Flicker: Standards and Test Methods

Depending on the details of the fluctuations, TLAs may consist of flicker, which is directly perceived as light fluctuation or stroboscopic effects, which is the misperception of motion, or both.

Assuming this phenomenon, particularly with the SSL technology, is viewed by the industry as an issue that needs to be addressed, several grey areas require clarification. One of these areas is the creation of a standard with appropriate measurement metrics.

Where Low Flicker Is Crucial and Where It Is Not

Different situations require a different focus on flicker, largely based on location, historical experience, likely exposure time and the sort of activities taking place.

In an outdoor environment, such as a street or a parking lot, there is little documentation of flicker complaints, and light sources with a high flicker may not have a negative impact in such situations. If the outdoor environment hosts evening sporting events, however, having a low flicker light source becomes important to avoid stroboscopic effects on the field.

Moving indoors, in an office or educational environment where individuals are exposed for a length of time to artificial light, while performing complex tasks, low flicker may decrease eye fatigue and be beneficial for migraine sufferers.

In an industrial environment the situation again needs careful consideration. In a warehouse with limited objects in motion and few visual tasks, low flicker is preferred but is not a necessity. In a production facility with many moving pieces of machinery, low flicker is an essential condition to avoid a misperception of moving parts.

Energy Saving Needs

According to the different kinds of environments and light needs, the lighting industry has developed dimmable products to help save energy.

Any dimming control, from a wall-box dimmer to an automated daylight harvesting system, has the potential for system mismatch and can introduce additional flicker. A phase-cut, wall-box dimmer has the most potential for additional flicker, although other methods can introduce at least some flicker.

While the application impacts of flicker have not been well studied, a good knowledge of light source and/or luminaire flicker characteristics, together with using good practices when considering the tasks of a space and the selection of lighting, may help to avoid discomfort among users. This is particularly important for LED installations that may be operating for many years.

Although some documents providing measurements metrics have been published on this topic, there are contradictions among them. Below is a brief summary of the main documents and the key aspects of each.

IEEE 1789

“IEEE 1789: IEEE Recommended Practices for Modulating Current in High-Brightness LEDs for Mitigating Health Risks to Viewers” is a document describing the challenges represented by flicker and some potential health impacts. It also provides recommendations for minimizing any risk of adverse effects. The document is not a standard and the recommendations given are very conservative, to the point that some traditional lamps are unable to meet the requirements. Despite this limitation, these recommendations can be useful when minimal flicker is required.

ENERGY STAR®

EPA’s ENERGY STAR program (developed by US DOE) introduced a CFL frequency requirement many years ago. Only recently has ENERGY STAR addressed specific recommendations for measuring flicker, with Lamps Version 1.0. The recommended practice for measuring requires flicker index and percentage, as well as testing with five different dimmers (in case of dimmable products). The program does not yet provide a specific requirement, but it is gathering information to introduce a specific flicker requirement in a future revision.

California Title 20 and Title 24

Shortly after ENERGY STAR released its recommended practices for the measurement of flicker, the state of California introduced its set of requirements, listed in title 24, Joint Appendix 10 test method for flicker.

California’s requirements include the test method “Joint Appendix 10” (JA10) that requires measuring the light output of a source or luminaire and dimmer for two seconds and then running the data through several complex calculations to evaluate flicker at multiple frequencies up to 400 Hz. Title 24 requires high efficacy sources to be “low flicker operation”, which means that the LED product will have less than 30% flicker at frequencies below 200 Hz. This requirement goes into effect in January 2017. Title 20, which also requires “low flicker operation”, will cover LED lamps, requiring that lamps be tested after being paired with controls.

IEC/TR 61547-1

This document adopts the same principle PstLM (illuminance measured with the light flicker-meter) described in the EMC standard IEC 61000-3-3 about the flicker metric definition and test method, utilizing a flicker-meter, as defined in the standard IEC 61000-4-15. The principle is to detect any possible flicker, due to voltage fluctuations on the grid, generated by variable loads, e.g, washing machines. The signals captured during the measurements through a light flicker-meter are acquired for a defined period and processed.

The outcome of the measurements is basically a numeric indication:
< 1 Acceptable because flicker is not visible (up to max 70 Hz)
= 1 As above
> 1 Unacceptable because flicker is visible and may cause adverse visual effects

This measurement metric is limited because it is based on a 60 W conventional incandescent lamp, currently banned from the market due to energy efficiency requirements. Furthermore, it does not include mains voltage fluctuation disturbances for voltages other than 230 V/50 Hz.

