U.S. Department of Energy - Energy Efficiency and Renewable Energy

Building Technologies Office – Information Resources

Text-Alternative Version: Dimming LEDs Webcast

Below is the text-alternative version of the "Dimming LEDs" webcast, held December 10, 2012.

Kelly Gordon: We're very happy to have as our speaker today Michael Poplawski, also of Pacific Northwest National Laboratory. Michael joined PNNL in 2009 as a senior engineer following 12 years in the commercial semiconductor industry. His work experience includes stints with domestic and foreign component manufacturers ranging in size from startup to conglomerate and responsibilities spanning a variety of device, circuit and system engineering functions. His current efforts are focused on supporting the daily solid-state lighting market development programs primarily in the areas of technology evaluation and demonstration, standards development, and lighting energy and use consumption. Michael –

Michael Poplawski: Thank you Kelly. And are the slides up?

Kelly Gordon: Yes, I can see them.

Michael Poplawski: Okay. So LED dimming, what you need to know. Let's start off by asking with all the other things there are to learn about LED sources and technology, why worry about dimming? Well as many people know LED dimming delivers many other value propositions to LED including additional energy savings, increased visual task performance, enhanced ambience, etc. And unique to LEDs, it has the potential to improve light source efficacy and light time. So there are many reasons to be interested in dimming LED sources.

So what's the big deal or why this webinar? Well, as many early adopters and users of this technology have found out, dimming LED sources in the real world, at least today can be challenging particularly with phase-cut dimmers. There's wide variation in LED source and dimmer characteristics and little can be assumed. Not all product claims are equal and at the end of the day it can be difficult to predict whether the LED source you're installing will work well with the dimmer or dimming equipment you're using or intend to use.

So what do you need to know? Let's review the key points that we're trying to hit during this webinar right up front and see if we can enhance them or deliver upon them throughout the next hour. LEDs are inherently dimmable. Dimming is not a limitation or weakness of this technology. However LEDs typically need a driver and if you're not familiar with what that is we'll talk more about that in a second.

Dimming an LED source can change the behavior of this driver and at the end of the day dimming performance is determined by the driver's capability and compatibility with the dimming equipment. Multiple compatibility issues are rooted in circuit level interactions between the LED driver and dimmer and we'll talk briefly about some of those.

And finally what you think you know may no longer be valid and this is particularly relevant for people who are experienced dimming light sources but perhaps aren't experienced dimming LED sources. Some of the rules have changed.

Now there is good news. You can dim today if you want to. Good LED dimming solutions are available but with various tradeoffs. Furthermore, like the rest of this technology, less many other aspects of this technology, new standards in dimming or control technologies are in development and hitting the market regularly and user experiences should only improve as time goes by. Today though your chances for success are likely gonna be correlated with your willingness or ability to learn new things. This in some cases may mean learning about issues you're unfamiliar with while learning about new standards and technologies. Your chances for success may also be correlated with your willingness and ability to evaluate products firsthand. This is not a new phenomena, many people in the lighting field have found that that's still been or always been the best way to evaluate certain characteristics of light sources such as their ability to render colors or to – how much glare they produce for example.

So, if you want to dim LEDs today here's some things you're probably gonna have to figure out or develop skill at figuring out. What are your options? Where can you go for information and guidance? What questions should you ask when you are talking to a product manufacturer? What potential tradeoffs are important or not important to your lighting application? What's your risk tolerance for the installation or project that you're working on and how much are you willing to learn?

So speaking of learning let's start with a brief review of some electrical circuit fundamentals. Here we're talking about how to control current in simple resistive loads. Now resistive loads are noted by the fact that they have linear current voltage relationships as shown in these plots on the upper right. In particular the current resistive load is just one over the resistance times the voltage. Now when the input voltage is AC or a sinusoidal wave, what you really care about is the RMF voltage.

Now another feature of resistive loads is that there's a time independency. So Irms equals 1 over R times Vrms and there's no time component in that equation. Resistive loads are also bidirectional which means that you can apply voltage in either direction or polarity, either polarity to the load and you'll get the same current. Again, so rewriting the equation Irms is 1 over R times Vrms, no sign dependence.

Incandescent sources which are the most commonly dimmed sources behave electrically like simple resistive loads which is actually very important to their use and deployment and the history that users have dimming them. Incandescent sources effectively then only care about Vrms. Electrically they present a resistance that appears constant at steady state. This resistance is just a function of filament temperature. Incandescent sources are also then bidirectional.

