light source caused by the alternating portion of the
electric current flowing through the light emitting element of the illuminant.
This 'illuminant' mentioned in the further course can in principle be any light source, so also computer monitors, televisions, etc.
Light flicker (in fact optical light flicker) may also be referred to as Tempororal Light Artifact (TLA)
or light modulation.
Light flicker is not to be confused with irregular "flaring" of low frequencies, as with wax candles.
In contrast to the incandescent light, which has the temperature rise / fall as an inertial element, the emitted light of the LED follows precisely the course of the current with which it is supplied.
The so-called electronic ballast is responsible for this, whose quality determines how well the lamps are regarding the flicker
degree. As an alternative to direct current, the LED can be pulsed with high frequency (> 30kHz non-perceptible) alternating current. If there is almost no electronic circuitry, the lamps
usually flicker with twice the mains frequency (like neon bulbs). A sufficient circuitry costs however money, which is practically forbidden by itself with China-cheap products.
In short: Depending on the design of the electronic ballast, the illuminants can flicker arbitrarily.
However, since there is no information about the flicker degree on the packaging or in the technical data of luminaires, both vendors and customers are kept in the dark. Only the manufacturer
really knows what he is doing and what he offers.
It is necessary to provide a customer and package friendly single value declarated as a percentage (or no unit), so that a simple purchase decision can be made.
The flicker degree must cover at least the following characteristics:
For comparison (and hardly anyone knows):
Even ordinary incadescent light bulbs flicker to a certain degree, because they are operated against Edison's idea with alternating voltage 230V/50Hz or 120V/60Hz. The CFD is between 5% and 10% (here 5.9%). We are accustomed to this little flickering, it does not harm according to the experience and thus provides the reference for LED retrofits.
The flicker index is in the case of pretty bad light sources (here a bad LED filament bulb)
above 0.5. This then makes it clear that these have a strong stroboscopic effect:
In the operating state they are partially longer pitch dark than bright alternating with twice the mains frequency.
Here the light emission example of a so-called driverless LED light whose flicker index is less than 0.3 but the CFD above 50%. There is also stroboscopic flickering. Most dimmable is only a part of this curve, so much worse than this picture.
After 'enjoying' stroboscopic light for hours, besides headache and other complaints feeling nauseous is possible, in the worst case, epilepsy seizures might occur.
Particularly noticeable is the perception of moving or rotating ojects. A simple hand movement is no longer perceived correctly as described here, but the hand appears several times in other places. Any head movement can lead to impairment of the orientation up to dizziness. LED filament lamps and small G9 lamps are particularly affected. Never since the invention of the incandescent lamp has there been so poor quality with respect to light flicker. The reviews in online shops also reflect the discrepancy between flickering and low-flicker illuminants.
By September, 20th 2015 Der Lichtpeter measured for the first time the highest possibility of light flickering: needle pulses with a frightful CFD of 124%! Only usable for stroboscopic effects in the discotheque, in a flash you have 20 fingers on one hand.
Der Lichtpeter has researched and unfortunately there is no measuring device that meets the above criteria and immediately indicates the degree of light flicker. There are some devices that determine the so-called percent flicker and flicker index according to the somewhat established method of the Illuminating Engineering Society IES: RP-16-10.
However, this method has decisive disadvantages:
- The basic frequency of the flicker is not taken into account.
- Two values ('Percent Flicker' and 'Flicker Index') are given, which only taken together make a good/bad decision. But there is not even an approach on how.
This is not manageable for the user and is not suitable for a technical specification.
A literature search from the middle of June 2015 to the end of August 2015 is completed with a critical assessment and a resulting measurement suggestion. It provides a reasonable measurement method resulting in a single measured value with (traffic light) rating, which can be declared in the technical data and finally on the packaging. Since October 2015 the world's first sensible measuring method to measure light flicker is available:
I carry out measurements for vendors and consumers. For this purpose, I have developed a processor-controlled semiautomatic machine, which performs all settings e.g. dimming.
Here is shown briefly, which measurement methods were known by the literature review:
A) Measurement methods (signal processing & measurement)
For measuring the light signal a special electronic circuit designed using a V(λ) photo diode directly connected to an adjustable transimpedance amplifier, which converts the photo current into a voltage. Subsequent is an anti-aliasing filter (low-pass filter) with a cutoff frequency suitable for sampling. The output signal is measured by a DSO. Depending on the calculation method, the signal is sampled at 2 kHz to 1 MHz. The duration should be at least 1 second.
The System by Der Lichtpeter scans at 500 kHz for 1s, which results in 500,000 measured values.
There are various methods for calculating the flicker from the measured signal. Most are time-based methods that do not take into account the frequency spectrum, and thus the waveform and the fundamental frequency.
B1) Time-based calculation methods
B1.1) Contrast methods using minimum, maximum and RMS values:
The companies Admesy B.V. and CHROMA ATE INC describe a method which relates the alternating portion (Max-Min or RMS) of the light emission to the DC proportion (mean value or ½Max + ½Min).
According to his website Dipl.-Ing. Dominik Schuierer determines the light flicker degree by calculating Vpp/Vrms.
B1.2) according to IES: RP-16-10:
B1.2.1) Flickering in percent is calculated as: (maximum - minimum) / (maximum + minimum).
B1.2.2) The flicker index is calculated from the curve: (area above the mean value) / total area.
B2) Frequency-based calculation methods
A common feature of these methods is that they undergo a Fourier analysis and then carry out further calculations.
B2.2) JEITA and VESA method
The companies Admesy B.V. and CHROMA ATE INC apply a frequency-dependent evaluation by setting all frequencies ≥ 65 Hz to zero. Thus, waveforms and the main problem of the flicker with twice the mains frequency are disregarded.
