Noise Monitoring at Work and on Site: What a Sensor Can Prove

Last updated: 25 August 2026

In short: UK law sets three noise thresholds at work, at 80, 85 and 87 dB(A). Battery powered noise monitoring sensors on a LoRaWAN network cost around £42 each and report sound levels for years without wiring. They are not legal measuring instruments. Their job is to show you which areas and which hours need a calibrated meter, and to prove a control measure actually worked.

Every supplier selling noise monitoring hardware will quote you a decibel range and a battery life. Almost none of them will tell you the one number that decides whether the device is any use to you in an enforcement conversation, which is its accuracy tolerance.

This guide covers what the Control of Noise at Work Regulations 2005 actually require, where continuous noise monitoring fits, and where it does not. It is written for facilities and site managers who have been asked to “get some noise monitoring in” and want to spend the budget once.

What does noise monitoring have to prove under UK law?

Noise monitoring has to prove two separate things: what your people are exposed to over a working day, and whether the controls you put in place reduced it. The Control of Noise at Work Regulations 2005 place the duty on the employer to assess the risk, not simply to own a meter.

The regulations came into force on 6 April 2006 and replaced the Noise at Work Regulations 1989, dropping each threshold by 5 dB(A). Personal exposure is what matters, measured as a daily or weekly average at the ear.

That distinction is the root of most wasted spend. A sensor bolted to a wall measures the room. A dosimeter clipped to a collar measures the person. Only the second one answers the legal question directly.

What are the noise exposure action values in the UK?

There are three. According to the Health and Safety Executive, the lower exposure action value is 80 dB(A), the upper exposure action value is 85 dB(A), and the exposure limit value is 87 dB(A).

At 80 dB(A) you must assess the risk to workers’ health and give them information and training. At 85 dB(A) you must provide hearing protection and establish hearing protection zones, and put regularly exposed workers into health surveillance.

The 87 dB(A) figure works differently from the other two. It is a limit rather than a trigger, and it is measured taking account of the reduction that hearing protection provides. Nobody may be exposed above it.

The scale of the underlying problem is smaller than the marketing suggests but has not gone away. HSE estimates 15,000 workers in Great Britain had work-related hearing problems across 2022/23 to 2024/25, with a confidence interval of 8,000 to 21,000, and 50 new cases of occupational deafness were assessed under the Industrial Injuries Disablement Benefit scheme in 2024.

Why a low-cost noise monitoring sensor is not a compliance instrument

Because of its accuracy tolerance. Legal and enforcement measurement in the UK relies on sound level meters built to IEC 61672, which sets two grades. At the 1 kHz reference frequency, a Class 1 meter must sit within ±1.1 dB and a Class 2 meter within ±1.4 dB.

A typical LoRaWAN noise monitoring sensor does not come close. The Milesight WS302 datasheet states an accuracy of ±3 dB between 0°C and +45°C, over a range of 30 to 120 dBA, with a frequency range of 30 Hz to 8 kHz.

Three decibels is not a rounding error in this context. Decibels are logarithmic, so a 3 dB error is roughly a doubling or halving of sound energy. A reading of 84 dB(A) from such a device could legitimately be anything from 81 to 87 dB(A), which spans the entire gap between “assess and train” and “you have breached the limit”.

This is not a fault in the product. It is a category difference, and it is the single most useful thing to understand before you buy.

What is continuous noise monitoring genuinely good for?

It is good for the things a calibrated meter is bad at, which is being everywhere, all the time, for years. A Class 1 survey gives you a defensible number for one location on one day. Continuous noise monitoring gives you pattern and duration across the whole site.

Four jobs noise monitoring does well:

• Triage. Finding which of forty rooms, bays or boundaries are anywhere near the action values, so the expensive survey goes to the right five.
• Timing. Showing that the problem is the 06:00 shift changeover, not the whole day. Exposure is a time-weighted average, so when noise happens matters as much as how loud it is.
• Verification. Proving an enclosure, damper or new machine mount actually changed the level, with weeks of before and after data rather than two spot readings.
• Alerting. Flagging in near real time that something has changed, such as a failed silencer or a contractor running a compressor by an open window.

None of those four require ±1.1 dB. All of them require presence, and presence is exactly what a £42 battery sensor buys you.

Three ways to do noise monitoring, compared

Most sites end up needing two of these three, not one. They answer different questions and they are priced accordingly.

A dosimeter and a Class 1 survey are episodic by nature, because both need a person present. Continuous noise monitoring is the only one of the three that keeps running after the consultant has gone home, and that persistence is its entire value proposition.