Applying Vrms in either direction produces the same current and the same average light output. There's an important caveat that is again beneficial to incandescent sources known as thermal persistence. Now normally if the current goes to zero and a resistor there's no power consumption. There can't be. But when current goes to zero in an incandescent source, despite the fact that current is zero and there's no power consumption, light output continues as long as the filament is hot. This can be tens or hundreds of milliseconds.

So I already talked about simple loads. Well what are complex loads? Complex loads contain complex electronic devices, things like capacitors and inductors. Complex loads contain devices which store energy and which have nonlinear current voltage relationships. And finally, complex loads contain devices with time dependencies, so the voltage and currents have time components if you were to write out their equations. The two most common components that comprise complex loads are capacitors and inductors and you can see in the bottom figures that they both store energy.

LEDs are complex loads. They're nonlinear devices as you can see on the plot on the right, the current and voltage relationship is not a simple linear straight line. In fact they have different current voltage relationships in different regions of operation. In certain regions small changes in voltage can equal large changes in currents. That tends to be the on state or the right portion of the plot in the upper right hand corner. So because of this the average current must typically be controlled.

LEDs are unidirectional or current only flows in one direction, and therefore light output only occurs for forward current. Here's another caveat here which is important to how this technology performs as a light source. When current goes to zero in a diode there is indeed no power consumption but there will also be no light output. So you need to pay careful attention to any time where the current is zero in a light emitting diode.

So I already mentioned LEDs typically need what's called a driver. And the driver performs the function of controlling the current to the LED. So as you see in the simple block diagram up front, an AC power source that delivers 120 volt rms voltage to an LED source will actually be delivering that voltage to a – well here it's called the black box but what is a driver which then transfers the power being delivered to the driver to the LED's and at the same time controls current to an average constant value.

Now nonlinear Iled and Vled relationships together with manufacturing variation and the forward voltage drop of the LED means LEDs again are best regulated by controlling their current. And a driver is comprised of power electronics components that are used to create circuits which again convert this AC voltage into a regulated LED constant average current.

Now I already mentioned one of the key points LEDs are dimmable and they're actually very easy to dim at the component level. There are two primary methods that are used to adjust the light output from an LED. One is referred to as constant current reduction, also known as CCR or analog dimming. In this approach the LED current is always on or there's always current flowing through the LED but the set points or the control point for the driver is just adjusted to deliver different levels of light output. So if you look at an LED data sheet you'll always find a plot similar to the one in the upper right-hand corner that shows the relationship between the average forward current and the relative luminous flux. So you can see for this particular diode that 750 milliamps is its nominal operation state so that's 100% of light output but if I reduce the set point and I operate the LED closer to 400 milliamps, I'll get 60% of the light output.

This approach has the benefit of potentially increasing the LED lifetime because it operates the LED when you dim because it operates the LED at lower current and temperature. It typically doesn't generate noise and it has potentially higher efficacy at lower dimming levels as shown in the plots at the right. Again if you look in an LED data sheet it's not uncommon to find a plot like this that shows the relationship between luminous efficacy and average forward current. And you can see in this particular case as you reduce the current from the nominal operating state, the efficacy actually goes up from 1 to a little less than 1.2.

Constant current reduction does not create flicker in LED's and we'll talk a little bit about what that means if you're unfamiliar with that term. One of the potential caveats though to constant current reduction as a dimming technique is because as you dim you are operating the LED at different current levels. In some cases or some LEDs, this can result in color shift because there can be a chromaticity dependency of the LED on current. However, again, as the technology continues to evolve and change, this color shift may reduce and may vary in different makes and models of LEDs. So this is probably something at should not be considered a given, that this color shift isn't objectionable. And certainly what is objectionable can vary by application.

Another challenge that constant current reduction sometimes has is it can be more difficult to regulate at deep dimming levels because of course deep dimming levels mean low average currents. And as we'll talk about later, LED drivers typically cannot maintain identical operation over a wide range of input variations. And at some cases while they can be very stable at nominal current, as you continue to decrease the current that the driver is trying to regulate, at some point it may struggle to maintain that average value consistently.

