newsletter link
mondo arc

Life after the bulb

Issue 51 Oct / Nov 2009

Dr Geoff Archenhold looks at LED drivers and the latest advances in LED technology that will make solid state lighting solutions cost effective enough to replace the CFL.

Unless you are a recluse you will have heard from virtually all media channels that the big one has finally happened. Yes, the light bulb invented (and patented in 1878) in the UK by Joseph Swan, is being phased out from all EU countries starting with the rather large and energy binging 100W+ incandescent light bulbs. But what will happen next to the lighting industry?

Of course, as this is a Solid-State Lighting column there can only be one feasible solution... LEDs!

Light Emitting Diodes and, within a decade, Organic LEDs will be the ONLY short to medium term sustainable answer to the replacement of the incandescent light bulb and unfortunately many of the global policy makers are going down a cul-de-sac by suggesting and recommending to consumers that CFLs are the answer. In past articles I have provided evidence as to why CFLs are such a poor lighting solution today in terms of functionality, quality of light, energy efficiency, sustainability and user acceptance. However, LED technology has rapidly come of age and the very latest products such as the new LED Light bulb replacement from Panasonic offers all of the advantages of the light bulb but in an energy efficient and highly sustainable form.

In South Korea, the government has taken a leadership role by setting the fledgling lighting industry to focus on LED technology through the promise that by 2013 at least 30% of the total lighting stock will be LED. This statement has amassed significant industry support and there are several substantial organisations not currently associated with the lighting industry developing global lighting strategies based on energy efficient products.

The lighting industry is going through a rapid pace of change as I stated over two years ago in this very column. LEDs will be common place in every application by 2012 and my latest market and technology analysis shows that this ‘digital lighting switch-over’ is still on track to be achieved.

This month’s article will start with a roundup of new technologies introduced to the market followed by a detailed discussion looking at LED system reliability and efficiency.

Advances in Core LED technology
LED technology is being advanced continuously enabling ever brighter, more efficient and lower cost LEDs to enter the marketplace. An interesting technology being researched by several groups around the world is that of creating white LEDs on a substrate (or plate) of silicon material. Most LEDs are fabricated by first depositing multiple GaN-based layers on top of a sapphire or silicon carbide (SiC) substrate. Silicon substrates offer potential cost advantages for LED manufacture especially as the majority of the electronics industry use silicon wafers for products such as microprocessors.

A new Chinese start-up company, Lattice Power Corporation, has demonstrated 1x1mm blue LED (453nm wavelength) grown on a 2-inch silicon wafer at a 350mA drive current that achieved greater than 458mW output power. The company has also demonstrated a 0.5x0.5mm LED that achieved a 74.8mW output power at 60mA drive current. By applying YAG phosphor, white LEDs made from 1x1mm chips have achieved 103lm at 350mA drive current, with a colour temperature of 5130K and CRI of 69 (see figure 1).

Latest LED Emitters
The last few months have seen a whole host of new LED emitter packages launched from a variety of manufacturers, all of them offering a variety of advantages, usually in higher LED powers or improved lumen performance.

The first LED is the new Acriche A4 from Seoul Semiconductor (see figure 2) who are now officially the sixth largest LED manufacturer in the world and climbing up the charts very quickly. The A4 has a luminous efficiency of 75lm/W, and is therefore capable of replacing warm-white incandescent and compact fluorescent light sources in many consumer and commercial applications. The A4 devices have a colour temperature of 3000K and efficiency of 75 lm/W, while also offering a high CRI of 85.

An advantage of the Acriche A4 series is the possibility to use Acriche products on AC mains power freely between 100V and 230V without additional conversion. Only a simple, inexpensive diode bridge is required.

Philips Lumileds has released a new LED, LUXEON® Altilon, specifically designed for the automotive sector. However, this may be a prelude to the development of a new range of LED array products for lighting similar to other LED companies such as CREE, Seoul Semiconductor, LEDEngin and Bridgelux to name but a few.

The Altilon, shown in figure 3, is made from 2 or 4 LED die and is capable of delivering more than 850 lumens at 1000mA (or 13.7W) with a colour temperature of 5600K in a body of 16.9mm x 15.9mm.
Epistar from Taiwan, has been developing an LED technology to reach high CRI and high efficacy simultaneously. They have demonstrated a white light LED with CRI of 80, CCT of 3000K with an efficacy equal to 80 lm/W. In 2009 Q4, Epistar expects to demonstrate a new product with CRI better than 80 and efficacy of 100 lm/W at 3000K.

