Are lux meter apps accurate enough for a lighting survey?

What a phone lux meter app measures

A phone has no part built to measure illuminance. Lux meter apps make an estimate from hardware that was built for something else.

• The camera. Many iPhone lux meter apps work out a figure from the camera's exposure settings: how sensitive the sensor was set and how long it was exposed. The camera is built to take good photographs, not to measure light.
• The ambient light sensor. On Android, apps can also read the small sensor the phone uses to dim its screen. It's there to make the screen comfortable to read, and it isn't calibrated as a measuring instrument.

Neither one does two things that a survey meter is designed to do. A survey meter's sensor is filtered so that it responds to the colours of light roughly as the human eye does. It is also corrected so that light arriving at a shallow angle is counted properly. Without those corrections, the same room can read differently under LED and fluorescent lamps, or as you tilt the phone.

What independent tests found

Two published tests are worth knowing about. Both used phones that are now a few years old, and newer phones may do better. We haven't found an independent test of current iPhones.

DIAL, 2022

DIAL, the German company behind the DIALux lighting software, tested seven lux meter apps on several iPhones and Android phones against a reference reading. The worst reading was more than double the reference: 1,063 lux shown where the meter read 500. Two phones of the same model, running the same app, gave different results. The error also changed from one light level to another, so a single correction factor couldn't fix it. DIAL's conclusion was that serious measurement of illuminance needs professional hardware.

Odenwald, 2020

A study of phone sensors for citizen science, published in Citizen Science: Theory and Practice, compared phone light readings with calibrated values. Above 5,000 lux, readings came within about 12%. Between 100 and 3,000 lux, the typical error was about 16%, even after calibration. In low light, readings could jump in steps of more than 25%. None of the apps it looked at could export their readings to a file.

Why that matters for a workplace

The figures in HSE's guidance for workplaces run from 5 to 500 lux (see HSG38 lighting levels). That is well below the 5,000 lux where phones came closest, and the low end is where they did worst. An office reading 16% high could come out above HSE's figure when the room is actually below it.

Where a phone app is fine

• Comparing two spots in the same room: is this corner darker than that one?
• Spotting that a lamp has gone out or a fitting is dimmer than its neighbours.
• Rough checks for photography or houseplants.

What it can't do is produce a reading that you can hand to someone else, tied to an instrument and a certificate they can check.

What a survey needs from its instrument

Use a calibrated lux meter, and record enough about it that a reader can trace every number back to it:

• the make, model and serial number
• the calibration certificate: its reference, the date, and when the next calibration is due
• the light-source setting, on meters that have one, matched to the lamps on site

The full method is in How to do a lighting survey.

Do lux meters need calibrating?

Yes. A calibration certificate records who checked the meter, against what reference, when, and when it is next due. Follow the interval in the maker's manual, or your client's if theirs is shorter. If a certificate has lapsed by the day of the survey, say so on the record rather than leave it out.

Can I use my iPhone as a lux meter?

For a rough look, yes. For a survey, use the iPhone to record a calibrated meter's readings rather than to take them.

Using a calibrated meter with an iPhone or iPad

That's the job Luxwork does. It's a lighting survey app for iPhone and iPad, and it doesn't use the phone's light sensor or camera to measure light. Readings come over Bluetooth from a Chauvin Arnoux C.A 1110 or a Hioki FT3425, or you type them from any other calibrated meter. Each reading lands on the plan where you took it, carrying the meter it came from and that meter's calibration record. If the calibration had lapsed, the report says so.