Experimental Evaluation and Classification
of LED lamps for Light Flicker Sensitivity
Jiri Drapela Roberto Langella Xiao Xu
Alfredo Testa Sasa Z. Djokic
Adam J. Collin
Dept. of Electrical Power Engineering Dept. of Ind. and Inform. Eng. School of Engineering
Brno University of Technology The University of Campania The University of Edinburgh
Brno, Czech Republic Aversa, Italy Edinburgh, Scotland, U.K.
[email protected] [email protected] [email protected]Abstract—The paper reports on the experimental evaluation and II. EXPERIMENTAL SET-UP
classification of LED lamps for light flicker sensitivity. After the
description of the experimental setup designed for tests, a
Experimental activities have been conducted in the
classification based on the driving circuits is proposed. For each laboratories of the Brno University of Technology.
class, the circuit description and the Gain Factor (GF) curves
A block diagram of the test system used for the GF curves
versus interharmonic frequencies are reported. Finally, a
comparison of the different GF curves is presented to show the measurement is shown in Fig.1. The test system hardware
Light Flicker sensitivity of the tested LED lamps. consists of a power amplifier 108-AMX from Pacific Power,
an analyzer LMG500 from ZES Zimmer, a PXI
Index Terms—LED lamps, light flicker, power quality. instrumentation system and a subsystem for lamps response
measurement. The PXI-8106 controls the test system
I. INTRODUCTION processes and sets, via a programmable waveform generator
Market trends confirm that light-emitting diode (LED) lamps PXI-5421 and 108-AMX amplifier, parameters of the
can be considered a "winning technology" for residential and fundamental and interharmonic voltages. The power source
commercial lighting applications worldwide [1]. In [2], a output feeds an LED lamp placed in an Ulbrich-type
comprehensive experimental analysis and classification, integrating sphere with a diameter of 2.5 m and supply
according to efficiency and power quality (PQ) indices, was voltage and lamp input current are measured. The radiant
conducted for 28 LED lamps (from 13 different power produced by the lamp is sensed by a photo-head
manufacturers) available on the EU residential lighting (human’s eye spectral correction, bandwidth: DC-327 kHz)
market. Results showed that LED lamps’ characteristics [9] and its output, which is proportional to the lamp luminous
differ significantly from other residential light sources (e.g. flux, is measured by DAQ PXI-4472 as part of the PXI
compact fluorescent lamps, CFLs) which typically have instrumentation system. Control of the measurement
similar electrical characteristics. procedure, including the source and analyzer control, data
Among PQ phenomena, light flicker (LF) is one of the reading from analyzer and spectra evaluation to obtain the GF
most important for distribution companies due to customer curves, is provided by a developed software running on the
complaints. It is well known that LED lamps (and CFLs) PXI platform.
behave differently from incandescent lamps [3]-[7] and that
new flickermeters need to be introduced to correctly measure
this phenomenon [8].
In this paper, the work done in [2] is completed to include Power amplifier
the LF sensitivity of LED lamps. The paper is organised as 108-AMX Analyzer LMG 500
follows: after the description of the experimental setup, a Supply voltage u i GPIB
Analog
signal
classification based on the lamp driving circuits is proposed. ∼φ
For each class, the circuit description and the gain factor (GF)
curves versus interharmonic frequencies are reported. Finally, LED
a comparison among the different GF curves is reported, to
Photo- NI PXI-1042Q
show the LF sensitivity of the tested LED lamps. with controller
head
Testing chamber PXI-8106
The paper was prepared at the SUN-EMC Lab. of the University of
Campania "Luigi Vanvitelli" and at the Centre for Research and Utilization
of Renewable Energy of the Brno University of Technology. The authors Figure 1. Test system for experimental measurements.
acknowledge financial support from Italian Ministry of University and
Research (PON03PE-00178-1) and from National Feasibility Programme I of
the Ministry of Education, Youth and Sport of the Czech Republic (LO1210).
978-1-5386-0517-2/18/$31.00 ©2018 IEEE
Authorized licensed use limited to: IEEE Xplore. Downloaded on September 09,2024 at 05:19:33 UTC from IEEE Xplore. Restrictions apply.