Now pulse width modulation (PWM) is another technique used to dim LEDs. And in this approach, as you can see in the upper right-hand plot, the LED current or average current through the LED when the LED is on is always the same. But different dimming levels are achieved by reducing the current from some nominal value to zero typically with some duty cycle. Duty cycle is just this ratio of on to off time. So if again you look at the plots on the right there are two examples of a PWM waveform, one where the duty cycle is 50% or it's – where it's on 50% and off 50% of the time and the plot below it is 25% so it's on 25% of the time and off 75% of the time. Of course the average value just varies with the duty cycle. So lower duty cycles result in lower average value use.

This approach again has the potential to increase LED lifetime because the LED is on physically for less time and typically will run at lower temperature. This approach also can achieve very good dimming regulation at deep dimming levels because when the LED again is on, it's at a nominal typically a very stable current level for the driver. Typically this approach does not produce color shift but again there are different reports and examples of color shift appearing or not appearing in sources dimmed through these techniques so that should not be considered a given either way.

Now because of course you are – this approach requires switching of current through an LED, it has the potential to generate noise. And probably most importantly the PWM frequency is important so the frequency is related to the period if you will in time over which this duty cycle is varied. And as you can see in these plots here, the period is around eight, a little more than eight milliseconds. Now, the frequency is important because it can lead to – or too low of a frequency can lead to undesirable flicker. And too high of a frequency can make it difficult to achieve low dimming levels.

Now we just talked about LEDs, what about dimming technologies? Many people are familiar with a variety of dimming technologies that have been used in the industry for a long time. Some of which are shown on this slide here. You'll note that there are different high level wiring configurations of these technologies that actually will become important as we consider their suitability for LEDs. One way to categorize existing dimming technologies is shown here. And I'm particularly referring to the difference to whether the dimming technology creates the control signal coincidence with the input AC power as shown on the upper half of the slide or whether the control signal and AC power is delivered to the dimmer and luminaire separately.

There are two types of technologies that deliver coincident AC power control signals, phase-cut AC sine waves. Many people may be familiar two variations, forward or reverse phase. There's also a very important difference between two wire and three wire dimmers that are shown in the figure in the upper right-hand corner. A two wire dimmer only has two wires, one going into the dimmer and one out while a three wire dimmer has three wires going into and out the dimmer. The extra wire that makes the key difference is whether neutral is a neutral wire is connected to the dimmer. We'll talk more about why that can be important.

Separate AC power and control technologies include fluorescent 3-wire, 0-10V, DALI, DMX512, and PWM. Now because they maintain separation between the input AC power and the control signal, many potential compatibility issues are eliminated or reduced with this approach. So because it's somewhat simpler for that aspect, let's look at what happens when you try to dim some LED luminaires with one of these approaches, the 0-10V control technology. This plot shows the results or the light source output from three different LED luminaires operated with the same 0-10V dimmer. So you see the control input on the X axis and light source output on the Y axis. When you can see some variation here even though they're all operated with the same dimmer, different light sources start at different output levels and drop down to different output levels on the low end and the shape of the dimming curve can vary also.

Let's look deeper at these three luminaires. These next set of plots show normalized light output versus time. As you operate the luminaire first on a switch and then at five different dim settings, again with the same 0-10V dimmer. So these are relative light output plots. The X axis is time, the scale is the same for every plot. So you can see on a switch you're getting normalized level of one in terms of light output or 100% and as you proceed to reduce the control setting the light output drops accordingly. You'll see the light output versus time appears to not vary at all.

Now here's a different LED luminaire which again behaves the same way as you adjust the control and reduce its setting. The average value which is shown as the mean in the legend there continues to drop accordingly. However, the light output versus time is not flat or not consistent. There's some modulation. But you see it's a small level of modulation and it relatively stays the same. It reduces slightly the amount of modulation as you dim.

Now here is a third luminaire, again operated on the same 0-10V dimmer. And here you see evidence of what we talked about earlier, pulse width modulation dimming. When I start to dim the light source using the control setting, all of a sudden I see something, in this case it's gonna be the driver turning the current through the LEDs on and off at some frequency. And the duty cycle changes if you look at again the ratio between on and off time such that as it reduces you reduce the average value. And again you can see that in the mean values.

So let's talk for a second about this phenomenon, flicker. What is flicker? Flicker is simply the variation in time of light output or modulation of luminous flux. It's present in all traditional commercial electric light sources running on AC power including the ones we're all familiar with, incandescent, halogen, fluorescent, etc. It's typically periodic and a property of the light source and it's there whether you're aware of it or not.

