Improvement of temporal light artefact effects in commercial LED lamps

Abstract

Temporal light modulation and its potential adverse visual effects on human observers, collectively known as temporal light artefacts (TLAs), have been limited by the European Union Ecodesign regulation. Limitations on two key TLA metrics, flicker and stroboscopic effect, were introduced in September 2021. This study analyses the impact of this regulation on commercial LED lamps by measuring multiple different consumer-grade LED E27 retrofit lamps by several manufacturers before and after the regulation. The lamps purchased after the Ecodesign regulation seem to favour a certain type of LED driver that provides the best TLA behaviour of all lamp types. The regulation, with respect to TLA, has resulted in the replacement of ill-behaving lamp types in the market by newer lamps with better TLA characteristics.

1. Introduction

Temporal light modulation (TLM) is any measurable change of light level or spectral distribution of light over time which has been linked to adverse effects on human observers. These effects can vary in severity from mild disturbance to epileptic seizures.^1–3 The suspected adverse health effects on humans include eye fatigue, dizziness, anxiety, panic attacks, disorientation, migraine and photosensitive seizures.¹ This is why the European Union (EU) Ecodesign regulation sets requirements for lighting equipment on the European market since September 2021.⁴

The EU Ecodesign regulation sets restrictions for two temporal light artefact (TLA) metrics: flicker and stroboscopic effect.⁴ TLAs are visible phenomena caused by the temporally modulated light sources. Flicker is defined as the perception of visual unsteadiness of a temporally fluctuating light stimulus by a static observer in a static environment.² Flicker can be observed as either temporal change in the luminous output of a light source or colour of the light, both caused by the fluctuation of the light source itself. The visibility of flicker is described with a metric called short-term flicker severity index $(P_{st, LM})$ . The EU Ecodesign has set $P_{st, LM} < 1$ as the limit for flicker. In practice, this limit indicates that a light source does not cause visible flicker on average, as $P_{st, LM} = 1$ is the 50% probability threshold for the flicker visibility for an average observer.⁵

The stroboscopic effect describes a TLA by a static observer in a non-static environment. The observer perceives changes in the motion of a moving object, for example fingers which are moving close to the eye and are illuminated by a temporally modulated source.² The stroboscopic effect causes apparent changes in the movement speed or direction of objects, for example the so-called wagon-wheel effect. Similarly to flicker, the stroboscopic effect has a metric that describes its visibility. The metric is called stroboscopic visibility measure $(M_{vs})$ , sometimes also noted as $SVM$ . This metric works similarly to $P_{st, LM}$ , so that $M_{vs} = 1$ describes the probability of 50% for an average observer to see the effect.^1,6 The Ecodesign regulation limit for the stroboscopic effect is $M_{vs} < 0.9$ and will be further reduced to $M_{vs} < 0.4$ in September 2024.⁴

In this work, we study the effect of the EU Ecodesign regulation on the commercial LED lamps by measuring a total of 80 different types of E27 base LED lamps purchased before and after the Ecodesign regulation was set in action. These lamps were characterized by the type of electronic design used to provide current to the LEDs, that is, their driver topology, and their TLA properties were determined with two different TLA meter implementations. The TLA behaviour of the lamps produced before and after the regulation was compared.

2. Measurement setup

Requirements for the measurement of flicker and stroboscopic effect have been given in IEC standards TR 61547-1⁵ and IEC TR 63158.⁶ Same equipment has been used for both metrics, as the measurement setup requirements are similar. The primary quantity of interest is the relative light waveform, which means that either illuminance or luminous flux can be measured.³ We decided to use a luminous flux-based measurement setup as depicted in Figure 1.

Figure 1

Schematic of the measurement setup used to measure TLA metrics $P_{st, LM}$ and $M_{vs}$ . Abbreviations EUT and CVC denote equipment under test and current-to-voltage converter, respectively

The lamp, EUT in Figure 1, is placed in an integrating sphere and connected to an AC power supply with a power analyser. The stability of the electrical power source is essential for the measurement of temporal modulation, as having variations in the input current will affect the lamp behaviour, which is also stated by CIE in TN 012.³ IEC standards TR 61547-1⁵ and TR 63158⁶ have set tolerances for the input voltage and for the frequency of the signal. For both, the tolerance is $\pm 0.5 %$ from the nominal value. As the power supply, we used a Chroma 61601 programmable power source, with voltage resolution of 0.1 V and frequency resolution of 0.01 Hz. The power analyser was utilized to monitor the input current and the input voltage waveforms.