III. PROPOSED CLASSIFICATION A. Type A: Full-wave Rectifier with Smoothing Capacitor
In [2], four LED lamp categories have been introduced for and Switch-Mode DC-DC Converter Circuit
classification purposes, based on: i) ranges of electrical In [2], a first category (Type A), based on a full-wave
characteristics; ii) similarity of input ac current waveforms rectifier with smoothing capacitor ac-dc stage and dc-dc
(distinctive waveshape features), and iii) topologies of the converter circuit (without power factor correction, PFC,
electrical circuits. circuits) was proposed. This circuit uses a simple full-bridge
diode rectifier, dc-link bulk storage capacitor (C1), variable
In this section, these categories are recalled and new sub- input protection (fuse, resistor, metal-oxide varistor, MOV),
categories are introduced to fully characterize the behaviour of input electromagnetic interference (EMI) filter and output
the LED lamps in view of the LF performance. For each switch mode dc-dc converter to set LED string
category and subcategory, the typical circuit and the GF curves voltage/current. The specific dc-dc converter topology will
are shown, referring to the the lamps whose characteristics are change based on the application and/or manufacturer.
reported in TABLE I. Light Flicker studies require a further subdivision between
TABLE I. TESTED LAMPS fixed-control (Type A1) and feedback-controlled (Type A2)
dc-dc converters regarding output regulation.
Prated Lum. flux Color Temp. Dimmable
Type Name 1) Type A1: Full-wave Rectifier with Smoothing
(W) (lm) (K) (Y/N)
A1 LED J 10 806 2700 N Capacitor and Fixed-control Switch-Mode DC-DC Converter
A2 LED B 6 410 3400 N Circuit (without PFC)
B LED 13 [2] 9 806 3000 N
C1 LED 18 [2] 8 1010 2700 Y
Figs. 2a and 2b show the typical circuit of Type A1 LED
D1a LED 27 [2] 14 1055 3000 Y lamps and GF curves of the typical 60W incandescent (INC)
D1b LED 24 [2] 15 1500 6000 N lamp (red dotted curve) and tested LED lamp versus
interharmonic frequency. The response of the Type A1 LED
In general, the GF is expressed as a relative change in the lamps is comparable to that observed in case of a typical
output flux level divided by a relative fluctuation in the voltage screw-base CFL with a built-in electronic ballast [3]. It is
for a given fluctuation frequency [5] - [7]. possible to observe from Fig. 2b that in the frequency range
from 0 to 100 Hz, where the INC lamp produces light flicker,
In [3], a new concept, called the luminous flux gain factor
the LED lamp is less sensitive in a limited range around the
(GFΦ) based on the concept of "interharmonic families", was fundamental frequency, while in the remaining part of the
introduced. The GFΦ of the luminous flux for a given range it is more sensitive. Concerning frequencies over
background interharmonic voltage component superimposed on 100 Hz, the GF curves, shaped into characteristic bells,
the fundamental is given by the following equation: decrease with interharmonic frequency.
Φ h −1 Φ DC
GFΦ = (1) fixed-
VIH V1 Input contr.
,
protection EMI filter dc-dc
where Φh-1 is the amplitude of the interharmonic component of Fuse converter LEDs
MOV without
the lamp’s luminous flux belonging to the VIsible Family [3] Vac C1
PFC
VIFh-1, ΦDC is the luminous flux dc component, and VIH and V1
are the background interharmonic and fundamental components
of the supply voltage, respectively. The mentioned voltage
interharmonic family produces luminous flux components at the a) typical circuit topology of Type A1 LED lamps
following frequencies, intervening visible range (0, 50) Hz:
fΦ = h ⋅ f − 1⋅ f . (2)
1 IH
In tests, a supply voltage with a fundamental component of
lamp rated voltage (230 or 240 Vrms, 50 Hz) and an
interharmonic component of VIH/V1*100=2% at frequencies from
1 up to 800 Hz (in steps of 1 Hz) was applied to each lamp. All
basic principles were respected during the measurements; such
as 20 minutes warm-up period of lamp, stabilization in every
measurement step, etc. The evaluation procedure to obtain the
GF curve can be found in [7]. Finally, it is important to recall
that interharmonic voltages are able to produce the same voltage
fluctuations caused by pure amplitude modulation [10].
b) gain factor curves of incandescent lamp and Type A1 LED lamps
Figure 2. Type A1 LED lamp: Typical circuit topology and GF curves.
Authorized licensed use limited to: IEEE Xplore. Downloaded on September 09,2024 at 05:19:33 UTC from IEEE Xplore. Restrictions apply.
As shown in [3], the shape is determined by the capacity B. Type B: Capacitive Dropper Circuit
of C1 and the load of the lamp's converter circuit. Figs. 4a and 4b show the typical circuit of Type B LED
2) Type A2: Full-wave Rectifier with Smoothing lamps and the GF curves of the typical 60W incandescent
Capacitor and Feedback-Controlled Switch-Mode DC-DC lamp (red dotted curve) and of the tested LED lamp versus
Converter Circuit (without PFC) interharmonic frequency.