The newly developed product, shown in figure 4, from Epistar includes one or several high-voltage monolithically integrated DC multiple-chip arrays combined with a bridge structure for AC drive. This new LED product will offer a significant leap in Warm White LED performance and enable very efficient incandescent bulb replacements to be manufactured.

Efficiency and reliability of LED systems

In the March 2009 column I briefly touched on the topic of reliability and mentioned that LED drivers are the weakest link in an LED fixture assuming the LED and thermal management have been taken care of properly. This month I will attempt to expand on what is meant by reliability and describe the limiting factors of driver efficiency and show how there is a trade-off between accuracy, efficiency and cost.

The definition of an LED driver refers to a complete LED driver subsystem enabling a user to attach high voltage AC mains in and is capable of driving a series of DC LEDs on one or more output channels using a variety of control techniques to obtain a desired lighting effect as shown in Figure 5.

The life of an LED driver is mainly determined by the lifetime of the electrolytic capacitors employed. Therefore, to achieve long life of the LED drivers, it is critical to select long-life, quality electrolytic capacitors (as discussed in March 2009 article). It is important to note that the life of electrolytic capacitors drops by 50% for every 10C increase in operating temperature and so thermal management of these components is extremely important.

Driver efficiency can affect reliability

As shown in table 1 the overall efficiency of the LED driver significantly falls as both stages drop in efficiency with anything less than 90% efficiencies in AC/DC and DC/DC stages resulting in significant losses in total driver efficiency eg <81%.

For LED driver PSUs with low wattages then the overall power efficiency of a driver may be managed. However, for larger wattages the inefficiencies are generated as heat within the driver and will therefore increase the temperature inside the driver.

Obviously the design of an LED driver becomes critical and the higher the efficiency of both stages the better. However, when one designs an LED driver the LED characteristics and input voltages varying during operation thus resulting in a difference between driver efficiency from design to operation. This variation becomes greater if the number of LEDs (or the forward voltage of the LED load) being driven changes significantly from the parameters used in the circuit design. 

The LED driver configuration used to generate the results in figure 6 is such that the maximum power on a channel is 50W and at 1A forward current the 12 white LEDs had a forward voltage of 36V providing approximately 36W.

Figure 6 demonstrates two important factors, the first is that as the LED forward voltages are varied the driver stage changes its efficiency and so, as the number of LEDs change from 12 to just 4 on a channel, we see the stages efficiency drop from 94% down to approximately 86%. The second factor demonstrates how the efficiency of the DC/DC stage varies for a fixed number of LED in a series chain but with reducing forward current through the LEDs. In the case of 12 LEDs, the efficiency varies from 94% at 1A down to 86% at 100mA. The efficiency curves are not perfectly linear as the forward voltage of the LEDs will vary as the current (or power) through the LED is changed and the sharp drop in efficiency shown between 200mA and 100mA is due to the sense resistor losses and the IC circuit’s own power consumption becoming dominant compared to the power of the LEDs. For example, at 100mA forward current the power through the 12 LEDs is only 3.6W.

The DC/DC driver used to obtain the efficiency figures reached 96% maximum efficiency when a full 50W of LED power was applied. The other 4% was lost due to the resistance in the sense resistor, MOSFET, short circuit protection and inductor. It is possible to increase the efficiency of the DC/DC stage still further however this usually results in much higher component costs so there is a trade-off between performance and price.

In figure 7 there are several key components that determine the efficiency of a DC/DC stage. These being:
• The sense resistor used to determine the actual current going through the LED chain and required for precise LED control.
• The resistance of the MOSFET when it is switched on which is termed RDS(on).
• A Short Circuit Protection resistance.
• The inductor used to power the LEDs when being switched.

As the various resistance values are in series the total resistance of the circuit when it is conducting or switched on is calculated as:
Total Resistance = Sense Resistor + MOSFET RDS(on) + Short Circuit Protection Resistance.

If the circuit in figure 7 has a forward current range between 350mA and 1A and is capable of driving 50W of LED power with 256 different intensity levels then the system efficiencies can be calculated and ‘tradeoffs’ in the design made.

If one examines the sense resistor component one would choose a resistance that is as small as possible in order to reduce the power dissipated across it and to maximise the efficiency of the stage. However, if the sense resistance is too small the LED current feedback circuit becomes very expensive as it has to measure and detect exceptionally small voltages.

In order to obtain an adequate sense resistor voltage of say 0.2V then one can use the equation Rsense = 0.2V / Forward current so for a forward current of 350mA one would require a sense resistor which is 0.57 Ohms and at a forward current of 1000mA one would need a sense resistor of 0.2 Ohms.

