verification path: supplier’s in-house lab capability (e.g., B&K instrumentation) plus a first-article third-party check for high-volume programs.
Suppliers who can document all six are the ones whose datasheets you can actually trust.
FAQ
Is a lower dB(A) number always a quieter fan?
Not necessarily. dB(A) at 1 m ignores tonal character, directivity, and test environment. A 38 dB(A) fan with a strong 2 kHz blade-passing tone can be subjectively more annoying than a 41 dB(A) fan with smooth broadband output. Compare fans under identical test conditions, and evaluate tonal content for noise-critical products.
Why is my fan quieter on the bench than installed in my product?
Installation effects — cavity resonance behind the grille, reflective panels, distorted inlet flow, and structure-borne vibration through rigid mounts — commonly add 3–8 dB. Always validate acoustics in the final assembly, not on the component.
How much noise reduction do I really get from slowing the fan down?
Aerodynamic noise scales steeply with speed (roughly the 5th–6th power of tip speed), so a 20% speed reduction typically yields 5–7 dB(A). This is the largest single lever, and it’s why EC/DC platforms with 0–10 V/PWM control dominate noise-sensitive applications. Partial-load running can mean a 6–10 dB(A) improvement versus continuous full-speed operation.
Do AC and EC cross flow fans of the same size have the same noise level?
At full speed they are broadly comparable in aerodynamic noise, but they differ in character: AC shaded-pole fans add a 100/120 Hz hum and cannot run below design speed, while EC fans with soft-start and quiet commutation logic eliminate much of the electromagnetic signature and can spend their lives at reduced speed.
What bearing life should I expect before noise degrades?
Quality sealed ball bearings in the LWCD/LWCA series are rated at 30,000–40,000 h L10 life. Bearing wear manifests first as rising mechanical noise and a growing 1× RPM signature. If your product life exceeds ~25,000 operating hours, specify aged-noise verification in your supplier agreement.
Key Takeaways
Cross flow fan noise is three problems — aerodynamic, mechanical, electromagnetic — wearing one datasheet number. Read dB(A) as a relative comparator, demand sound data at your true operating point, and spend your engineering budget on the levers with real dB returns: speed control via a larger slower EC fan, grille detuning, G6.3 balancing with sealed ball bearings, and inlet conditioning. Done in that order, the difference between “compliant” and “comfortable” is usually 8–10 dB — and entirely within your control.