B2.3) IEEE 1789
According to IEEE 1789, there are frequency-dependent thresholds for the classification of a light source in RP1="recommended" and RP2="no observable effect level".
RP1 is fulfilled when the modulation below 90 Hz is less than 0.025×f as well as between 90 Hz and 1250 Hz less than 0.08×f.
RP2 is fulfilled when the modulation below 90 Hz is less than 0.01×f as well as between 90 Hz and 3000 Hz less than 0.0333×f.
Physiologically implausible is the leap by a factor of 3.3 at 90 Hz.
To IEEE 1789 it also makes no difference whether the light is spikelike, sinusoidal or rectangular, but in practice it does.
B2.4) Calculation according to LRC
The ASSIST group of the Lighting Research Center evaluates the frequency components obtained according to a characteristic curve which is based on the flicker fusion threshold.
A so-called "metric value" results from the Pythagorean Sum of the evaluated frequency components, which in turn is subjected to a further evaluation formula from which the probability of detection of the flicker is expressed as a percentage. In this case, too, the clumsy choice of the upwardly limited flicker fusion frequency of 70 Hz does not take account of the problem of flickering with twice the mains frequency.
Since mid-April 2016, the UL has issued a low flicker mark.
In principle according to this method, all light emissions are regarded as low flickering which are either above 200 Hz or which components below 200 Hz have a modulation (% flicker) of less than 30%. However, it happens, for example, that low-frequency emissions of approx. 30 Hz, 15%, which are clearly perceived as strong flicker by anyone, are regarded as low-flicker. In my survey service, however, I can presage whether the UL would award the certificate.
B2.6) The calculation according to the International Commission on Illumination CIE, lighting department, practically explained in a webinar: The CIE created the term "Temporal Light Artifacts" (TLA), which is called "PstLM" in the frequency range up to 50 Hz and "SVM" in the frequency range from 80 Hz to 2 KHz.
Here the range of 10 Hz to 2 kHz is covered, but the disadvantages are the non-closed gap between 50 Hz and 80 Hz as well as the problem of the necessary communication of two values in order to make a single assessment. Both values must be less than 1.0 to classify the light emitted as imperceptible. The Commissioners Phillips, Osram, SEU (CN), Chroma (TW), PNNL (US), ETSIB (ES) are pushing this method, which is why it could become a standard despite the disadvantages.
At a working meeting of the International Electrotechnical Commission in April 2018, Der Lichtpeter made a proposal on how to combine the two values PstLM and SVM into one (TLA), which was considered technically correct.
B2.7) Der Lichtpeter: CFD
The flicker frequency can be precisely calculated with the frequency-based CFDFB , which
Der Lichtpeter gives in all ratings and measurement reports. It can be determined with a powerful computer and specially developed software. The computation of the CFD is given its own web page for its complexity.
Here is only a brief summary:
- The measured values are decomposed by a discrete or a Fast Fourier transformation.
- The complex frequency components are converted into frequency-amplitude values.
- The amplitude values are normalized to the DC component.
- The normalized frequencies are weighted with respect to the human perception with a special polynomial function.
- The RMS is calculated from the weighted frequency amplitudes, the result is converted as a percentage.
- The final result is the CFDFB, which is generally communicated with CFD.
- The CFD is a single percentage value that takes into account all frequencies from 10 Hz to 250 kHz.
1) For pure constant current, the compact flickering degree is 0.
2) For a 100 Hz sinewave, the following can be assumed: CFD ≈ Modulation / 2.
3) For a 100 Hz rectangle in the duty cycle of 50% / 50% the CFD is approx. 70%.
4) In the case of non-sinusoidal waveforms (for example, needle pulses), it is >100%
5) It responds to the waveform and stroboscopic effects.
6) It takes into account the flicker fundamental frequency, which at 100 Hz is neutrally approx. 1
7) The calculation of frequencies up to 20 kHz also takes into account the phantom array effect.
8) The CFD results in a single user-friendly and easy-to-handle confidential unit. So it goes well into the technical data and is great on the packaging.
Only the calculation method according to CAT24JA10 (see above) comes in its quality somewhat closer to the CFD.
On over 700 measurements of the over
500 measured light sources it can be seen:
The correlation between CFD at about
10% = "good" (low flicker) and
UL "low optical flicker" is over 95%.
This is also due to the fact that most of the illuminants flicker at about 100 Hz,
which both methods at this point similarly take into account.
It is thus shown that the CFD is excellent as a measuring method for optical ligh flicker compared to the CAT24JA10 method. This is because the CFD provides a true reading compared to a simple "low flicker pass / fail".
The CFD has been declared in the technical data for the first time by illuminant manufacturers, even before an indication of light flicker was given by any other manufacturer / vendor.
The Compact Flicker Degree allows the customer to easily judge the light source. On this web page for the CFD calculation, the CFD is graded in 5 categories according to the traffic light system.
According to this CFD algorithm, Der Lichtpeter also tested some artificial signals (e.g. those from these documents) and found that all values are mathematically plausible. Correspondingly, the over 500 measured illuminants also provide plausible values and match the indicated behavior of the 5 categories.
You can use the quick tests to get a first-aid overview of whether the light is flickering too much.
There was a time of no information on color temperature in Kelvin or the color rendering index in Ra for LED lamps. If one of the values is missing today, then probably hardly anyone will buy the
This should also be so for light flicker, each manufacturer should specify it in the technical data.
If the CFD is low, then this is a good sales argument. If it is not available, it can be interpreted as an inferior quality.
One should ask his vendor for the amount of light flicker or if cannot provide it: Do not buy.