LED driver architectures determine SSL flicker, explain ZHAOQI MAO, LANE GE, and GARY HUA, but techniques that suppress ripple can prove cost effective and deliver comfortable LED-based lighting.

Replacing traditional incandescent and fluorescent lights with more efficient, and longer-lasting LED-based solid-state lighting (SSL) is an undeniable trend in the lighting industry. However, since SSL fixtures are directly connected to the AC line, like with legacy lighting, there is a risk that 100-Hz or 120-Hz flicker could occur as a result of driving current ripple at the supply`s output. Flicker can make people uncomfortable, causing headaches and other maladies even though the human eye may not detect the flicker. Careful LED driver design can minimize flicker and help ensure that SSL delivers on its energy-efficiency promise through broad deployment.

Indeed, LED and luminaire manufacturers are keen to solve the flicker problem, and they have turned largely to the driver manufacturers for a solution because ultimately the driver determines the extent of flicker. This article will explain the root cause and characterization of flicker, describe how it occurs in LED lighting, and explain how engineers can trade off different driver topologies in an effort to find the right cost/benefit combination. We will describe a ripple suppressor circuit that is a cost-effective and flexible way of implementing flicker-free LED lighting.

Market expectations

It is a well-acknowledged trend that LED lighting will replace legacy sources including incandescent bulbs and fluorescent bulbs and tubes in the next few years. Since LED is a new-generation light source with electronic roots, the market expects not only the higher SSL system-level efficacy but also a better lighting environment.

Like the legacy lighting technologies, however, most LED lights are directly connected to the AC mains that operates at 50 or 60 Hz in different global regions. Even after rectification to frequencies of 100 or 120 Hz, any line-associated flicker may be detectable by the human eye due to the relatively low frequencies involved. Indeed, the rectified line may lead to strobe flicker, which can cause the previously mentioned distress in humans.

In particular, LED driver designs that rely on a single-stage architecture to handle both power factor correction (PFC) and output drive current are especially susceptible to flicker. Among other causes, flicker is affected by the LED ripple current. But there are numerous methods to eliminate the problem including a ripple suppressor circuit. SSL product developers need to evaluate the approaches and choose a driver that meets the application requirements in cost and performance. Different levels of flicker are acceptable in different applications. A system design that is unacceptable in indoor applications due to flicker may perform fine in an outdoor street or area lighting role.

The effect of flicker

Now let`s examine flicker. According to the paper "A review of the literature on light flicker: Ergonomics, biological attributes, potential health effects, and methods" that was published by the Institute of Electrical and Electronic Engineers (IEEE) in 2010, light flicker frequency in the 3–70-Hz range is highly perceptible by human beings and this could make them very uncomfortable (http://bit.ly/1iL2vjq and http://bit.ly/NAtjXq). Even repetitive flashing lights and static repetitive geometric patterns may induce seizures in these individuals, and the occurrence rate is around 0.025%. This type of flicker can be easily solved by the driver, and normally we regard the driver as unstable if that kind of frequency can be seen in the output current ripple waveform.

Now, however, people are beginning to pay more attention to long-term exposure under higher-frequency flicker in the 70–160-Hz range. Such flicker can cause malaise, headaches, and visual impairment. Some researchers even claim the retina can sense flicker up to 200 Hz, but tests have shown that above 160 Hz the health effects of flicker are negligible. Because of the previously mentioned rectified line frequencies of 100 and 120 Hz, we will focus on mitigating flicker in that range here. And in actuality the human health impact of 100- or 120-Hz flicker is a function of not only frequency but also physical and physiological factors.

Defining flicker

We first need to understand how flicker is characterized. The Illuminating Engineering Society of North America (IESNA) released the definition of "percent flicker" and "flicker index" in the ninth edition of The IESNA Lighting Handbook. Fig. 1 shows how the metrics are defined.

FIG. 1. The IES defines flicker index.

Percent flicker is a relative measure of the cyclic variation in output of a light source (i.e., percent modulation). This is also sometimes referred to as the "modulation index." Referring to Fig. 1, you calculate percent flicker based on the maximum (A) and minimum (B) light output levels. You divide the sum of A and B by the difference of the two to obtain a percentage.

Flicker index is defined in the IESNA handbook as a "reliable relative measure of the cyclic variation in output of various sources at a given power frequency. It takes into account the waveform of the light output as well as its amplitude." The flicker index assumes values from 0 to 1.0, with 0 for steady light output. Higher values indicate an increased possibility of noticeable lamp flicker, as well as stroboscopic effect. Referring again to Fig. 1, you calculate the index by dividing Area 1 by the sum of Area 1 and Area 2.

As noted previously, in addition to the frequency, the flicker index has a significant effect on how light makes people feel. Higher flicker index means more sensitivity to the human eye and a poorer comfort level. Table 1 shows the typical flicker index of different light engines according to the paper "The evaluation of flicker in LED luminaires" written by Michael Grather, president of Luminaire Testing Laboratory, and posted on the Council for Optical Radiation Measurements (CORM) website.

SSL light source output

Having defined flicker, let`s consider how an LED light source operates. LED light output is almost linear with the drive current. Peruse any high-power LED data sheet and you can see the linearity in graphs that plot forward current relative to luminous flux. Such a plot makes it quite obvious that the drive current is a critical source of LED light flicker, and constant current supply is the primary job of the LED driver.