The common mistake is treating these as competing purchases. Continuous noise monitoring makes the other two cheaper, because it tells you exactly where to point them instead of surveying the whole site on a hunch.

Noise monitoring on construction sites and Section 61 consent

Construction adds a second, separate regime aimed at neighbours rather than workers. Under Section 61 of the Control of Pollution Act 1974, a contractor can apply to the local authority for prior consent, setting out the works, the method, and the steps proposed to minimise noise.

The authority has 28 days to respond and may attach conditions, limit the scope, or restrict the duration. Breaching a condition is an offence, and a refusal or a conditional grant can be appealed to a magistrates’ court within 21 days.

Consent conditions usually specify limits at named receptor positions, and those compliance measurements are made with calibrated Class 1 equipment at agreed boundary or facade locations. British Standard BS 5228 is the methodology that local authorities reference for predicting and controlling construction noise and vibration.

Continuous noise monitoring sits alongside that, not instead of it. A boundary sensor that alerts the site manager at 07:50 that levels are climbing is a way to avoid breaching a condition. It is not the evidence you submit to prove you did not.

How much does noise monitoring cost?

A LoRaWAN sound level sensor is around £42 excluding VAT. The gateway that collects it is the larger line item, from roughly £99 for a compact indoor unit up to £658 for an outdoor model, and one gateway serves a large number of sensors.

The realistic comparison is not sensor versus meter, it is sensor versus repeat consultancy. A single Class 1 boundary survey typically costs more than a handful of sensors and a gateway, and it tells you about one day. If your answer to “has it got worse?” is currently “we would have to book someone”, continuous noise monitoring is doing a different job for a different budget.

Delivery is a flat £14.25 across the UK, and orders are typically dispatched within 14 working days.

How do you deploy noise monitoring sensors without killing the battery?

Choose your time weighting deliberately, because it dominates battery life. This is the detail buried in the datasheet that catches people out.

On the WS302, at a ten minute reporting interval and 25°C, battery life is around 5 years at SF7 on EU868 with fast time weighting disabled, and around 3.3 years at SF10. Turn fast time weighting on and the same device gives around 1.7 years at SF7 and 1.5 years at SF10.

Two lessons follow. First, “more than 4 years” is only true in the slower configuration, so decide whether you actually need fast weighting before you quote a replacement cycle to anyone. Second, spreading factor matters: a sensor far from its gateway spends longer transmitting and pays for it in years.

Three practical placement rules. Mount at roughly working head height rather than ceiling level where you care about exposure. Keep the microphone away from hard reflective surfaces and out of direct airflow. Log a Class 2 or better meter next to each sensor for one shift at commissioning, so you know each unit’s offset in its actual position.

Frequently asked questions

Can I use a LoRaWAN sensor for Control of Noise at Work assessments?

Not as the assessment itself. With a ±3 dB tolerance it falls outside IEC 61672 Class 1 and Class 2, so it cannot produce the defensible personal exposure figure the regulations require. Use it to decide where and when to measure properly.

What is the difference between LAeq and LAmax?

LAeq is the equivalent continuous A-weighted level, an energy average over the measurement period, and it is what exposure calculations are built on. LAmax is the highest level reached. A quiet workshop with one loud press can have an acceptable LAeq and an alarming LAmax.

Should I use A weighting or C weighting?

A weighting for almost everything, because it approximates human hearing sensitivity and it is what the action values are expressed in. C weighting is the right choice when you are assessing peak or low frequency content, such as impulsive noise or plant rumble.

Do noise monitoring sensors record conversations?

Noise monitoring devices of this type report sound pressure levels as numbers, not audio. There is no recording and nothing to play back, which is what makes them straightforward to place in occupied areas. Confirm the behaviour of any specific model before deploying it where people work.

How many noise monitoring sensors do I need?

One per acoustically distinct area, not one per room. Two adjoining offices on the same side of the same plant room behave as one zone. A boundary facing housing is always its own zone.

The honest summary

Continuous noise monitoring is a screening and verification layer, not a compliance instrument. Bought with that expectation it is cheap, useful and hard to regret. Bought as a substitute for a calibrated survey it will let you down at exactly the wrong moment.

If you are assessing workplace exposure risk more broadly, our guides to real-time dust monitoring and COSHH and indoor air quality monitoring cover the same screening logic for airborne hazards. If the network side is new to you, start with what LoRaWAN is and where it fits.

The Milesight WS302 sound level sensor is in stock. Add it to a quote and we will come back within one working day.