According to IEC standards,^5,6 the light form of the EUT needs to be measured with a light sensor with an optical $V (λ)$ filter. The measurements were carried out with a PRC-Krochmann TH1 photometer from the sphere output. This detector has a diffuser with a diameter of 8 mm, and it has a temperature controlled $V (λ)$ filter integrated into the device. The detector has a silicon photodiode to convert the detected light into photocurrent.

The photocurrent is connected to a CVC that has two main purposes. Firstly, it utilizes transimpedance amplification to convert the photocurrent to voltage and to amplify the signal to a suitable level for data acquisition. The amplification of the transimpedance amplifier can be selected between $10^{3} V / A$ and $10^{9} V / A$ . The second purpose for the CVC is to apply low pass filtering to the signal. In the IEC standards,^5,6 the cut-off frequency $f_{c}$ of the amplifier is recommended to be $f_{c} = 3$ kHz for the stroboscopic effect and 2 kHz for the flicker to reduce anti-aliasing of the signal. We used the lowest possible cut-off frequency, 10 kHz, from the CVC to measure both TLAs at the same time.

For the data acquisition, a Teledyne LeCroy HDO4054 oscilloscope was used. This oscilloscope has 12-bit vertical resolution and a 1.25 GHz maximum sampling rate. The oscilloscope converts the voltage signal to digitized data which holds two important parameters, the duration of the measurement $T_{m}$ and the number of samples $N_{s}$ . These parameters are directly linked to the sampling frequency $f_{s}$ as shown in Equation (1).

f_{s} = \frac{N_{s}}{T_{m}} .

(1)

The digitized data from the oscilloscope were then collected and analysed in two different sets of MATLAB TLA software implementations, which convert the measured waveforms into TLA metrics. The first set includes the ones recommended in IEC standards.^5,6 These can be found as MATLAB toolboxes for $P_{st, LM}$ and $M_{vs}$ , respectively.^7,8 The other set includes improved implementations published by Mantela et al.,⁹ which show better response to the test waveforms given in IEC standards.^5,6 The latter implementations can be found in a GitHub repository.¹⁰ The calculation for $M_{vs}$ has been validated against further improved implementations involving zero padding of 15 times (15×) the signal length. The zero padding is used to improve the frequency resolution for the fast Fourier transform that is used in the implementations. The two further improved implementations were well in agreement with each other, with an average difference of $\pm 0.005$ between implementations in the scale of $M_{vs}$ . Both sets of TLA implementations give both $P_{st, LM}$ and $M_{vs}$ values.

3. Measurement procedure

The measured lamps were installed into a lamp holder inside the integrating sphere. They were let to stabilize with driving current on for at least 30 min before the measurement took place. The oscilloscope was adjusted to fit the light waveform within the 12-bit range of the vertical scale. Resolution of 12 bits is sufficient for TLA measurements according to IEC^5,6 and Zong and Miller.¹¹ The data from the oscilloscope were downloaded to a computer. Both TLA implementations used to calculate $P_{st, LM}$ and $M_{vs}$ values were created in MATLAB; thus, we used MATLAB as the software platform for the data analysis.

The input voltage and current of each lamp were stored from the power analyser with a sampling rate of 32 kHz. The measured waveforms were used to identify the driver topology of each lamp.

4. Measured artefacts

We measured 80 different types of commercial E27 retrofit LED lamps that varied in manufacturer, nominal power and nominal luminous flux. Sixty lamp types were purchased during or before 2016, well before the EU Ecodesign regulation started. The other 20 lamp types were purchased in December 2021. The new lamps were purchased anonymously from multiple different grocery stores, home appliance stores and hardware stores in Finland.

