Noise Sensor Calibration: Field Checks Without a Lab
Share
- Microphone sensitivity drifts slowly downward, so a sensor that reads low makes a loud site look compliant — the worst direction to drift.
- A 94 dB / 1 kHz acoustic calibrator check takes a minute: within about half a decibel is healthy, 1–2 dB is a warning, and more than that means lab calibration or replacement.
- Read the check at the dashboard or controller, not just at the sensor — most "calibration" problems are actually scaling errors in the chain.
- Check the windscreen first when readings come back low at high frequencies — a hardened, clogged windscreen is cheaper to replace than a microphone.
- Never hide drift by typing an offset into the controller; record the error and take the sensor out of service if it is outside tolerance.
Table of contents
A noise sensor that has never been checked is a guess with a cable. Microphone sensitivity drifts — slowly, silently, and in one direction — and a sensor that read true when it was installed can be several decibels off a year later without anything visibly wrong. You do not need a calibration laboratory to catch that drift. A few field checks, done with care and recorded, are enough to know whether the readings can still be trusted, and which sensor needs to go back for a proper calibration.
Why microphones drift
The sensing element in a field noise microphone is a thin diaphragm, and it lives outdoors. Humidity, dust, and temperature cycling all change its stiffness a little at a time, and the electronics behind it drift too. The result is usually a slow loss of sensitivity: the sensor reports lower than reality, which is the worst direction to drift, because it makes a loud site look compliant. A sudden jump is easier to spot — a jump means damage or water ingress — but slow drift only shows up if you check.
Windscreens add their own drift. A foam windscreen that has spent a year in sun and rain hardens, clogs with dust, and absorbs high frequencies. The sensor is fine; the cover is not. Replacing a windscreen is cheap, and it is the first thing to try when a field check comes back low at the high end but near-normal at low levels.
The field calibrator check
The standard field tool is an acoustic calibrator: a small device that fits over the microphone and produces a known tone, usually 94 dB at 1 kHz. The procedure takes a minute. Fit the calibrator, note the reading, remove it, and repeat. A sensor within about half a decibel of the calibrator's stated level is healthy; one to two decibels off is a warning to log and watch; more than that, and the sensor should be taken out of service for a laboratory calibration or replacement. The key detail is to record the numbers — a check with no record is a ritual, not a measurement.
If the sensor's output is a Modbus register or a 4–20 mA loop rather than a display, the check is the same, just read from the other end: apply the calibrator, read the value at the controller or the dashboard, and compare it against what the sensor should report. This also tests the whole chain — sensor, wiring, scaling, and logging — which is where most "calibration" problems actually live. A sensor that is perfect at the microphone and wrong at the dashboard has a scaling error, not a microphone problem.
Checks without a calibrator
Not every site owns a calibrator, and there are still useful checks. The most reliable is a side-by-side comparison: put a handheld sound level meter next to the fixed sensor and read both at the same time, in the same sound field, over the same averaging window. A steady source helps — a running fan, a machine at constant load — because the two instruments must hear the same sound. Agreement within a decibel or two is good; a consistent offset in one direction is a calibration problem, whichever instrument is at fault.
A second check costs nothing: compare the sensor against itself over time. Pull the long-term trend and look at the quiet periods — night-time background, shutdown days, weekends. If the floor of the readings has crept down over months while nothing on site changed, suspect drift. This check cannot tell you the absolute error, but it can tell you something changed, which is usually the question that matters.
A third check is a known sound source used as a rough reference. It will not give you a laboratory number, but a tone or a machine whose level you recorded when the sensor was new, replayed in the same position, will show whether the sensor still agrees with its own history. Keep the method identical each time — same distance, same source, same settings — or the comparison is meaningless.
What a field check cannot do
Field checks catch drift and damage; they do not replace laboratory calibration. A single-frequency check at 1 kHz says nothing about the sensor's response at low or high frequencies, and it cannot verify the A-weighting curve or the averaging behaviour. For compliance reporting, a laboratory certificate on a stated interval is still required — typically annual for most environmental monitoring, more often in harsh conditions. The field check is what keeps you honest between those certificates.
Also, never "calibrate" a sensor by typing an offset into the controller to make it match a one-off comparison. That hides drift instead of fixing it. An offset applied to a drifting sensor is a number chosen to make today look right, and it will be wrong again next month. Record the error, and if it is outside tolerance, take the sensor out of service.
How often, and what to record
For a fixed outdoor noise sensor, a field check every three to six months is a sensible rhythm; monthly in dusty, coastal, or high-humidity sites. Check after any event that could damage the microphone — a storm, a wash-down, a bird strike, a nearby lightning strike — and after any maintenance that touched the sensor. Each check gets a dated record: calibrator level, sensor reading, difference, windscreen condition, and who did it. A year of those records is the audit trail that makes the data defensible.
Field calibration checklist
- Check with a 94 dB / 1 kHz acoustic calibrator, or side-by-side with a handheld meter on a steady source.
- Read the value at the dashboard or controller, not just at the sensor — test the whole chain.
- Replace a hardened or clogged windscreen before blaming the microphone.
- Record the calibrator level, the sensor reading, and the difference, with a date.
- Treat a consistent offset as a calibration problem, not a reason to type in a fudge factor.
- Check after storms, wash-downs, and any maintenance on the sensor.
- Keep the laboratory calibration on schedule — field checks fill the gaps, they do not replace it.
An RS485 noise sensor with a 30–120 dB range that is checked on a schedule and checked honestly gives readings you can stand behind. The calibrator is cheap, the check takes a minute, and the record is what turns a number into evidence.
Our engineers can help you pick the right model, check compatibility, and get your installation right the first time.
RS485 Noise Sensor (30~120dB Range)View product →
Related Blogs
The Site Got a Noise Complaint but Your Sensor Showed Nothing: Compliance Monitoring on Construction Sites
Your Rain Gauge Under-Reads After a Storm: The Drip Test and Adjusting Screws
Two Humidity Sensors Disagree by 10%: Calibrating with Salt Solutions and Reference Checks
Your UV Readings Keep Creeping Down Year After Year: Sensor Ageing and Drift