Flicker and strobe effects became a historical chapter of lighting technology after the advent of electronic ballasts for fluorescent lamps. However, light-emitting diodes, with their steep voltammogram characteristics and rapid response of luminous flux to changes in current, have brought these phenomena back to the fore. In this part, the effect of typical power supply and dimming methods of light-emitting diodes on the time course of their luminous flux will be described.

DC power supply

This trivial method of powering LEDs is more inconvenient in terms of temporal artefacts, as the luminous flux produced does not change over time. The LED or module is connected using a resistor or current regulator to a voltage source. Many types of adhesive LED strips are designed to be powered by a constant voltage, typically 12 or 24 V. Series combinations of several LEDs are connected to the power bus via a resistor, or in higher-end strips, a transistor current regulator with thermal protection.

A similar method of power supply is also used for some LED bulb replacements for mains voltage. Here, a linear regulator supplies a series connection of approximately 100 diodes from a rectified and smoothed mains voltage. The thermal losses on the stabilizer are acceptable at power inputs up to about 10 W. Besides zero crosstalk, very little electromagnetic interference is an advantage.

DC dimming

CCR (Constant current reduction, also Amplitude dimming) is the dimming of LEDs using linear current control, see Figure 1. This method of power supply also produces no flicker. However, as the current decreases, the spectrum of the emitted light changes slightly and the differences between individual diodes increase. A potential weakness is the instability of the control at low intensities below 3% of maximum. This can be addressed by pulsing the output (Hybrid Dimming [1]), or in combination with output filtering (Filtered Dimming [1]), see Figure 2. In the linear region, again the advantage is the low radiated electromagnetic interference, which allows the use of longer wires between the LED and the power component.

@obr1@
Fig. 1. Linear current control

@obr2@
Fig. 2. Different control methods at low intensity

Pulse Width Modulation Dimming

PWM (Pulse Width Modulation) represents in lighting technology a periodic interruption of the light flux with sufficient frequency to apply Talbot's law, i.e. that a person perceives only the mean value of the light quantity. The mean value can easily be changed by changing the alternation. Simple dimmers or multichannel colour mixing devices can be constructed on this principle. The advantage is very low losses, because the switching transistors are in either the on state (minimum voltage and full current) or the open state (full voltage and no current). Losses are mainly incurred during transitions between these states, leading manufacturers to try to use lower frequencies (a few hundred hertz). A potential weakness is again the low intensities, where the light is emitted in the form of short intense flashes. The light pulses are detected by the retina and transmitted to the visual cortex of the brain up to a frequency of about 4 kHz. For some power supplies, the PWM frequency is therefore increased to 8 kHz at the lowest intensities. The disadvantage of PWM is the steepness of the edges of the current waveform, which carries a large proportion of harmonic components and can cause large electromagnetic interference. The solution is output filtering and shortening the length of the wires.

PWM dimming modification

For multi-channel power supplies, a PWM with a higher frequency is sometimes used for the warm light channel, see Figure 3. According to the Kruithof curve, it can be expected to be used more often at lower intensities. The flicker is more noticeable in the red region of the spectrum when viewed directly and less noticeable with increasing frequency.

@obr3@
Fig. 3. Dual-channel PWM dimming with different frequencies (Helvar LL60/2-E-DA-iC)

Power supply by switching inverter

Switching converters in the BUCK (decreasing) or BOOST (increasing) topology are often used to excite LEDs e.g. in table lamps powered from a mains adapter. The output current has a triangular waveform with a large DC component. Ripple at operating frequencies in the tens to hundreds of kilohertz produces no visible timing artifacts, see Figure 4. Although these circuits consist of only a few components, the design is extremely sensitive to the arrangement of the components and their connections. Thus, an improperly designed circuit of a few watts can be a strong source of electromagnetic interference. The inverters in the most widely used mains-voltage LED bulb replacements operate on the same principle; the typical operating frequency of several of the E27 socket types tested was 100 kHz and that of the miniaturised E14 socket version was 200 kHz.

@obr4@
Fig. 4. LED current flowing when powered by switching converter

Dimming of switching converter

The integrated circuits for these converters usually include a PWM dimming input (see relevant paragraph). In addition, some allow dimming with a DC voltage, which can be obtained by low-pass filtering the PWM from the control circuit to prevent mixing. Unsmoothed rectified current supply

Two-way rectified voltage and biasing resistors are often used in line voltage tapes or Christmas chains. In some cases, the filter capacitor behind the rectifier is missing, has low capacitance or has lost its original characteristics through operation. The diode light then pulses at twice the mains frequency, which may not be visible in direct view, but the flicker is noticeable in the peripheral field, see Figure 5.

@obr5@
Fig. 5. LED current flowing when powered by un-damped DC voltage

Unsmoothed rectified current supply with section switching

Ever since the beginning of LED lighting, manufacturers have been trying to develop a power supply circuit that would do without electrolytic capacitors (which are the main source of failures and limited lifetime of electronics) and, if possible, without inductances. Imagine a string of 100 LEDs with Vf = 3.2 V connected in series. This can be connected to a rectified mains voltage (umax = 325 V) using a small resistor. At the peaks of the rectified voltage, the diodes will light up to their full power. As the voltage drops, however, they will fade rapidly. If some of the diodes were bridged with decreasing voltage, the remaining diodes would glow much more. This bridging of sections can be repeated several times to extend the portion of the period for which at least some of the diodes are lit. The culmination of efforts in this area are probably the Seoul Acrich circuits [2], which connect directly after the rectifier bridge and need neither an electrolytic capacitor nor inductance to operate; see Fig. 6 for the luminous flux waveform.

@obr6@
Fig. 6. Light flux path of Seoul Acrich 2

Survey of field mixing

When selecting a light source with an integrated inverter, luminaire or power component, it is advisable to monitor their mixing to ensure that it is acceptable for the intended use. Portable flicker meters, called flickermeters, or the flicker measurement functions of handheld spectrometers can be used for this purpose. For indicative testing, the author has built a flicker detector with a photodiode, an operational amplifier, and a small audio amplifier for a built-in loudspeaker that allows flicker to be "heard". This opened up a whole new world in which every light fixture, display or indicator light makes its characteristic sound. Considering that the brain registers these temporal changes in lighting, the question is what it actually does with them. These signals are very diverse, and luminous objects making no "sound" are surprisingly rare.

Two myths about shuffling

The incandescent light supposedly doesn't flicker

The instantaneous filament temperature of the bulb fluctuates slightly with twice the mains frequency, and so both the luminous flux and the chromaticity temperature change over time. The waviness factor (kf) of the bulb light tends to be around 5%.

The light of fluorescent lamps with electronic ballast (EP) is said not to flicker

The operating frequency of the EP in the order of tens of kilohertz suggests that temporal changes in light flux cannot be registered by the human eye. However, the output of the ballast is fed from a rectifier or PFC circuit, the output of which can be rippled in rhythm with the mains voltage. This ripple increases with the ageing of the filter capacitors in the power supply part of the ballast and in practice values of light ripple factor in excess of 5 % have been measured.

Literature:

[1] Helvar LED Luminaire Solutions [online]. [cit. 2018-11-06]. Available from: https://goo.gl/7W6bWc

[2] Seoul Acrich [online]. [cited 2018-11-06].

Author. Antonín Fuksa, NASLI & Blue step
Published in Světlo 1/2019