As LEDs need DC to function, a power supply is needed to convert the AC mains electricity to DC. This can be achieved in multiple different ways, but in commercial lamps four basic topologies are used as described by Xu et al.¹² and further by Drapela et al.¹³ The four groups are defined with letters A to D as:

Type A: Full-wave rectifier with smoothing capacitor and DC–DC converter

Type B: Capacitive dropper circuit

Type C: Linear constant current regulation straight circuit

Type D: Switch-mode driver circuit

The driver topology used directly affects the input current waveform, thus by measuring that with a power analyser, the lamps can be classified into these four basic groups. Representative current waveforms for each topology can be seen in Figure 2, and their corresponding illuminance waveforms in Figure 3. The differences in the electrical waveforms between the driver types are related to the way each driver achieves the AC to DC conversion. For example, the apparent ‘noisiness’ of current in type D driver is due to an active power factor correction.¹²

Figure 2

Input voltage and typical input current waveforms for LED driver topologies A to D as classified by Xu et al.¹²

Figure 3

Measured illuminance waveforms for each driver topology A, B, C and D

5. Results

All the 80 lamp types measured were classified with respect to their power supply topologies. The distribution of these types can be seen in Table 1. In both lamp groups, type A lamps are clearly the most popular. With other lamp topologies, B and D types only appear in the old lamp types, whereas type C lamps are only found in the new lamps.

Table 1

Distribution of A, B, C and D type lamps in Pre-Ecodesign and Post-Ecodesign lamps

Type	Pre-Ecodesign lamps	Post-Ecodesign lamps
A	42	18
B	4	0
C	0	2
D	14	0

The older 60 lamps had their nominal luminous flux values between 150 lm and 2452 lm, the nominal power levels were between 1 W and 20 W, and they were produced by 22 different manufacturers. The newer 20 lamps had nominal luminous flux values between 30 lm and 1521 lm, and nominal power levels ranging between 0.6 W and 14 W. They were from nine different manufacturers.

The differences in $P_{st, LM}$ and $M_{vs}$ values and the driver topologies used before and after the Ecodesign regulation are shown in Figures 4 and 5, respectively. The blue box describes the difference between upper and lower limit of the middle 50% of the interquartile range (IQR). The whiskers of these boxes show the values within the 1.5 × IQR range. All values outside this range are considered statistically outliers and marked with red + signs. The red line inside the blue box describes the median value of all results.

Figure 4

Difference in $P_{st, LM}$ values and driver topologies before and after the EU Ecodesign regulation

Figure 5

Difference in $M_{vs}$ values and driver topologies before and after the EU Ecodesign regulation. The dashed and dotted lines show the limits of September 2021 and September 2024 set by the EU Ecodesign regulation, respectively.⁴ All new lamps with type A driver have $M_{vs}$ values between 0 and 0.04

As can be seen in Figure 4 for $P_{st, LM}$ , both old and new lamps show $P_{st, LM}$ values well within the Ecodesign regulation limit of 1. Small differences can be seen in type A lamps, newer lamps provide slightly smaller median $P_{st, LM}$ value, whiskers are shorter and have less outliers. The other types have slightly higher $P_{st, LM}$ than type A lamps.

More significant differences can be seen between the old and the new lamps when we compare $M_{vs}$ values in Figure 5. The B and D type lamps have significantly higher $M_{vs}$ when compared to the type A lamps. The D type lamps have the worst $M_{vs}$ values, having most of the lamps over the limit of 0.9. Type B lamps are, on average, slightly below the 0.9 limit, but still over the coming 0.4 limit line. Type A lamps have just a few outliers over the 0.4 limit line, thus having the best behaviour of the old lamps.

In the newer lamps, both lamp types show good $M_{vs}$ characteristics. Type C lamps, while having worse $P_{st, LM}$ values than B or D type lamps, have significantly better $M_{vs}$ values being around the coming 0.4 limit, and on the average below the limit. Similar to the $P_{st, LM}$ values with type A lamps, they have also improved in their $M_{vs}$ values, having in practice $M_{vs} \approx 0$ .