Now here we're talking about, and the graphs we're depicting, photometric flicker which is not to be confused with electrical flicker whereby noise on an AC distribution line directly creates additional modulation on resistive loads because of their resistive nature. In this case the flicker is not property of the light source. Measurement and reporting of flicker is not a standard practice for commercially available light sources but because it does vary and can vary significantly across light sources, it may be something you want to take note of and ask about when you're evaluating and/or dimming an LED light source. In particular because as you saw in that last plot, a particular LED light source that may not flicker at all or have again, let's look back here, none, almost no modulation when operated at a switch can all of a sudden have modulation or flicker when operated on a dimmer.

Now we don't have time to go into much more detail about flicker but if you ask yourself who cares about flicker or why should I care, well certainly anyone who is sensitive to flicker cares about it. And certainly anyone who is responsible for human health, well-being and or performance in spaces with electric lighting will care about it too. There are different at-risk populations for specific impairments. The most well known impairment is photosensitive epilepsy which appears in about 1 in 4,000 individuals but also migraine suffers and not all at risk populations are identified. Young people and autistic people are also known to be in general more sensitive to flickering light sources. So let's continue on.

We noted in the previous example some variation in different properties of dim luminaries when operated on a 0-10V dimmer. Why does this occur? Well let's step back for a second and look at a block diagram for a system, again with separate AC power and control signal like a 0-10V dimmer and LED load. There's actually a number of things going on. A user is adjusting some input to set the dimming level. So that may be a dial, a slider, typing in some digital number.

So that input is going into the control and therefore then the control creates a control signal. So there's a transfer function. Right, they control input from the user and then the control output which is the control signal comes out. Whenever you have a transfer function, that function can be simple and linear as shown in the little icon below but it doesn't always have to be and it isn't. In fact it isn't for many products that are on the market.

Similarly when a light source receives a control signal, then it has to create light so it does some modulation or adjustment of its properties to create different levels of light output. So there's another transfer function. So that's sort of important to remember that there's two separate transfer functions that go from the input user to the output of the light source output or lumens.

So what about these transfer functions? Well in fact, if you look at equipment on the market, there is some variation in how for example a control processes the input from the user to the output signal. This plot shows kind of three idealized versions of commonly seen transfer functions. One of course is the simple linear relationship but there are also controls that produce what is called a square law relationship as shown in the yellow curve or an S curve relationship that is shown here in the blue curve. So different control inputs produce different control outputs based on the decision or the architecture or design of the control.

And here is an example that in fact shows three different manufacturers' controls, 0-10V controls in fact, where the output voltage was measured with respect to the user input. In this case, the adjustment of the slider. And you can see again the kind of idealized curves shown with the fat lines, but then the actual measurements shown with the measurement points. And in this case there's three different manufacturers whose 0-10V dimmer produces something pretty close to the three different idealized wave forms or control shapes that I described on the previous slide.

Now similarly LED source manufacturers in some cases target different transfer functions or the light source. We'll talk a little bit more about why both control and light source manufacturers are doing this but what's important is again because LED sources have a driver, something, again a black box that is – has the ability and in fact is typically necessary to regulate output current. But it can do so in a variety of different ways. In some cases manufacturers are exercising that capability in producing different transfer functions. So again here different lights, relative light source output for the same light source input. In this case, the 0-10V light source input.

In the example here, again measured on real products, you see that the product made by manufacturer A appears to have a square law type relationship while the product made by manufacturer B appears to have a almost linear relationship between 0-10V signal and light source output. So what happens when I put the two together? Because that's how we operate light sources in the real world. We want to just adjust the input control and see a change in output light.

While here is an example that shows a square law input control together with a linear LED light source. And the results, again control input on the Y axis, what we really care about, light source output on the X axis, looks pretty decent. This looks like an acceptable dimming curve. However if you're not careful, this can also happen. If you pair up a square law control with a square law light source, you might get a dimming curve that looks like this which would not be very useful or very satisfactory. And the important point here is to recognize that this combination may appear to not dim at all. It's almost down to no light output at 80% of the control signal and there are many potential reasons why that could be the case but here it's just one example of where it's just bad matching between the control device and the LED source.

So again, why these different transfer functions? Well manufacturers or lighting source developers are trying to take advantage or optimize the fact that we don't see light or perceive light in a linear fashion. And this plot here from the IES Lighting Handbook depicts that very clearly, that at 40%, that when you measure for example 20% light output or 20% footcandles for example, a user observing that light source will only have perceived a reduction to be 45%. So there's a difference between measured light from a light meter or luminance and perceived lights which is the visual interpretation and affected by adaptation and eye dilation.