Therefore one can calculate the power loss due to the sense resistors by using the equation Ploss = Forward Current2 * Sense Resistor which at 350mA would be 0.07W and for 1000mA it would be 0.2W.

Another way of calculating the efficiency loss is to use the voltage drop across the sense resistor and compare this against the total forward voltage of the LEDs. For example, the loss in efficiency due to the sense resistors is 0.2V/48V or just over 0.4% for both forward currents.
If either the sense resistor is increased or the LED forward voltage is reduced the efficiency will change. For example, if one only had 1 LED in a chain then the efficiency loss due to the sense resistor can be calculated as 0.2V/3V or 6.6%. If one chose a sense resistor that meant the sense voltage was now 0.5V then with just one LED connected the lost efficiency would be 0.5V/3V or 16.6%. This demonstrates why the efficiency shown in figure 6 drops when different numbers of LEDs are placed on the DC/DC stage.

As sense resistors decrease in resistance value from 1 Ohm their costs and tolerances increase and so most sense resistor values are chosen between 0.5 and 1 ohm in value.

It is also important to note that in order to obtain very accurate forward current through the LEDs the current sensing system needs to be able to measure very small voltage at a minimum defined by the equation CurrentResolution = Vsense / number of Intensity Steps so for 0.2V and 256 steps the current measurement system needs to be able to measure a voltage of 0.2V/256 or 781µV.

Referring back to figure 7, the total resistance of all components in a typical circuit would be Rsense=0.5 Ohm, RDS(on)=0.15 Ohm and SCP Resistance=0.5 Ohm = 1.15 Ohm. Therefore, at a maximum forward current of 350mA the power loss would be 0.4025V/48V or approximately 1%. At 1 Amp the power loss is 1.15V/48V or approximately 2.4%.

Therefore, to obtain the highest efficiency it is important to utilise the best components possible at a budget that is acceptable for the design.
I would also suggest that if an end-user is offered an LED driver for £10 then the components used within it are not of a high specification so expect the efficiency to be low, performance poor and a finite lifetime. Another specification to look for on a driver datasheet is the maximum ambient operating temperature; the lower the temperature the poorer the driver efficiency is in general or the driver is using low quality components and the more likely it will fail. Most quality LED drivers have a maximum ambient operating temperature between 45ºC and 60ºC so anything less needs to be evaluated for lifetime and quality.

Failure rates of LED drivers
As stated in the March 2009 article there are several Achilles’ heals within LED drivers including electrolytic capacitors and components such as MOSFETs that can run close to their maximum temperature specifications however it is currently difficult to define the lifetime of an LED driver.

There are several methods for defining reliability that use very sophisticated mathematics and statistics however in practice it is common to find products that fail either straight away or after a significant period of use and this distribution is commonly referred to as a bathtub distribution. Figure 8 shows the reliability ‘bathtub curve’ which models the cradle to grave instantaneous failure rates vs. time.
If one follows the slope from the start to where it begins to flatten out this can be considered the first period. The first period is characterised by a decreasing failure rate. It is what occurs during the early life of a population of units. The weaker units fail leaving a population that is more rigorous. This first period is also called infant mortality period. The next period is the flat portion of the graph. It is called the normal life. Failures occur more in a random sequence during this time. It is difficult to predict which failure mode will manifest, but the rate of failures is predictable. The third period begins at the point where the slope begins to increase and extends to the end of the graph. This is what happens when units become old and begin to fail at an increasing rate.

Early Life Period
In order to reduce the number of units that fail in applications many of the quality LED driver manufacturers will include production techniques to identify the products that will fail during the early part of the product life. Many of the techniques used include: burn-in (to stress devices under constant operating conditions); power cycling (to stress devices under the surges of turn-on and turn-off); temperature cycling (to mechanically and electrically stress devices over the temperature extremes); vibration; testing at the thermal destruct limits; highly accelerated stress and life testing; etc.

There is always the risk that, although the most up to date techniques are used in design and manufacture, early failures will still occur.

Useful Life Period
As the product matures, the weaker units die off, the failure rate becomes nearly constant, and modules have entered what is considered the normal life period. This period is characterised by a relatively constant failure rate. The length of this period is referred to as the system life of a product or component. It is during this period of time that the lowest failure rate occurs. The useful life period is the most common time frame for making reliability predictions.

Wear-out Period
As components begin to fatigue or wear-out, failures occur at increasing rates. Wear-out in power supplies is usually caused by the breakdown of electrical components that are subject to physical wear and electrical and thermal stress. It is this area of the graph that the Mean Time Between Failure (MTBF) rates calculated in the useful life period no longer apply.

