Sound level meter selection turns on three binding engineering gates: IEC 61672-1 Class 1 or Class 2 accuracy tolerance, A/C/Z frequency weighting, and a matched acoustic calibrator at 94 dB or 114 dB reference [S7]. Get any one wrong and the data is inadmissible for the standard you are trying to evidence.
The buying universe spans Class 1 laboratory-grade units from Brüel & Kjær, Cirrus, and 01dB, down to Type 2 industrial handhelds from NoiseMeters Inc — a North American distributor exporting to 108 countries — and Chinese OEM units quoted at US$18-70 per piece with a 10-piece MOQ on Made-in-China.com [S3][S4]. Open-source Python scripts such as the arupiot/Sound-level-meter project on GitHub (11 stars, 2 forks) only confirm that PC microphone-based SPL readouts lack the calibrator chain required by IEC 61672 [S1].
IEC 61672 Class 1 vs Class 2: tolerance, frequency, and what each is FOR
IEC 61672-1 divides integrating-averaging sound level meters into Class 1 (precision, ±0.7 dB reference-tone tolerance) and Class 2 (general-purpose, ±1.5 dB) over the 20 Hz to 20 kHz nominal range [S7]. Class 1 is the only choice for occupational noise dosimetry evidence under ISO 9612, environmental surveys under IEC 61672-3 field implementation, and any submission to a regulator that will pull the calibration certificate.
Class 2 is the right call when the work is comparative — machine-to-machine trending, factory walk-downs, HVAC balance checks — and where the consequence of a 1.5 dB error is a maintenance ticket, not a hearing-loss claim. A common mistake: buying a Class 2 meter to support a community-noise complaint, then discovering the local authority requires a Class 1 traceable measurement to GB 22337 or equivalent [S5]. The class you purchase must be written into the measurement plan on day one, not decided after the dispute.
Frequency weighting: A, C, Z and the 20 µPa reference
Every credible meter offers A, C, and Z (formerly linear) frequency weightings, with A-weighting approximating equal-loudness at 40 phon and the standard metric for occupational exposure under ISO 9612 and most community-noise ordinances [S7]. C-weighting stays flat across a wider band and is mandatory for peak sound pressure (LCpeak) measurements used in hearing-conservation programs where impulsive noise drives the dose.
The reference sound pressure is 20 µPa RMS — the 1 kHz threshold of human hearing — and 1 Pa equates to 94 dB SPL on that reference, a fact that explains why every acoustic calibrator field-checks at 94 dB or 114 dB at 1 kHz [S7]. Z-weighting (zero-weighting, i.e. flat 10 Hz–20 kHz) is the modern replacement for the older "Linear" or "Flat" mode and is the only weighting admissible when the post-processing will apply its own filter chain. A meter that only offers A and C cannot produce a Z-weighted spectrum file; do not assume all three are present on a sub-US$100 OEM unit [S4].
Time constants: Fast, Slow, and Impulse
IEC 61672 also defines the time-averaging constants: Fast (125 ms), Slow (1 s), and Impulse (35 ms rise, 1.5 s decay), and the meter must implement all three to claim compliance. Fast is the default for occupational noise exposure, Slow smooths out fluctuating sources like passing traffic, and Impulse is reserved for short-duration events such as drop-forge impacts or pyrotechnic discharges. [S3]
For construction-site boundary noise under China's GB 12523 framework, both the equivalent continuous A-weighted level LAeq over the measurement window and the maximum A-weighted level LAmax must be reported, which means the meter must log LAeq and LAmax simultaneously rather than only display an instantaneous dB reading [S5]. Brüel & Kjær's accessory catalog includes weather station kits, windscreens, and extension cables for exactly this kind of unattended long-term deployment, where 24-hour LAeq logging drives the compliance outcome [S2].
Microphone, preamp, and calibrator stack
The microphone capsule is the weak link: a 1/2-inch free-field pre-polarized microphone is the de-facto standard, with 1/4-inch capsules reserved for high-SPL work above 140 dB. Field calibration before and after every measurement session is non-negotiable — a 94 dB / 1 kHz acoustic calibrator must be listed in the case alongside the meter, and a Class 1 calibrator must be used to verify a Class 1 meter. [S1]
For outdoor environmental monitoring, Brüel & Kjær's all-weather microphone kit, windscreen, and desiccant-protected preamp assembly keep the capsule within IEC 61094-6 environmental limits for temperature, humidity, and static pressure, which is the same logic behind NoiseMeters' dedicated construction-site noise monitoring systems that download LAeq data over USB or cellular backhaul [S2][S3]. Skipping the outdoor kit because the budget is tight is the single most common way a Class 1 measurement campaign produces Class 2-quality data.