Figs. 3a and 3b show the typical circuit of Type A2 LED Type B lamps are very cheap and produced for the low-
lamps and the GF curves of the typical 60W incandescent end market due their circuit simplicity. The lack of feedback
lamp (red dotted curve) and the tested LED lamp versus control and the basic principle of a capacitor divider on which
interharmonic frequency. The LED B GF curve is multiplied the lamp design is based, make this category the most
by 10 to make the shape visible. sensitive to voltage fluctuations.
From Fig. 3b, it is possible to observe that the tested lamp This is evident in Fig. 4b, where it is possible to observe
is of very limited sensitivity to produce LF, compared with that Type B lamps exhibit characteristics similar to Type A1
incandescent lamps. This is due to the active control of the lamps. However, the gain is about twice as high. Moreover,
dc-dc converter by means of output feedback loop, which is in the range from 0 to 100 Hz, it is closer to the gain of the
able to compensate the dc voltage variations due to supply standard incandescent lamp. For frequencies higher than
side voltage fluctuations caused by interharmonics. 100 Hz, the sensitivity (GF curve) decreases with frequency.
Even if the bell like shape of GF curve in each doubled
harmonic range, due to energy transfer via the dc bulk
capacitor, is still present and observable in figure, the strong
LF attenuation effect compared to Type A1 is evident, as the
LF related sensitivity is about 20 times lower.
feedback R1
Input
Input contr.
protection EMI filter protection
dc-dc
Fuse converter LEDs Fuse LEDs
Vac MOV C1
Vac MOV C1 without C2 Dz
PFC
R2
a) typical circuit topology of Type B LED lamps
a) typical circuit topology of Type A2 LED lamps
b) gain factor curves of incandescent lamp and Type A2 LED lamps b) gain factor of incandescent lamp and Type B LED lamps
Figure 3. Type A2 LED lamp: Typical circuit topology and gain factor Figure 4. Type B LED lamps: Typical circuit topology and gain factor
curves. curves.
Authorized licensed use limited to: IEEE Xplore. Downloaded on September 09,2024 at 05:19:33 UTC from IEEE Xplore. Restrictions apply.
C. Type C: CCR Straight Circuit Figs. 5a and 5b show the typical circuit of Type C1 LED
In [2], another category based on a constant current lamps and the GF curves of the typical 60W incandescent
regulator (CCR) circuit was introduced. Light Flicker studies lamp (red dotted curve) and the tested LED lamp versus
require a further subdivision between linear and discrete CCR interharmonic frequency.
(a cascade switching), denoted as Type C1 and Type C2, It is very interesting to observe that due to the absence of
respectively, but in this paper only results for Type C1 are any energy storage (no dc capacitor) the GF (Fig. 5b) remains
reported. constant in each frequency sub-range of 100 Hz width
1) Type C1: Linear CCR Straight Circuit (double of supply voltage fundamental frequency) with
different values in all frequency sub-ranges. While the GF
The linear CCR is a type of a self-biased transistor, level of Type C lamps is inversely proportional to the current
maintaining the flowing-through current constant (i.e. CCR) conduction time, the GF curve scattering is related to the
over a wider voltage range, as required to supply the LED CCR regulation loop response to input voltage fluctuation. As
string. To avoid damage of the CCR transistor, an with Type A1 lamps, Type C1 lamps are less sensitive in a
overvoltage protection (OVP) circuit is integrated and limited range around the fundamental frequency when
activated when the CCR voltage is larger than a threshold compared to incandescent lamps, but are more sensitive in the
voltage. In order to achieve the sinusoidally-shaped remaining part of the range.
waveform of ac line current and circuit efficiency, the CCR is
typically fed by an ac-dc stage consisting of a full-bridge D. Type D: Active Power-Factor-Corrected Single or Multi-
uncontrolled rectifier with no output (dc side) capacitor. Stage Driver Circuit
Therefore, the line current corresponds to the current drawn
In [2], a single category covering all LED lamps utilizing
by the CCR, but its conduction range in each half-period of
any switch-mode driver circuit with active power factor
the system frequency is reduced by the CCR - LED string
correction (aPFC) was introduced.
chain bias voltage [2].
However, study of Type D LED lamp LF performance
requires a further subdivision, distinguishing between single-
stage (S-S) and double-stage (D-S) topologies, both providing
aPFC.
