Photometers work by placing a fi lter over the top of a silicon detector that turns the wavelength response of the detector to one that mimics the way the eye responds to light (the photopic response curve). The photometer turns the incident watts of light into lumens, which is important for lighting applications.
When there are differences between the response of the photometer and that of the photopic response curve, the lumen figure may be wrong.
Most cheap luxmeters have filters with strong deviations from the photopic response curve (see figure), which results in large errors for all spectra, except the pre-calibrated tungsten, and have especially large errors for devices that do not have smooth output spectra – such as fluorescent and LED light sources – that tend to have output concentrated around the region where the errors are largest.
Of course, many of the new energy-efficient light sources do not have smooth spectral profiles and they present the greatest problems because the lighting industry tends to depend on the older photometer-based luxmeter technology.
Even the best photometers such as the ones designed by the UK National Physical Laboratory which helps set the standards for lighting around the world, have errors. These errors tend to occur largely in the blue region, which is unfortunately where LEDs emit most of their light (see figure).
If the shape changes, the effect of the error also changes and a new calibration will be necessary.
Most luxmeters are factory calibrated against a tungsten source at a warm colour temperature of about 2900K. A tungsten source has a relatively smooth output with changing wavelength, as you can see below. The calibration essentially corrects for the error in thefilter response and is known as the spectral correction factor. Once corrected, the luxmeter will work well for tungsten light with correlated colour temperature around 2900K.
Another issue is the error in measurement of the incident lux – lumens falling onto the detector divided by the detector area, lm/m2 – at different incident lux levels.
For instance, when assessing installations or products ranging from 10 to 10,000lx, the user will assume that the luxmeter has been calibrated across that wide range of lux levels. However, many cheaper luxmeters have significant errors at different lux levels as a result of the non-linear response of the detector and the electronic amplification circuitry, which may not have been corrected.
Such devices can be altered by calibration, however, these calibration companies cannot alter the inherent nature of the photometer, which requires spectral correction for the spectrum of each light source measured.
So, what’s the solution? Well, it’s definitely worth spending a little more money on a reasonably priced luxmeter. Having luxmeters calibrated for the type of light spectrum that is generally used in your installations will improve accuracy.
If you have more budged and need to work with many different light sources, a spectrometer-based (or spectrophotometer/sprectroradiometer-based) system is the best. This will give accurate measurements of any light source and you can also measure light quality (CRI).
Errors as high as 40 per cent can be seen between cheap luxmeters and proper spectrally resolved measurements will become more important as on-task lux levels are more tightly controlled and energy loading in Buildings Regulations becomes more common.
When buying a luxmeter, ask to see the filter response function across the range of incident lux levels. If the supplier cannot provide this information then maybe it is worth considering another supplier.
When dealing with a test lab, ask if it used spectral correction for any of its photometric measurements. When having photometric measurements done, it is important that systems can be adjusted for the differing spectra of light sources.
Q: Red, green and blue LED datasheets have the optical output described in milliwatts. How can this be converted into lumens?
Again, the answer lies in the photopic response curve. A single spike at a particular wavelength can be converted easily – here at the green wavelength (555nm) 1W of light is worth 683lm, 1W of red (625nm) is worth 247lm and 1W of blue light (460nm) is worth a measly 50lm. The table also shows typical values at some other wavelengths. There is a large spread of values in the efficacy figures.
However, only laser sources tend to have very narrow spikey spectra and LEDs tend to have broader spectra. The figure shows a blue LED superimposed on the photopic response curve.
In this case, it is the overlap between the photopic response curve and the W/nm values from the LED datasheet spectra, that are used to convert form watts to lumens. We do this by converting the W/nm figure to lm/nm at each individual wavelength and then integrating all the values to produce a total lumen value, which is the luminous flux output.
Next week we will address some questions about the information in photometric reports and how they have been derived from the measurement labs.