The TLA implementations by Mantela et al.^9,10 for $P_{st, LM}$ and $M_{vs}$ were used for the results in Figures 4 and 5 because their output values agree well with the expected values when using given test waveforms of the IEC standard.^5,6 However, comparison of the results was also made to those obtained using the IEC recommended implementations.^7,8 The EU mandates the manufacturers to declare the $P_{st, LM}$ and $M_{vs}$ values of their light sources in European Product Registry for Energy Labelling database.¹⁴ However, it seems that the manufacturers have characterized their products by the Ecodesign limits of $P_{st, LM}$ and $M_{vs}$ , instead of measured values.

Our measured $P_{st, LM}$ values for real LED lamps were small as can be seen in Figure 4. Therefore, when using the above-mentioned TLA implementations with this set of lamps, there are differences reaching 0.01 in only six lamp types in the results of the $P_{st, LM}$ implementations, while other values agree within the resolution of two decimals. When considering $M_{vs}$ values, the agreement with the implementations is mostly within ±0.01, apart from a single lamp, where the novel implementation with zero padding had a 0.05 higher $M_{vs}$ value. The data for all lamps are given in Tables A1 and A2 of the Appendix.

6. Conclusions

The TLA behaviour of LED lamps before and after the EU Ecodesign regulation was studied. The overall situation of the TLA artefacts in consumer-grade lamps has improved by manufacturers favouring the type A driver topology and bringing new type C lamps to the market. Manufacturers have also stopped producing B and D types of lamps that produce higher $M_{vs}$ values than the regulation allows. These actions have reduced the amount of TLAs in consumer-grade LED lamps.

The $P_{st, LM}$ values did not differ significantly between old and new lamps, which is expected, as the manufacturers have probably considered flicker well before the regulation, as flicker is much more noticeable than the stroboscopic effect. Thus, the regulation in practice limited the amount of $M_{vs}$ in the lamps. Large value of $M_{vs}$ is much harder to notice, because the higher frequencies cannot be directly observed, but require an external moving object to be visualized.

Footnotes

Appendix

Table A2.

Results of new lamps measured for TLA behaviour

Lamp ID	N1	EL1	K1	EL2	L1	R1	EP1	EL3	A1	EP2
P (W)	0.6	1	2.9	3	3	4.5	4.6	5	6	7
$ϕ_{v}$ (lm)	30	90	250	250	250	470	470	470	470	806
Type	A	A	C	A	A	A	A	A	A	A
$P_{st, LM (A)}$	0.14	0.05	0.11	0.02	0.03	0.04	0.01	0.04	0.07	0.05
$P_{st, LM (IEC)}$	0.14	0.05	0.11	0.02	0.03	0.04	0.01	0.04	0.07	0.05
$M_{vs (A)}$	0.02	<0.01	0.60	0.02	<0.01	<0.01	<0.01	0.02	<0.01	<0.01
$M_{vs (IEC)}$	0.03	0.01	0.60	0.02	<0.01	0.01	<0.01	0.02	<0.01	<0.01

Lamp ID	P1	R2	R3	K2	EP3	P2	L2	C1	EP4	EP5
P (W)	7	7	7	8	9	9.4	9.5	10	14	14
$ϕ_{v}$ (lm)	806	806	806	806	806	806	806	1055	1521	1521
Type	A	A	A	C	A	A	A	A	A	A
$P_{st, LM (A)}$	0.03	0.02	0.09	0.20	0.03	0.01	0.01	0.06	0.03	0.03
$P_{st, LM (IEC)}$	0.03	0.02	0.08	0.20	0.02	0.01	0.01	0.06	0.03	0.03
$M_{vs (A)}$	<0.01	<0.01	<0.01	0.04	<0.01	<0.01	<0.01	0.04	0.01	0.02
$M_{vs (IEC)}$	<0.01	<0.01	<0.01	0.04	<0.01	<0.01	<0.01	0.03	0.01	0.02

The nominal power, nominal luminous flux, the lamp driver topology and results obtained with both TLA implementations are presented. Subscript (A) shows results obtained with the software from the Github repository¹⁰ and subscript (IEC) shows results obtained with the IEC recommended implementations.^7,8 The $M_{vs (A)}$ values are from the implementation with 15× zero padding.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was carried out within the EMPIR programme project 20NRM01 MetTLM. This project 20NRM01 MetTLM has received funding from the EMPIR programme, co-financed by the participating states and the European Union’s Horizon 2020 research and innovation programme. The work is part of the Research Council of Finland Flagship Programme, Photonics Research and Innovation (PREIN), decision number 346529, Aalto University.