1) Type D1: Single-Stage Switch-Mode Driver Circuit
Single-stage aPFC topologies are very common for low
power self-ballasted lamps. The typical circuit of Type D1
LED lamps is shown in Fig. 6a. The S-S aPFC circuits
originate from D-S topologies composed of two separate
switch-mode dc-dc converters, where each performs a
dedicated role. The first converter, starting from the ac side,
a) typical circuit topology of Type C1 LED lamps serves as PFC and a pre-regulator, while the second (the
output) provides load feeding according to the specific
requirements. Merging both together, an S-S circuit is
obtained, which usually cannot provide all of the D-S circuit
functionalities properly. Therefore, S-S drivers, taking into
account applicable topologies, can offer either better ac line
current waveform, or better regulation on constant output, but
not both. This will have an effect on LF performance. In this
view, two separate D1 categories (D1a and D1b), based on
the converter control strategy, have to be introduced. Type
D1a LED lamps are of “(PFC) aimed” control, while type
D1b LED lamps are of “regulation on constant output (CO)
aimed” control.
Figs. 6b and 6c show the GF curves of the typical 60W
incandescent lamp (red dotted curve) and the tested Type D
LED lamp versus interharmonic frequency.
b) gain factor of incandescent lamp and Type C1 LED lamps
Concerning the GF (Fig. 6b), it is possible to observe that
Figure 5. Type C1 LED lamps: Typical circuit topology and gain factor in the frequency range 0-100 Hz the tested lamp belonging to
curves. Type D1a is less sensitive to voltage interharmonics, than the
Authorized licensed use limited to: IEEE Xplore. Downloaded on September 09,2024 at 05:19:33 UTC from IEEE Xplore. Restrictions apply.
incandescent bulb. It can also be seen that the PFC-focused As shown in Fig. 6c, Type D1b lamps behave differently.
control strategy is characterized by: i) bell shape GF curve It is possible to observe that in the frequency range from 0 to
segments, connected to the non-driven charging of a dc bulk 100 Hz, where the INC lamp produces light flicker, the LED
capacitor (as in the case of Types A and B lamps), which has lamp is, in the relevant range closer to the fundamental
to be present and is placed in the output stage, and ii) frequency, less sensitive. Concerning frequencies over
smoothed scattering of the GF curve segments related to the 100 Hz, the GF curve decreases quickly with interharmonic
dominant input current control mode, similar to Type C frequency. The change in the GF shape in each segment,
lamps. Compared to Types A1, B and C lamps, the sensitivity compared to the Type D1a lamp, is due to a dominant
of Type D1 lamps to light flicker related to voltage constant output regulation mode that actively compensates
disturbances is reasonably reduced. any variation in the output level introduced by input voltage
fluctuations.
single-stage
2) Type D2: Switch-Mode Driver Circuit (double stage)
(PFC)
dc-dc The Type D2 double-stage driver circuit topology is very
converter
or
common in the case of external LED drivers and is also
(CO) employed in electronic ballasts for linear fluorescent lamps,
dc-dc or in high intensity discharge lamps with input power over
converter 25 W. However, the double stage topology is not presently
considered in self-ballasted LED lamps for household
a) typical circuit topology of Type D1 LED lamps
applications. Therefore, it is out of the paper scope. However,
it is important to note that a D-S topology typically performs
output stabilization at the first (PFC) stage, or even at both
stages, resulting in very small sensitivity of the D-S LED
drivers to flicker related voltage disturbances (even lower
than of Type A2 lamps), as they provide low distortion of ac
line current with very high power factor.
IV. COMPARISON OF GF CURVES
In order to compare light flicker sensitivity of different
lamps, two frequency ranges have been selected. The first
frequency range is from dc (0 Hz) to 100 Hz, corresponding
to the only range in which incandescent lamps are sensitive to
voltage fluctuations and produce LF.
In this frequency range, the GF curves of the tested LED
lamps are normalized with reference to the GF curve of the
b) gain factor of incandescent lamp and Type D1a LED lamps standard incandescent lamp, as shown in Fig. 7.
It is possible to observe that:
• as previously commented, Type A1, Type C and
Type D1b LED lamps are more sensitive than
standard incandescent lamps symmetrically outside
the frequency ranges from 24 Hz to 74 Hz, from
32 Hz to 68 Hz, and from 24 Hz to 74 Hz,
respectively;
• Type B LED lamps exhibit a non-symmetrical
sensitivity with respect to incandescent lamps, which
is mostly lower from dc to 50 Hz and mostly higher
from 50 (62) Hz to 100 Hz;
• Types A2 and D1a are less sensitive than
incandescent lamps, mainly due to the dc-dc
converter control of the driving current of the LED
c) gain factor of incandescent lamp and Type D1b LED lamps
chain.