So the key point here is when you're having a discussion with someone about dimming or a dimming level and someone says, "I'm 50% dimmed," you need to ask yourself or the person, what do you mean by that? Is that the dimmer position? Is that the change in energy consumption? Is that measured light or is that possibly even perceived light?

Okay, here I wanna take a word or a second to talk about another potential variation in LED sources as you dim them. I mentioned a number of them up front. Here we're talking about power quality. So I mentioned dimming an LED source can change the behavior of the driver or the set of power electronics that regulates current to the LEDs. The efficiency of the driver can actually degrade as you dim it but often it's offset by improving LED efficacy as we saw in those plots from LED data sheets. You've also seen an example of how flicker can be induced or increased by dimming. If you remember, the third luminaire controlled by the 0-10V signal when operated on a switch the driver was producing almost no flicker but when operated at different dimming levels it pulsed with modulation and created flicker.

So similarly power quality, at least as quantitatively evaluated by the power factor and total harmonic distortion metrics can also be degraded as you dim. The key word there in many of these cases is ten. These things don't always happen but they can happen. And I've already mentioned why, LED drivers typically cannot maintain consistent performance over a wide range of conditions. So their performance is gonna vary a bit over temperature. It can vary over connected load so how many LEDs are connected to the driver. It can also vary a bit over input voltage.

So this is probably most relevant to outdoor luminaires which often have universal drivers or universal or multiple input voltages. And again the driver typically won't operate the same at 120V for example as it may at 240V. Now this driver performance of course, like everything else, varies with technology and cost. So it's again not a consistent rule. Some drivers perform very well in a particular metric across a certain range and others may perform less well and there may be tradeoffs.

So here this begs the question for another short sidebar about what is power quality. We don't have time to go into this in great detail but real briefly, typically people are talking about displacements and distortions to voltage in current waveforms. I already mentioned that common metrics for power quality or power factor and total harmonic distortion. These measurements if you will, they're metrics for measurements. Power factors simply relates the active power to the apparent power for a given load. Now it's important to note that lower power factor loads do not themselves consume more energy, i.e. the load doesn't consume more energy but they do draw more RMS current and that can be important.

Total harmonic distortion, it's important to know that there's actually two types. There can be distortion of the voltage or the current waveform. Voltage waveform distortions are typically created by electricity generators while current waveform distortions are typically created by loads. And then you see some standard, somewhat standard specifications when necessary for both voltage THD and current THD.

So who cares about power quality? Well certainly electric producers and consumers may care and typically it's because of these increased current requirements, because transporting additional current can result in additional resistive losses. Now nationally or if you look across the grid, electricity transports on average result in about 10% loss from the generating source to the end load. So losses in general aren't huge but increased current can reduce and slightly increase transport losses.

Of course the increased current can also have ramifications for the infrastructure carrying that current. You may need bigger wires, circuit breakers and transformers with higher ratings, etc. Unfortunately this is a, these increased current requirements are not simple to calculate or identify the system issue.

Another potential ramification to power quality is that it can in some cases, again can, lead to electronic equipment damage or degraded performance but these things are not common. Now lighting equipment manufacturers of course care about power quality also. They in fact have worked together with industry including electricity producers to create voluntary requirements for lighting equipment that's in ANSI82.77, most recently published in 2002. Due to the rapid deployment of LED technology and in many ways it varies as we've been discussing here, this document is currently under revision.

Now like many things there can be system design tradeoffs for some LED sources for additional or increased power quality requirements. And there can be cost and size constraints for some LED sources also. So it should be taken into account as always when creating specifications because just asking for very high performance in this case may create a tradeoff in something you care more about.

So we've talked about 0-10V dimming and looked at some LED luminaire examples but what about incandescent dimming? This is the technology that is most widely used or most widely dimmed in the current infrastructure. How is it dimmed? Well, as I already mentioned, incandescents are typically dimmed with coincident AC power and control signal approaches such as the phase-cut AC sine wave or reduced amplitude AC sine wave approach. And the control block diagram that we looked at previously in this case looks a little bit simpler as shown below because power and control are delivered to the light source.

