Expose the global conspiracy or how to measure the luminous flux of LEDs on the knee

All of you have probably heard about the global conspiracy . Masons, aliens and Jews The manufacturers of light bulbs entered it a hundred years ago so that the light bulbs would not last forever, but would burn out every month and eat a lot of electricity. And only now the conspiracy is shattered and bulb magnates are crushed by the great empire of China, which has flooded the whole world with eternal and economical LED lamps. But do not relax - the world plot does not surrender. Now he came in the form of the Great Led Lazhi Lodge Lies In short, everyone lies (s).

Jokes, jokes, and to some extent lie, probably, all manufacturers of LED-lighting. Someone brazenly and frankly, someone so slightly moans - but somehow, it seems, there is not a single company that would not overestimate the parameters of their products. In many ways - someone just writes beautiful numbers from the lantern, sometimes transcendent from the point of view of common sense. And someone - just writes the characteristics are quite truthful, but obtained in conditions far from the real conditions of operation. For example, the luminous flux measured at a temperature of 25 ° C in a pulsed mode. One way or another, a 15-20% "allowance for lies" will have to be given.

Illumination is easy to measure, luminous flux is difficult and expensive. It is necessary to collect all the light emitted by the lamp and equally take into account the rays in all directions. That is, we need a photodetector in the form of a hollow sphere with the same photosensitivity of each part of its surface. The manufacture of such a photometric sphere and its subsequent calibration is a very difficult task.

Another approach is to measure the radiation pattern of the light source point by point and integrate it over the entire sphere. But it is not easy: you must have a solid dark room with dark walls. And a goniometric head with two axes is needed, preferably with an automatic drive, in order not to get rid of the angles manually for each of several hundred points.

However, there are a couple of particular cases that are often found in practice and for which it can be limited to one dimension. I want to tell one of them to the habrasoobshchestvo.

This particular case is a flat cosine emitter. Cosine is such a radiator, the brightness of which does not depend on the angle between the normal to its surface and the direction of the observer. The radiation pattern of such a radiator is determined solely by the geometry - namely, the apparent surface area. And for a flat cosine emitter, there is a simple relationship between the luminous flux and luminous intensity in the direction normal to the plane:

.

That is, it is enough to measure the illumination in a meter from the light source with a luxmeter and multiply it by 3.14 - and we already have the luminous flux value (or, if the distance is not equal to a meter, it will have to be taken into account according to the inverse square law). Of course, the light source must be much less than the distance to the luxmeter - otherwise the inverse square law will not work and the measurement result will be overestimated.

What light sources can be considered as sufficient cosine emitters with sufficient accuracy for practice? These are practically any white lighting LEDs without a lens and flat assemblies based on them. All kinds of Chinese 5730, 2835, 5050, 3030 and others, which are usually found in LED lamps with aliexpress, and are also sold there separately in coils for pennies - this is it. And also matrices. And the Chinese square at 10 watts, and Cree CXA and CXB. But for any LEDs with a lens, as well as for LEDs without phosphor (for example, RGB), this method is not suitable - their radiation pattern is significantly different from cosine. Flat luminaires, recessed into the ceiling and covered with milk glass, also correspond well to this model.

So let's measure something already. As guinea pigs we have today:

1. Chinese assembly for 90 watts of 156 5730 LEDs (each with two 13x30 mil crystals) with a built-in driver for the CYT3000B. According to the assurances of the Chinese, should give 9200 lm.

The power consumption of the instruments - 85 W, and stay on it .

2. Matrix CXA2530, new version, 3000 kelvin, Ra> 80. The luminous flux at 800 mA and 85 ° C according to the datasheet is at least 3440 lm, and at 25 ° C (there is no such temperature, unless the LED itself gets cold to a temperature below zero, thermal resistance will not give) at least 4150 lm.

We start at a current of 800 mA, the power consumption was 28.64 watts .

3. HPR20D-19K20 - as old as a mammoth (bought in 2010, if not earlier), the 20-watt matrix from HueyJann, similar to the current 10-watt matrix, differs from them in a large number of crystals under the phosphor - there are 16 of them instead of nine ( 4 pieces in series in each of the four chains connected in parallel). Declared 1830 lm at a current of 1.7 A, it's really not brighter by eye than the CXA2011 with a power input of 11 watts.

We run on the passport current of 1.7 A, the voltage was 12.2 V, power 20.74 watts .

Illumination is measured with a UT382 (Uni-T) light meter, on the “eye” of which we put on a black paper hood so as not to catch the light reflected from the walls in an unprepared room. Distance in all cases - meter. The results in the table.

It turns out that the luminous flux of the Chinese assembly corresponds to the stated (within the error of the luxmeter), the Cree's matrix also has everything within the datasheet (considering that its temperature is unknown), but the HueyJann matrix of the promised lumens is not close.

But something dimly worried me: 9000 with a tail lumen at 85 watts, considering the driver efficiency of 80% and despite the fact that the LEDs are far from lightweight, half a watt per case, and the peak current is twice the average (no filter capacitor these boards do not) - this is very cool. In addition, it is somehow not visible from this assembly of a much greater illumination in the room compared to a chandelier, in which there are five light bulbs of 950 lm each (energy saving).

Suspicion falls on the luxmeter - not all of them adequately measure LED sources. Those made on the basis of the BPW21R photodiode have a very approximate correspondence of the spectral sensitivity to the standard visibility curve, and the relative sensitivity to radiation of 450 nm (this is the wavelength corresponding to the blue peak present in the spectrum of almost all white LEDs) exceeds the relative sensitivity of the eye in this area several times. In this device, the photodetector is different, which was one of the criteria when choosing an instrument, but still we go to labor protection and take another luxmeter. It turned out to be TKA-Lux. In his method of verification contains a check of the spectral characteristics, that is, it must comply with the visibility curve with a standardized error. We repeat the measurements with him. Here are the results:

Well what can I say? Not only manufacturers of LED lamps are lying, but also my luxmeter. And lying, as expected, in different ways for different LEDs. For the CXA2530 matrix, the difference with the professional device is minimal, rather within the error limits of both instruments. But this matrix has an almost imperceptible dip in the spectrum, if you look through the CD (actually, of course, it does exist). But the rest of the experimental "failed" decently. And now it is perfectly clear that they do not reach the declared lumens more than noticeably: the Chinese 90-watt assembly - by 25%, and the HPR20D-19K20 matrix - almost doubled.

From this we can draw the following conclusions :

1. Yes, in the manner described it is possible to estimate the luminous flux emitted by the LEDs, arrays and assemblies (within the particular case described).
2. With the measurement of illumination from LEDs with a light meter, one must be careful and make sure that it has the correct spectral sensitivity curve. For all lie (s).
3. If the measurements show that the Chinese product achieved the stated characteristics, then it is quite likely that the device is graduated in Chinese suites :).

If you want in the same way to evaluate the luminous flux of the LED bulb with a hemispherical lens, you need to remove the lens. Beneath it are likely to be quite suitable LEDs. But the lens itself makes a loss of 15-20 percent or more of the light flux.

Yes, and the last. The described method is in no way metrologically rigorous or accurate. It is estimated and nothing more. That is why I did not give an error analysis here.