Mean Time Between Failures (MTBF) and Mean Time To Failure (MTTF)

Reliability is often quantified as MTBF (Mean Time Between Failures) for repairable products and MTTF (Mean Time To Failure) for non-repairable products. A correct understanding of MTBF is important. For example, a power supply with an MTBF of 40,000 hours does not mean that the power supply should last for an average of 40,000 hours. An MTBF of 40,000 hours, or 1 year for 1 module, becomes 40,000/2 for two modules and 40,000/4 for four modules. Sometimes failure rates are measured in percent failed per million hours of operation instead of MTBF. The FIT is equivalent to one failure per billion device hours, which is equivalent to a MTBF of 1,000,000,000 hours.

In another example, if 10,000 units operated in the field for 1000 hours with 10 failures, the MTBF would be 1 million hours. This does not suggest that any unit will be expected to operate for 114 years. Viewing MTBF from a different perspective, if a product is determined to have an MTBF of 250,000 hours and 1000 units are deployed in the field, on average a failure could be expected about every 10 days if the products are operated around the clock or about once a month if they are operated 8 hours per day.

The formula for calculating the MTBF is:
Ø = T/R.
Where: Ø = MTBF; T = total time;
R = number of failures.
MTTF stands for Mean Time To Failure. To distinguish between the two, the concept of suspensions must first be understood. In reliability calculations, a suspension occurs when a destructive test or observation has been completed without observing a failure.
MTBF calculations do not consider suspensions whereas MTTF does. MTTF is the number of total hours of service of all devices divided by the number of devices. It is only when all the parts fail with the same failure mode that MTBF converges to MTTF.
Y = T/N
Where: Y = MTTF; T = total time;
N = Number of units under test.

For example, suppose 10 devices are tested for 500 hours. During the test 2 failures occur.
The estimate of the MTBF is:
Ø = (10*500)/2 = 2,500 hours / failure.
Whereas for MTTF:
Ø = (10*500)/10 = 500 hours / failure.
If the MTBF is known, one can calculate the failure rate as the inverse of the MTBF. The
formula for (h) is:
h = 1/Ø = r/T
where r is the number of failures.

Driver lifetime
The lifetime of a driver indicates how long a product should be expected to survive under normal operating conditions. It is the period of time between starting to use the device and the beginning of the wear-out phase. This is determined by the life expectancy of components used in assembly of the unit. The weakest component with the shortest life expectancy determines the life of the whole driver. For power supplies, electrolytic capacitors typically have the shortest lifetime expectancy.

Power Factor and Power Factor Correction
Power Factor (abbreviated PF) is the ratio of real power to apparent power in an AC power system and is expressed as a number between 0 and 1. Real power is the actual power drawn by the load whereas apparent power is the product of the load current and load voltage. Since the voltage and current may be out of phase this product may be significantly greater than the real power.

Practically, a load with low power factor draws more current than a load with a high power factor for the same amount of useful power transferred. The higher currents increase the energy lost in the power distribution system, and require larger wires and other equipment. Because of the costs of larger equipment and wasted energy, electrical utilities will usually charge a higher cost to industrial or commercial customers where there is a low power factor.

The PF is very important when calculating the energy efficiency of a lighting fixture, for example if two luminaires both produces 1000 lumens output and consume say 10W of apparent power (ie; I * V) each then they would appear to produce the same efficacy of 100 lumens per watt. However, if the PF of each driver were different, for example 0.6 and 0.9 then the real power would be 10/0.6 or 16.6W for the first luminaire whilst the second luminaire would consume 10/0.9 or 11.1W resulting in a significant difference in luminous efficacy of 60.24 lm/w and 90.1 lm/w respectively. Typically, CFL lamps have a PF of 0.6 and quality LED drivers have a PFC of 0.9 resulting in considerable real power savings when using LED based products that have exactly the same lumens and apparent power specifications as that of CFL based systems.

PFC is the abbreviation for Power Factor Correction. In order to maintain a high power factor, switch mode power supplies (including LED drivers) must employ some form of power factor correction.

Most power drivers or ballasts with passive PFC can achieve power factor of about 0.7–0.75 however drivers with active PFC can reach up to 0.99 power factor, whilst a driver without any power factor correction has a power factor of only about 0.55–0.65.

To comply with current EU standard EN61000-3-2, all switched-mode power supplies with output power more than 75 W must include passive PFC, at least. 80 PLUS power supply certification requires a power factor of 0.9 or more.