Figure 6. Type D1: Switch-Mode Driver Circuit (with a-PFC) and gain
factor curves
Authorized licensed use limited to: IEEE Xplore. Downloaded on September 09,2024 at 05:19:33 UTC from IEEE Xplore. Restrictions apply.
Concerning the second frequency range, from 100 Hz to classification based on the driving circuits from [2] has been
800 Hz, where the GF curve of the incandescent lamps are revised and new subcategories of LED lamps are proposed.
equal to zero, normalization of GF curves of the tested lamps For each (sub)category, the circuit description and the GF
is not applied, as shown in Fig. 8. curves versus interharmonic frequencies have been reported.
Finally, a comparison between the different GF curves has
It is possible to observe from Fig. 8 that the sensitivity of
been reported, to show the ranges of LF sensitivity for all
different LED lamps depends on the driver type and varies in
tested LED lamps.
wide ranges. Particularly, all tested LED lamps, except Type
B, are less sensitive than Type C. Type B lamps exhibit It is clear from the presented results that the rapid
higher or lower sensitivity depending on the frequency development of LED lamp technologies, together with the
ranges. larger number of producers has resulted in dramatic
differences in lamps’ LF performances. The main conclusion
of the paper is in line with the previous findings from [2], i.e.
that LF sensitivity of LED lamps is strongly influenced by the
lamp design and the implemented circuit topology.
Accordingly, the LED lamp LF performance classification
presented in this paper can be combined with the previous
classification of LED lamp harmonic emission and
operational performance, in order to define a preliminary
selection tool for lighting system designers. Furthermore,
both the methodology and results presented in this paper can
be used for practical evaluation of the actual PQ and flicker
performance of different LED lamps. This is the subject of
further work by the authors.
REFERENCES
[1] Dept for Business, Energy and Ind. Strategy, “National Statistics:
Energy Consumption in the UK”, Energy and Climate Change &
Figure 7. Comparison of GF curves of alltested LED lamps normalized Energy Consumption in the UK: Evidence and Analysis, Nov. 2016.
with respect to the incandescent lamp GF in the frequency range [0-100 Hz]. [2] X. Xu, A. Collin and S. Z. Djokic, R. Langella, A. Testa, J. Drapela,
“Experimental Evaluation and Classification of LED Lamps for Typical
Residential Applications,” IEEE ISGT EuropeT2017, Turin, Italy, Sept.
2017.
[3] J. Drapela, R. Langella, J. Slezingr, A. Testa. " Generalized Lamp
Model for Light Flicker Studies", Electric Power Systems Research,
154, 2018, pp. 413-422.
[4] IEEE Recommended Practice 1453.2012 —Adoption of IEC 61000-4-
15:2010, Electromagnetic compatibility (EMC)—Testing and
measurement techniques—Flickermeter— Functional and design
specifications.
[5] CIGRE WG C4.111 Review-of-LV-and-MV-Compatibility-Levels-for-
Voltage-Fluctuation. [Online] http://c4.cigre.org/WG-Area/WG-C4.111
[6] IEEE Standard P1453-2015, Recommended Practice for the Analysis of
Fluctuating Installations on Power Systems.
[7] J. Drapela, M. Kratky, L., Weidinger, M., Zavodny, Light Flicker of
Fluorescent Lamps with Different Types of Ballasts Caused by
Interharmonics. 2005 IEEE St. Petersburg PowerTech Proceedings. St.
Petersburg, Russia, IEEE PES, 2005, 7pp., ISBN 5-93208-034-0
[8] J. Drapela, R. Langella, J. Slezingr and A. Testa, "A Tunable
Flickermeter to Account for Different Lamp Technologies", IEEE Tran.
on Power Delivery, Volume: 32, Issue: 2, April 2017, pages 872 - 880.
Figure 8. Comparison of GF curves of all tested LED lamps in the [9] J. Drapela, J. Slezingr, "A Light - flickermeter -Part II: Realization and
frequency range from 1 Hz to 800 Hz. Verification". Electric Power Engineering (EPE) 2010, Brno, pp. 459-
464. ISBN: 978-80-214-4094- 4
V. CONCLUSIONS [10] R. Langella, A. Testa, “Amplitude and Phase Modulation Effects of
Waveform Distortion in Power Systems”, Electrical Power Quality and
In this paper, the previous work done by the same authors Utilization Journal, Volume XIII, N° 1, 2007, pp. 25-32.
on the efficiency and PQ characteristics of commercially
available LED lamps for residential applications in [2] has
been completed to include the LF related sensitivity of LED
lamps. After the description of the experimental setup, a
Authorized licensed use limited to: IEEE Xplore. Downloaded on September 09,2024 at 05:19:33 UTC from IEEE Xplore. Restrictions apply.
