Richard Hayes is seething about testing! What’s going on?
Let’s clear this up once and for all. Most of the explanations I see regarding the distinction between absolute and relative measurement methods contain some good information – but miss some important facts. To really understand this topic, you need to understand the different ways of gathering photometric quantities. Mind you, it does help if you’ve spent the last 35 years of your life actively engaged in photometry, but for those out there who haven’t, let’s try to simplify everything
This discussion has nothing to do with the presentation of the data; a properly formatted and measured absolute data file (i.e. an IES file) will produce exactly the same results as a properly formatted and measured relative file (i.e. EuLumdat or LDT file). Anybody who tells you that only an IES file can be used for LEDs is completely wrong . . . don’t ever talk to them again.
There seems to be one of these things floating around the industry at the moment, sometimes repeated by people who should know better. The statement says you can only measure an LED luminaire in absolute systems. This is totally incorrect, though I have even seen it repeated in a Standard under ‘definitions’.
There are two pieces of equipment most commonly used in photometric measurements of lamps and luminaires. The Goniophotometer and the Integrator. Contrary to popular belief both can be used to make absolute or relative measurements.
In the old days when we had lamps (remember them?) the luminaire was placed on a Gonio with the intensities collected in arbitrary units. Readings were then scaled by measuring the output of the lamps in an Integrator. This yielded a data set in cd/klm, so called relative photometry. No measurements that are made are in any particular units, but the calculations between the two measurements enable us to derive meaningful data.
The relative data file must always sum to 1000 lumens. This system had an advantage that it used a derivation of LOR (Light Output Ratio) so that efficiency comparisons could be made, and the data could be applied to different lamps of the same physical and electrical characteristics, but with different lumen output, when used in the same luminaire. Lamp manufacturers catalogue data for the lamp output could be applied to the data for lighting design purposes.
These days we still tend to put the luminaire on a Gonio, the only difference being that the intensities are collected in candelas. To do this, the Gonio system has to be calibrated. This was originally favoured for LED luminaires because LEDs could not be removed for a separate lamp comparison, relative-type, test. Hence no meaningful LOR, which will always be 1. The data would only be applicable to that luminaire, with those LEDs, on that day.
Calibration of a Gonio is usually done by reference measurements to a standard lamp. The standard lamp has a calibration in Candela and Lumens, data which is traceable to an accredited third party. These individual intensity readings are summed or integrated over the surface of a sphere to give an indirect measurement of Lumen output. That last part is very important, the Lumen output determination in the so-called absolute method is a derived function.
Let us suppose also that we can place the standard lamp used to calibrate our Gonio above into our Integrator; we can then calibrate the Integrator in Lumens. Then we can measure absolute Lumens.
But if I apply the relative method to an LED luminaire I can measure its intensity in arbitrary units, then use an Integrator to determine the lumen output and I can still produce a perfectly valid relative photometric data set. This has a big advantage. It may be that the same physical LED at the same drive parameters may have a different colour temperature and hence a different lumen output, so by using this data the measurement can be made applicable to all of the colour variants of that LED in that luminaire.
I hope you can realise that making these simple comparisons makes the distinction between absolute and relative photometry meaningless.
The questions we should really be asking is how does the testing process affect the integrity of the data produced.
There has been a move in the industry over recent years to say that LED luminaire should be measured under absolute conditions and that this means using a calibrated Goniophotometer. The current measurement standards are heavily biased towards the use of Gonio systems over integrators.
No measurement system is perfect, both Gonios and Integrators have problems which need to be addressed and understood. One is not better than the other; sometimes the measurement will bias towards one system, sometimes towards the other.
Dead spots all gonios have a support mechanism for the luminaire. At some angles this obscures the sensor. If you can’t collect all of the light then the sum (i.e. the lumen output) will be wrong.
Movement: all gonios move either the luminaire or the sensor. Both are temperature sensitive, so moving air has to be restricted or corrected for. This often restricts the speed of rotation of gonio systems.
Orientation: some popular gonio systems measure the luminaire at other than its intended mounting orientation. This will affect lumen output.
Temperature: all luminaires and lamps are temperature sensitive as far as light output is concerned. Gonios are big and may have large moving parts. Keeping the air temperature stable is very difficult. Changes in temperature during the test affect lumen output.
Stray light: there is no such thing as a zero reflective surface. Light from the test piece bounces around the gonio room, some reaches the sensor. If you collect some of the light directly and by reflection from surrounding surfaces, the sum is wrong and the Lumen output is in error
Absorption: the luminaire absorbs some light bouncing around the system. This needs to be corrected for.
Differential reflectance: Integrators depend on high even diffuse reflectance and any variation will affect the output. This can be particularly noticeable with very narrow beam type luminaires
Spectral error: the Integrator coating needs to be optically even in reflectance to all wavelengths that the system measures.
Signal output: Integrators attenuate the signal to the sensor by very large factors. This can be problematical.
Calibration: although Integrators can be calibrated, this is usually only applicable to one particular setup. Changing the baffles or supports can mean a re-calibration. Any Integrator system procedure should take this into account.
But Integrators do have one big advantage: they seal the luminaire within a chamber, therefore the temperature inside the chamber can be very accurately controlled – much more accurately than in a big gonio room. Would anyone like to bet that I can maintain a greater degree of temperature control in a sealed 3 m integrating sphere than in a 10m x 10m x 10 m Gonio room with large moving arms? (editor’s note: Richard has never been known to give a sucker an even break)
So what does all this mean; all of the sources of error above can be compensated for in a properly run laboratory. A lumen output derived from a properly set-up Gonio when compared to a properly run Integrator should agree to well within 2 %. If I had to measure a 6 degree LED spot I would favour a Gonio for determining Lumen output, but that Gonio had better be capable of discriminating 0.5° steps. If I want accurate Lumen output of a medium beam LED luminaire I would favour an Integrator test.
Ideally in both cases I would use both systems and compare the results, if they differed, I would look for unaccounted errors in the systems. One is not better than the other, absolute is not better that relative, the distinctions between the two methods are arbitrary and meaningless.