80 PLUS power supply certification

An interesting voluntary cooperation within the US computer industry created the 80 PLUS initiative to promote more electrical energy efficient computer power supply units (PSU). It certifies products that have more than 80% energy efficiency at 20%, 50% and 100% of rated load, and a power factor of 0.9 or greater at 100% load. That is, PSUs that waste 20% or less electrical energy as heat at the specified load levels, thus reducing electricity use and bills compared to less efficient PSUs.

Having high standards within the LED driver community is essential for maintaining a high quality reputation of products and will undoubtedly reduce the prospects of disappointed end-users who were early adopters of LED technology that were low cost and poor quality.

Round up
From some of the recent breakthroughs in LED technology highlighted this month it is possible to deliver a cost effective LED lighting solution for most applications as the efficacy, quality and reliability are improving at a significant rate. This innovation in the LED lighting industry is continuing at a pace at all levels within the supply chain enabling a roadmap of improvements over the next few years which will result in high light output products that offer high CRI lighting at virtually all colour temperatures with ever decreasing production costs.
This innovation is exciting for the LED industry but great for the end user who will be able to use light in ever more intelligent ways to save energy. We have seen this month that end-users, such as the Holiday Inn, are able to choose LED lighting solutions and demonstrate significant electricity savings as well as reduce their carbon commitment that offers a great benefit for all.

We have seen that the power factor of LED drivers is an important parameter to determine the power consumed by lighting products and the reason why the majority of CFL technology is poor compared to high quality LED lighting systems even if the systems consume the same apparent power and emit the same amount of light.

I would like to leave you this month with the thought of how much work still has to be done within the LED industry in order to raise quality. I was recently at an event held that matched LED companies and end users from the public sector together and one of the companies told me they could buy a 20 or 30W LED driver for less than $5 from China and their aspirations were to manufacture an LED downlight for less than $20 complete within 12 months. After asking several questions about quality and sustainability it became very clear that these were not at the top of this company’s agenda which is of great concern.

I predict there is a real danger for the reputation of the LED industry and hence the wider lighting community when end users see LED lighting products that claimed 50,000 hours life start to fail in a few years time. Unfortunately, it may be too late by this time to repair the damage to the industry’s reputation and I urge the reputable companies to follow the lead of IT industry and form voluntary certifications outlining quality solutions that end-users can purchase with a high degree of confidence.

Unfortunately, as I stated to the LED manufacturer at the event, a 20W $5 LED driver is of course possible to achieve but a $5 high quality and reliable LED driver is NOT!

As a response to the quality agenda, I am working with the Lighting Association Test Laboratories and several LED Test and Measurement companies to host a Live testing event during next year’s LED & Light Fair exhibition taking place in Birmingham in June 2010. The event will demonstrate testing on all types of lighting and the best efficiency products will highlighted by a ‘Top Gear’ style highest efficiency table. The event will not only demonstrate how one should test luminaires for colour, CRI and efficiency but show the end user they can procure quality products. If you are interesting in learning more about the event or would like to discuss issues relating to the industry then please do not hesitate to contact me at

Finally, we should say Adios to the incandescent light bulb and wish it well in retirement as it has been a great ambassador to the lighting industry which is why it has been around for such a long time. I think we should warn the youthful pretender (‘the LED bulb’) that will start to hold the reins of the lighting industry that it’s not that easy achieving reliability and cost effectiveness together - not just yet anyway!

Geoff Archenhold is an adviser to the UK Government on LED technology and helps LED companies raise investment from the finance community. He is an investor in an LED driver company and an LED fixture company, a Lighting Energy Consultancy and euroLEDs Events LLP.

The views expressed in this article are entirely those of Geoff Archenhold and not necessarily those of mondo*arc.


Figure 1: Lattice Power’s Cool White LED performance from a GaN/Si 1mm x 1mm LED

  • Figure 2: The Achriche A4 - Warm White High CRI, AC LED

  • Figure 3: The 850 lumen 4-chip Luxeon Altilon White LED from Philips Lumileds

  • Figure 4: A sample packaged with newly developed chips from Epistar

  • Figure 5: Simple block diagram of an LED Driver system

  • Figure 6: How the DC/DC stage efficiency is reduced according to LED current and forward voltage

  • Figure 7: A typical DC/DC LED driver stage showing the key components

  • Figure 8: A typical reliability bathtub curve

  • Table 1: Overall drivver efficiency table when combining AC-DC and DC-DC stages

  • Table 2: 80 PLUS Certification levels of PSUs
Related Articles


Follow us on…

Follow Mondo Arc Magazine on Twitter Follow Mondo Arc Magazine on Facebook Follow Mondo Arc Magazine on Linked In

mondo arc india

darc awards DWLF IALD PLDC LRO