Data logging, dosimetry, and integration with monitoring systems
A handheld meter that only shows a real-time dB number is not a survey instrument; it is a spot-check tool. Real surveys need logged LAeq, Ln percentiles (L10, L50, L90, L95), LCpeak, and time-stamped 1-second or faster broadband records, with storage for at least 24 hours of continuous data at 1 Hz. Noise dosimeters take this further by shoulder-mounting on a worker and accumulating dose per ISO 9612 exchange-rate criteria (3 dB or 5 dB). [S3]
For unattended boundary monitoring — a hospital noise dashboard, a wind-farm complaint response, a quarry perimeter check — look for a noise monitor with a wall-mounted display, a noise warning sign output, and remote download, exactly the product lines marketed by NoiseMeters Inc and by Brüel & Kjær's environmental hire fleet [S2][S3]. The PC-based Python script approach (arupiot/Sound-level-meter on GitHub) cannot meet these logging requirements because it lacks IEC 61672 traceable calibration, a Class 1 capsule, and any documented time-base accuracy [S1].
Selection matrix: which meter for which duty
Match the IEC 61672 class and feature stack to the duty before you request a quote. Use this decision table as the first cut: [S1]
Occupational dosimetry (ISO 9612) → Class 1, A + C + Z weightings, Fast/Slow/Impulse, LCpeak logging, paired with a Class 1 calibrator [S7]. Community/environmental surveys (IEC 61672-3) → Class 1, A + C + Z, LAeq + Ln percentiles + LAmax logging, outdoor microphone kit [S2][S5]. Construction-site boundary (GB 12523) → Class 2 minimum, A + C, simultaneous LAeq and LAmax, weather protection, 24-hour logging [S3][S5]. Factory walk-down / machine trending → Class 2, A + C, Fast/Slow, no logging required, no calibrator needed daily. Music venue / club noise → Class 2, A + C + Z, LCpeak fast logging, noise warning sign output [S3]. Hospital / patient rest monitoring → Class 2, A-weighted LAeq data-logging, sign display integration, USB download [S3].
Where not to spend: do not buy a Class 1 meter for in-house machine-to-machine comparative checks; a calibrated Class 2 unit gives equivalent trending value at a fraction of the cost, and the US$18-70 per piece OEM tier on Made-in-China.com is a sensible shortlist for non-legal metrology [S4]. Conversely, do not specify a Class 2 OEM unit for any submission to a regulator that pulls calibration certificates; the cost saving disappears the first time the data is challenged.
Standardisation, calibration traceability, and certification stack
Every compliance-grade meter must ship with a traceable calibration certificate from an ISO/IEC 17025-accredited lab, with the uncertainty stated in dB and the reference standards (typically a reference sound source traceable to a national metrology institute) listed by serial number. The certificate is valid for 12 months on most vendors' recommended cycles; high-use or field-deployable units often run on a 6-month cycle. [S1]
Standards you will see in the certificate and on the data sheet: IEC 61672-1 (electroacoustic performance), IEC 61672-2 (pattern evaluation), IEC 61672-3 (field implementation), IEC 61094 (microphone specifications), ISO 9612 (occupational exposure), and — for community noise in China — GB 22337 and the construction-site boundary standard GB 12523, both of which define daytime (06:00–22:00) and night-time (22:00–06:00) windows and require maximum A-weighted level reporting alongside LAeq [S5]. A 1 Pa sound pressure reading corresponds to 94 dB SPL on the 20 µPa reference, and any calibrator that does not state its reference pressure and tolerance should be rejected on sight [S7].
Two trackable signals for the next quarter: first, watch whether IEC 61672-3 amendment cycles tighten the outdoor weather-correction tables — the field-implementation standard is the one most often cited in disputes and the most likely to be revised. Second, monitor Chinese OEM product listings on Made-in-China.com for new Class 1 units entering the US$200-400 band; the price floor on credible Class 1 hardware is the single most important commercial variable in this market [S4]. For related specification logic on other field instruments, see the anemometer selection criteria guide, and for a deeper view of cost-stack trade-offs in measurement hardware, the anemometer price and cost guide 2026 is a useful parallel. The core sound level meter encyclopedia entry consolidates the IEC 61672, ISO 9612, and microphone-class definitions used throughout this article.
Spec-level background on the components involved: radar level meter, and tdr level meter.