Someone may think that using a stabilized power supply is enough to solve the issue; however, ripple on the output may still cause flicker. A revision of this document has recently begun.

CIE TC 1-83

This document, which takes into account the visual aspects of time-modulated lighting systems, specifies TLA metrics and terms, and it includes the metrics PstLM (flicker) and SVM (stroboscopic effect). The metric for the detection and measurement of flicker proposed in this document is similar to that contained in the IEC/TR 61547-1 (under revision).

Japan DENAN Law

DENAN Appendix 8 and J60598-1(H26): JIS C 8105-1: 2010 + Amd. 1 (2013).

This document addresses flicker detection in these terms:
SSL products are deemed not to create a sensation that the light output is flickering if: the lighting fixture has a cyclic frequency of 100 Hz or more with no output failure (5 percent% or less of the peak value of light output) or a cyclic frequency of 500 Hz or more of light output.

Considerations

Although some of the above mentioned documents do not account for non-visible flicker (>70 Hz), which is considered stroboscopic effect, non-visible flicker has not been proven completely safe. In fact, if we assume that a small portion of the population may be affected by serious health effects, and others by less serious side effects, e.g. eyestrain or, headache, it becomes clear that we are now dealing with the quality of the light itself.

A product with no visible flicker may still cause issues in certain situations, e.g. installations open to the public where a high quality of light is required and video recording may be necessary for security reasons. Again, stroboscopic effect is an unwanted side effect in sports environments where the motion of objects may be wrongly perceived, e.g. phantom array in motorsports and games with moving balls.

One of the major issues not yet addressed by the available documents, is the evaluation of dimmable products, e.g. phase cut dimming and PWM products. In particular, reliable measurements metrics should carefully apply limits in the SSL luminaires dimmed through PWM techniques (100% modulation depth), where the typical operating dimming frequency is in the range of 200-400, up to 800 Hz, because it can be difficult to address TLAs up to 1,25 KHz.

Many other questions may arise, particularly from lighting designers and architects who wish to combine several products and/or light sources. One of these could be “what is happening in an installation with several luminaires where one or more products are affected by flicker or stroboscopic effect issues?” A typical metric relies on type testing (except for California Title 24 that requires three samples to be tested, while one sample is usually verified). No specific tests have been conducted so far on multiple TLAs, because if several products of the same genre and model are affected by flicker, it is reasonable to assume that the flicker effect will be multiplied rather than lowered. Although no specific tests have been conducted to address the multiple flicker issue, it can be assumed that perception may be negatively affected. Unfortunately, type tests are not able to solve this issue.

In conclusion, with TLA issues reducing the quality of today's light, the lighting industry is asking for reliable metrics, and, eventually, the application of limits. Ideally, these metrics would be differentiated by the destination of use of the lighting fixtures, the type of environment (e.g. public or private) and the duration of exposure to the artificial light.

A worldwide-recognized standard with appropriate test methods would be the best solution for the lighting industry. After taking the published literature into consideration, the development of a possible metric is described below.

Flicker Percent (FP) and Flicker Index (FI)

Flicker can be more or less apparent depending on several factors, primarily the relevant amount of variation in the light per cycle, the proportions of the lighting waveform, and the frequency (or frequencies) at which the light variation occurs. To describe the variation within a cycle, there are two primary measures: flicker percent and flicker index. Flicker percent is the measure of the maximum light vs. the minimum light in a cycle. This only accounts for the minimum and maximum light outputs, and does not differentiate between waveforms. This is the simplest form of flicker to determine. Flicker index is another common metric for describing the behavior in terms of the amount of light that a product produces over a given cycle. Flicker index requires more calculations than flicker percent, as there is consideration given to the shape of the waveform. Flicker index considers the area of the waveform above and below the average light output.

The difference in perceptibility in these two metrics is one that is still in debate. However, it is generally acknowledged that the perceptibility of both is dependent on the frequency at which a product operates.

With the shift to electronic ballasts operating at 40 kHz or more for greater efficiency, flicker issues were largely eliminated from fluorescent lighting. However, flicker has reappeared with LEDs. There is a balance between size, cost and lifetime when designing drivers, and many LED drivers operate at lower frequencies or contain lower frequency components that can cause perceptible flicker.

Figure 1: Example light waveform that helps to understand the definition of flicker [1]

These metrics do not quantify TLA effects correctly and objectively. Flicker percent and flicker index are not selective (i.e., they do not distinguish between “flicker” and “stroboscopic effect”) and do not account for the effect of frequency-dependent sensitivity or the wave shape of the light output, additional verifications are required to cover these aspects.