ORCID iDs

V Mantela

J Askola

P Kärhä

P Dekker

References

Institute of Electrical and Electronics Engineers. IEEE Recommended Practices for Modulating Current in High-Brightness LEDs for Mitigating Health Risks to Viewers. 1789-2015. New York, NY: IEEE, 2015.

Commission International de l’Éclairage. Visual Aspects of Time Modulated Lighting Systems – Definitions and Measurement Models. CIE TN 006:2016. Vienna: CIE, 2016.

Commission International de l’Éclairage. Guidance on the Measurement of Temporal Light Modulation of Light Sources and Lighting Systems. CIE TN 012:2021. Vienna: CIE, 2021.

Directorate-General for Energy. Commission Regulation (EU) 2019/2020 of 1 October 2019 Laying Down Ecodesign Requirements for Light Sources and Separate Control Gears Pursuant to Directive 2009/125/EC of the European Parliament and the Council and Repealing Commission Regulations. Brussels, Belgium: European Union, 2019.

International Electrotechnical Commission. Equipment for General Lighting Purposes – EMC Immunity Requirements – Part 1: Objective Light Flickermeter and Voltage Fluctuation Immunity Test Method. IEC TR 61547-1:2020 Ed. 3.0. Geneva, Switzerland: IEC, 2020.

International Electrotechnical Commission. Equipment for General Lighting Purposes – Objective Testing Method for Stroboscopic Effects of Lighting Equipment. IEC TR 63158:2018 Ed. 1.0. Geneva, Switzerland: IEC, 2018.

Banerjee

Light flicker assessment toolbox. Retrieved 23 May 2021 from https://www.mathworks.com/matlabcentral/fileexchange/63445-light-flicker-assessment-toolbox

Banerjee

Stroboscopic effect visibility measure toolbox. Retrieved 23 May 2021 from https://www.mathworks.com/matlabcentral/fileexchange/59242-stroboscopic-effect-visibility-measure-toolbox

Mantela

Nordlund

Askola

Kärhä

Ikonen

Digital implementations for determination of temporal light artefacts of LED luminaires. Lighting Research and Technology. First published 29 November 2023. DOI: 10.1177/14771535231212404

10.

Nordlund

Mantela

Aalto implementations for flickermeter and SVM meter [Online database]. GitHub: https://github.com/MantelV/Aalto-implementations-for-flickermeter-and-SVM-meter (2022, Updated April 2024, accessed 20 May 2024; with updated code using zero padding in the SVM meter, the deviation⁹ of M_VS test waveform Sin-4 improves from −0.0038 to −0.0001). DOI: https://doi.org/10.5281/zenodo.8380929

11.

Zong

Miller

Evaluation of data acquisition systems for measurement of temporal light modulations, in Proceedings of the CIE Symposium on Advances in Measurement of Temporal Light Modulation, CIE x049:2022, DOI 10.25039/x49.2022.P03.

12.

Collin

Djokic

Langella

Testa

Drapela

Experimental evaluation and classification of LED lamps for typical residential applications. 2017 IEEE PES Innovative Smart Grid Technologies Conference Europe (ISGT-Europe), Torino, Italy, 2017: 1–6.

13.

Drapela

Langella

Testa

Collin

Sjokic

SZ.

Experimental evaluation and classification of LED lamps for light flicker sensitivity. 2018 18th International Conference on Harmonics and Quality of Power (ICHQP), Ljubljana, Slovenia, 2018: 1–6.

14.

European Union. European Product Registry for Energy Labelling. Retrieved 9 July 2024 from https://eprel.ec.europa.eu/screen/product/lightsources