Why not direct AC drive your LED string?

publish: 2016-07-27 Source: http://en.spartaled.com
更多 0

Why not direct AC drive your LED string?


The whys and wherefores of AC-direct driving of LED lighting systems explained by Francois Mirand


True to the history of other types of semiconductor products, LEDs offer better performance at a lower cost with every passing year.



So much so, in fact, that the other elements of a conventional LED lighting fixture – the housing, connectors, PCB, optics, and most of all, the power supply – have assumed a large share of the total system cost which five years ago would have been dominated by the cost of the LEDs alone.


This has driven attempts to create a far cheaper architecture, dramatically reducing the cost of the power supply. In luminaire types such as low-end downlighters, linear fixtures and bulkheads, low cost is the most important attribute to buyers. Here, the concept of ‘AC-direct LED systems’ has been developed to eliminate entirely the need for a conventional (but bulky and expensive) AC-DC switch-mode power supply (SMPS).


The first implementations of AC-direct LED systems, however, gave their designers tricky design problems to solve, and severely limited their freedom to optimise operational features and the luminaire’s form factor.

Now an improved architecture for AC-direct LED systems appears to offer both competitive performance and design flexibility. But does it come at an acceptable cost?


TAP1506_AC-direct LED systems_Fig 1



Figure 1: a simple AC-direct implementation with four LEDs


Why design out the SMPS?A conventional high-brightness LED is a low-voltage device, typically having a forward voltage of around 3V. Mains power is supplied at approximately 120V or 240V, depending on which part of the world the user is in.


A luminaire therefore requires a way to step the AC mains voltage down to the DC forward voltage of an LED or string of LEDs, typically at below 60V.


The most common way to do this to date has been with an SMPS. The SMPS’s chief advantage is its excellent efficiency, sometimes exceeding 90%, combined with its ability to provide electrical isolation (for safety purposes).


Its efficiency is an attribute of the switch-mode architecture, which saves energy in, and discharges energy from, inductors and capacitors at very high speed, thus storing power at the peak of the AC voltage swing rather than dissipating it, as a simple linear regulator does.


Unfortunately, an SMPS requires bulky inductors or transformers and capacitors. What is more, the electrolytic capacitor type used in power supplies has a much more limited lifespan than the other components – it is commonly the first component in an LED lighting system to fail. EMI counter-measures are also required, because SMPS normally switch at high frequencies of 20kHz or more.


And an SMPS cannot be mounted easily on the same PCB as the LEDs. A metal-core PCB’s (MCPCB) single-sided copper construction makes SMPS circuit routing very difficult, and the passive devices required tend not to be available in surface-mount packages. FR4 (the most common PCB substrate type) offers more routing layers, but isolation and thermal management may become complicated with this type of PCB. In any case, the large passive components in an SMPS will create shadows if placed too close to the LEDs.


An SMPS, then, is bulky, creates some design difficulties, and is relatively expensive both in its bill-of-materials and assembly costs.


A simpler approach is to have no stepping down


This gave manufacturers of power-regulation ICs the incentive to develop a simpler architecture which drives high-voltage LEDs directly from a mains voltage without stepping this voltage down.

The AC-direct regulation scheme manages to achieve this by switching on the LEDs in sequence following the mains sinusoidal voltage. Suitable LEDs for this application include the LUXEON 3535 HV from Lumileds, as well as the 5630 HV, 5250 HV and 3030N HV series from LG Innotek.


The simplest implementation of the concept is shown schematically in Figures 1 and 2.



Figure 2: the AC-direct regulation IC progressively switches on each LED until all are illuminated at the mains voltage’s peak.


The AC mains voltage is rectified (so the negative half of the cycle becomes positive). This rectified output is a 100Hz half sine wave, typically swinging between 0V and around 325V if the nominal mains voltage is 230V.


In each half-cycle, the first high-voltage LED (or LED string) is switched on once the voltage has reached around 70V, the second is added at 140V, the third at 210V. All four LEDs are illuminated for as long as the voltage is at or above 280V.


This regulation scheme means that there is excess voltage which must be dissipated through a linear regulator, generating waste heat.


Such power losses necessarily reduce system efficiency when compared to an SMPS solution. But in practice, such an AC-direct LED design can achieve total system efficacy of as much as 100lm/W, compared to 130lm/W for an equivalent fixture using an SMPS.


In a simple, low-cost application such as a 600lm downlighter, perhaps a more instructive comparison is with the equivalent dichroic halogen bulb, which offers typical system efficacy of around 10lm/W: a huge reduction in power consumption is possible if such a bulb is replaced by an AC-direct high-voltage LED lamp.

And compared to the SMPS equivalent, the AC-direct design is much smaller, much simpler, and much easier to assemble, since it requires just a single PCB (see Figure 3).