Several studies demonstrated that, even if a population has different threshold sensitivities, flicker is visible up to 70-80 Hz; above this value we need to discuss stroboscopic effect.

ASSIST metric, developed by the Lighting Research Center, is an additional metric that completes the other two, providing a numeric value that can be used to predict the flicker perception.

Testing Procedure

Test setup
The product to be evaluated requires a controlled environment consisting of a test enclosure able to maintain a constant temperature of 25°C ±2°C.

Test enclosure
The test enclosure shall not have to admit stray light. The measurements that will be obtained are relative, an integrating sphere is not strictly required, however it is ecommended to test the product in the sphere, this will help to grant better repeatability of measurements.

Photodetector
The photodetector fits the International Commission on Illumination (CIE) regarding the spectral luminous efficiency curve, linearity of response over the measurement range and response time of the sensor.

Signal amplifier
If necessary, a signal amplifier may be placed between the photodetector and the data collection device.

Device for data collection
Appropriate instruments like a digital oscilloscope with data storage capability or similar equipment able to store high frequency data from the photodetector may be needed.

Input power supply
Input power to UUT (unit under test), is required at the rated primary voltage and frequency within 0.5% for both voltage and frequency. When ballasts are labeled for a range of primary voltages, the ballasts are to be operated at the primary application voltage. The voltage will have a sinusoidal wave shape and a voltage total harmonic distortion (THD) of no greater than 3%.

Sampling parameters
Measured data are to be recorded in a digital file with an interval between each measurement no greater than 0.00005 sec (50 microseconds) corresponding to an equipment measurement rate of no less than 20 kHz. The equipment measurement period shall be greater than or equal to two seconds. In case of dimmable products, record lighting measurements (in foot-candles or volts) from test equipment with readings are taken at intervals of no greater than 50 microseconds for each dimming level after the products have stabilized. These readings are compiled for an equipment period of no less than two seconds and recorded into a comma separated data file (*.csv).

Signals acquired
These are then to be processed using the ASSIST metric, to determine the likelihood of flicker detection (as a percentage). In addition to the device-level flicker, lighting system design can have an influence on the amount and type of flicker that is experienced by an observer. Methods of dimming can introduce additional flicker into the light output. Phase-cut type dimmers alter the incoming power to the lamp, driver or ballast, which will often alter the light frequency components coming out of the product. Some drivers and ballasts do a very good job of smoothing this; others transfer more of the cut wave to the light output. A common method of dimming in LED drivers is pulse-width modulation, or rapidly changing the LED light output to make the light appear dimmer, but also introducing additional moderate frequency components (typically 400-800 Hz). The above measurements must be repeated at different dimming levels.

The ASSIST metric takes into account the wave shape of the light waveform. The sensitivity curve used for sinusoids is quite similar to the PstLM sensitivity curve as recalled in the IEC/TR 61547-1. Frequency domain processing: weighted summing of spectral components. Phase information of the various frequency components is not taken into account, which is relevant for complex waveforms.

Figure 2: The necessary steps that a good metric must include to obtain reliable data

The possible limits of this metric may be:

• Duration of a few seconds of the waveform is not sufficient for detecting low-frequency flicker. Longer acquisition time is possible, but will require a measuring instrument with appropriate data processing capacity
• Flicker due to single-event modulations is not detected. This has to be thoroughly studied to determine if should be considered a random event or if it might be perceived.
  Figure 3: The UL packaging mark

UL Marketing Claim Verification Verified Program

In response to numerous inquiries and requests, UL recently introduced a marketing claim verification program that enables manufacturers to test their products for a neutral, third party assessment of flicker performance. UL Verification addresses low optical flicker in lighting products having well dfined limits.

Limits for the UL Verification:

• 10% for products powered by mains voltage with a sinusoidal 60 Hz frequency
• 8% for products powered by mains voltage with a sinusoidal 50 Hz frequency

Products that successfully have their flicker claims verified may include a packaging mark of “Low Optical Flicker Less than X%” where “X” indicates the percentage detected as a result of the tests.

Though many gray areas still exist regarding TLAs, market understanding continues to grow and evolve. A worldwide metric with standardized test methods would allow the global lighting market to better rate and understand product differences, but more work is needed before a metric can be implemented. In the meantime, UL continues to work with the global lighting community in an effort to increase understanding, improve testing, and help ensure the safety of those using artificial light.

References
[1] Adapted from IES Handbook