When we discuss 100–120-Hz flicker, most often we are focused on indoor lighting applications. There are quite a few LED driver schemes for indoor lighting that can provide a constant current. For example, simple current-limiting resistors, linear semiconductor regulation, and switching pulse-width modulation (PWM) regulation following AC rectification are all possibilities. But those schemes are outside our scope here because they are not able to provide the power factor (PF) required for indoor commercial applications. Typically, commercial applications require a PF greater than 0.9. More and more countries and standards associations like Energy Star and DesignLights Consortium (DLC) require lighting to have greater than 0.9 PF value. We predict that any lamps and luminaires with lower than 0.9 PF will be phased out soon.

Driver topologies

Given our PF requirement, let`s consider some driver topologies that can be used in indoor SSL products along with the cost and performance implications of each. The various approaches are summarized in Table 2. We will also introduce a new scheme to lower the large ripple that is common in single-stage drivers.

FIG. 2. Passive or valley-fill PFC stage plus a DC/DC-converter stage in an LED driver.

Passive PFC plus switching DC/DC. Fig. 2 depicts a two-stage design that includes a passive PFC stage along with a switching DC/DC converter second stage. This structure is widely used in low-cost offline adapters and chargers. The PF design is often referred to as valley fill as capacitors keep the output from falling to low levels. Thanks to the valley fill circuitry and the bulk capacitor, the current ripple of this scheme is small and easy to control. The drawback of the passive scheme is that it is not suitable for higher power over 20W because of inherently poor PF electromagnetic compatibility (EMC) issues at higher power levels. Such designs are incapable of passing the IEC EN61000-3-2 (Harmonic current emission test) class C standard. In addition, the passive scheme is not suitable to achieve wide universal input voltage ranges such as 100–240 VAC.

FIG. 3. Single-stage active PFC driver architecture.

Single-stage active PFC. The single-stage approach with active PFC that is depicted in Fig. 3 is a widely adopted topology for wide input range LED drivers. The topology delivers good power-conversion efficiency and PF value with wide load range as well. The drawback is the high current ripple that leads to visible or invisible 100–120-Hz flicker. Good designs can reduce the current ripple to a relatively low value; however, the ripple is normally still higher than the previous two-stage scheme. One interesting feature of the single-stage topology is that the ripple is greatly affected by the different voltage and current characteristics that are specific to each LED load. Driver designers are seeking better ways to control the ripple in the single-stage design.

FIG. 4. Active PFC stage plus a DC/DC-converter stage in an LED driver.

Active PFC plus switching DC/DC. One way to solve the output ripple problem is to add an active second stage behind the active PFC stage. Such a topology is depicted in Fig. 4 with the addition of a DC/DC converter stage. But the additional DC/DC stage in the driver comes with a cost increase of 15–20%. This circuitry greatly lowers the output current ripple and makes the output almost an ideal DC at the expense of losing 2–3% efficiency. Moreover, this structure can cover most of the power levels required in indoor applications and is widely used.

FIG. 5. Ripple suppressor block diagram.

Single-stage PFC and ripple suppressor. Ideally, SSL system developers would prefer a lower-cost method to reduce the output ripple, and that would bring us back to a single-stage approach. Fortunately, there is another good solution for lowering the output current ripple with a circuit that`s far simpler than a switching DC/DC stage. You can segment a single-stage design with a relatively simple linear ripple-suppressor circuit such as the one depicted in Fig. 5.

The modified-single-stage topology utilizes a uniquely designed linear regulator, which can greatly reduce the output current ripple from single-stage PFC constant-current output with only 2–3% efficiency loss. The approach offers additional benefits. Adding a switching DC/DC stage to a driver handicaps the EMC performance in most of the cases, while adding the linear regulator does not. The better EMC performance allows the ripple suppressor to be flexibly utilized with existing single-stage LED drivers by SSL manufacturers. The addition of the circuit to the output is far more cost effective than buying another driver or switching DC/DC converter to get much better light output.

This ripple suppressor circuit is used in series with the single-stage PFC output and is primarily composed by a power MOSFET, a current sensing resistor, and an error amplifier. The sensing resistor gets the current ripple signal and if the ripple is larger than the set value, the output of the error amplifier adjusts the voltage on the MOSFET so as to make the ripple smaller. Those simple components can be encapsulated into a very small package or even made into an integrated circuit.

FIG. 6. Single-stage driver output current.

Driver test results

Consider tests on the ripple suppressor circuit that demonstrate the value. Fig. 6 depicts the voltage and current output of a typical single-stage driver. The 42W design is delivering 700 mA into an LED load. The ripple is very apparent in the current waveform.

FIG. 7. Driver output current with the ripple suppressor circuitry.

Fig. 7 depicts the output waveforms from the same 42W driver with the ripple suppressor added to the output. You can see how the suppressor effectively clips the ripple amplitude to an acceptable level, thereby eliminating the flicker through reduction of both the flicker percentage and index.

As the LED lighting industry develops, features like high efficiency and long life alone cannot satisfy the market. People are looking for a better lighting environment, especially when it is related to health. For certain places like offices and living rooms, elimination of strobe flicker is even more important.

There are multiple ways to create good DC current with low ripple to drive LEDs; each method has advantages and drawbacks. The key advantage of the ripple suppressor is that it provides a very simple and flexible way to reduce the flicker of the design we already have at a minimal and very reasonable cost.

<Source : http://www.ledsmagazine.com>