TAP1506_AC-direct LED systems_Fig 3b



Figure 3: an LED bulb using an AC-direct regulation scheme







Nevertheless, in addition to reduced efficacy, the simple form of AC-direct LED regulation has other drawbacks which assume considerable importance in some applications.











One is heat. As Figure 1 shows, current regulation and power dissipation are all concentrated in a single IC. In normal operation, this makes the IC an extreme ‘hot spot’. This calls for special counter-measures to avoid the risk of over-temperature damage, including the use of an MCPCB and a board layout with wide gaps between other components and the regulation IC.This centralised form of AC-direct regulation also lacks flexibility in terms of functionality and performance optimisation, as well as component selection. This is evident, for instance, in the constraints on board layout: the single regulator IC occupies a large board footprint, and is therefore difficult to accommodate in linear formats such as strip lights and fluorescent tube replacements. The scheme is also limited in its flexibility to implement functions such as


These drawbacks have led to the development of an enhanced version of the AC-direct LED regulation scheme: rather than the centralised implementation shown in Figure 1, power IC manufacturer Exar (through its iML subsidiary) has developed a distributed architecture. (Figure 4)












The principle of operation is the same as in a centralised control scheme: high-voltage LEDs are switched on and off in a sequence during each mains half-cycle. But in the distributed control scheme, a separate IC such as the iML8684 regulates each high-voltage LED string.












Balancing cost and efficacy, most implementations adopt a three-step regulation scheme (with three HV LED strings), but two- or four-step schemes, or even schemes with more than four steps, are also possible.

The prime advantage of the distributed control scheme is its design flexibility. The flexibility of the circuit topology, for instance, allows the system designer to accommodate various LED arrangements, and to achieve the best balance between performance, functionality and cost.

The distributed approach allows the selection of virtually any LED, both the classic low-voltage types as well as multi-junction high-voltage LEDs. This gives the designer a much wider choice of package styles and of performance specifications (flux, efficacy, colour temperature and colour rendering).

It also supports compact implementations using a small number of high-flux LEDs, as well as designs with a more diffuse light output using a higher number of lower-flux LEDs.

In addition, a reduction in flicker and phase-dimmer compatibility can both be obtained with the balanced circuit structure supported by the distributed architecture or by adding a single capacitor, and it is easy to add a bleeder for better compatibility with triac dimmer schemes.





The flexibility of the distributed architecture for AC-direct LED systems is also realised through the increased freedom to choose PCB type, form factor and layout.




TAP1506_AC-direct LED systems_Fig 4



Fig. 4: three-step regulation, with three regulator ICs and a ballast transistor, is the most common way to implement distributed AC-direct regulation.




This is because the distributed architecture can readily be implemented entirely with surface-mount components, and its simple electrical circuit can relatively easily be laid out on a single-sided PCB. In the distributed architecture, heat is dissipated more widely by a larger number of driver ICs, eliminating the single hot spot found in conventional AC-direct LED systems.



This means that distributed schemes can often use an FR4 PCB, while achieving a more compact layout with smaller spacing between components. In addition, the small, low-profile components used in distributed schemes permit the designer to realise fixtures in any form factor, including linear strips and uniform diffuse light engines free of dark shadows.



The distributed scheme also offers other advantages over centralised regulation. The Exar topology has inherent high surge (transient voltage) immunity of >750V. In a centralised scheme, by contrast, surge immunity requires the addition of a large, discrete Metal-Oxide Varistor (MOV). An MOV is prone to long-term reliability problems as it has a finite ability to absorb the energy of the transient voltage surges that are common on AC mains. To make matters worse, these varistors are not always available in surface-mount versions, and are expensive.


Concerns over flicker



In both centralised and distributed implementations of AC-direct LED regulation, lighting designers have expressed some concern over the issue of flicker. In both schemes, the LEDs go dark every mains half-cycle, (that is, every 1/100th of a second in a 50Hz mains system), when the voltage is at or near to 0V.




In fact, the flicker occurs for such a short interval that in most applications it will not be perceptible by any user. In the low-cost applications for which AC-direct regulation is intended, the low level of flicker will tend to be tolerable.




Adopting a balanced circuit topology, however, and adding a small reservoir capacitor can both mitigate flicker as well as improve efficacy and LED usage, thus balancing part of the extra bill-of-materials cost of the distributed architecture.



Distributed scheme: cost/performance trade-off
The distributed version of AC-direct regulation pioneered and patented by Exar offers the designer a unique freedom to achieve design optimisation, as well as giving a large degree of control over the trade-offs between performance, functionality and cost.The simplest implementation of the distributed solution tends to be cheaper than the centralised approach, while more sophisticated designs offer the opportunity to add valuable extra features. For example, functions such as presence detection or DALI dimming control may both be accommodated in a distributed AC-direct LED system.Every application will have its own balance of performance and cost requirements, but the unparalleled design flexibility of the distributed approach merits consideration by system designers who want to implement AC-direct regulation in a new generation of low-cost LED fixtures.