Now what's the difference between phase-cut and sine wave dimming? Well phase-cut is by far the most commonly deployed of all dimming technologies and in fact it has the largest installed base which NEMA estimates at over 150 million. The difference between the two is seen in the plot in the upper right-hand corner where the phase-cut dimmer removes a portion of the AC sine wave to embed the control signal. A sine wave dimmer reduces the amplitude, here you see nominally from one down to .5 or a 50% control signal. It reduces the amplitude of the input voltage.

But equally to how common and widely deployed phase-cut dimmers are, sine wave dimmers are very, very rare and the reason may be seen to some degree in the images at the bottom of the slide. The phase-cut dimmer is typically, it looks – or it can be found in the typical wall box that we all have, probably a couple of in our house and a sine wave dimmer, again while they still exist in the market are typically used for high end applications and look like something that you might install in a rack system.

So what's really important about phase-cut dimming is to note that it was designed for incandescent sources. And really it takes advantage of the feature of resistive loads that I mentioned previously but they only care about RMS. So if you look at the little cartoon figure here, I can create a input voltage wave form with any shape I want if the RMS voltage is the same I'm gonna get the same average light output from an incandescent source. So in fact this is what a phase-cut dimmer does. It's basically a high performance inexpensive Vrms adjuster and here you see in the plot, 120V in, 50% phase-cut roughly results in 60V out. I get about 50% light output.

The dimmer determines the dimming performance almost all by itself because there's very little variation in incandescent lighting sources. Now here you see some plots or dimming curves if you will of a single incandescent source operated with different phase-cut dimmers. And what's important to note here is that the dimming range is pretty similar. Again the property of the light source, its ability to dim, is not affected to a large degree by the dimmer. But what is affected is the shape of the curve and we've already seen why, because different control manufacturers create different transfer functions for their dimming control.

Another interesting point about incandescent sources and phase-cut dimmers is typically the efficacy of the incandescent source actually is reduced and can be reduced significantly as it's dimmed, as shown in the plots here. So the relative efficacy can drop even much below 20% as you dim to very low levels. Now what if I wanna use this most commonly deployed dimming technology, phase-cut dimming with LED light sources? While again as we already talked about, because LEDs are nonlinear loads, if I were to just put a different RMS voltage waveform shape, straight into an LED, I would definitely get different average light outputs for different shapes. And this is again the reason why you need a driver, or again shown here as the black box, something that processes the input power and controls current to the LED.

So if you're gonna use a phase-cut dimmer, you're again using it to basically create a control signal, right, which is the RMS or the conduction angle and then you're relying upon the driver to interpret that control signal and adjust the output of the LED, again probably through CCR or PWM approach appropriately. But as we already mentioned as one of our key points, the capability and compatibility of the driver determines dimming performance for LEDs. So here is again some examples.

This is a single LED light source, again operated on a variety of phase-cut dimmers. And here you see that both dimming range and curves can vary significantly and much more than you would see traditionally with the incandescent sources that these dimmers were designed for. Now on the other hand, typically LED's maintain their efficacy when dim so here are these same – the same source operated on different dimmers typically maintains its efficacy all the way down to the low end dimming level.

So what about flicker? So here are some examples of an LED lamp operated on a phase-cut dimmer, similar plots to the ones we looked at previously. And you can see again as you operate the control signal and dim the lamp, the initial modulation in the lamp starts to change and at some point you see the lamp is trying to employ pulse-width modulation to dim itself. So if I look at this same lamp on a different dimmer, I see the same general behavior. The lamp is again using pulse-width modulation to dim itself but at the same dimmer setting I'm getting slightly different output and slightly different wave form shapes. So the flicker is different or is affected by which dimmer I'm using to dim this lamp. Go back and forth again real quick so you can see this on the bottom half of the plots.

So what's the big deal again? Well as we've seen through many of the examples here, dimming LED sources can be challenging in particular with phase-cut dimmers. While there's some variation that can appear in dimmed LED light source output that's just due to the pairing of the control with the light source, there are additional set up or different things that can happen when operated on a phase-cut dimmer. At the end of the day this is all due to again, wide variation and LED source and dimmer characteristics. This is dimmer characteristics is really, the variation is we're talking about phase type dimmers here much more so than 0-10V in the other dimming technologies.

And we mentioned little can be assumed. Not all claims are equal but at the end of the day the problem is that it can be difficult to predict. And what you wanna predict is what's the performance gonna be and you assume or would like to not have to worry about compatibility and the line between what's performance and compatibility can be kind of fuzzy but we've already talked about some performance things that can vary, things like dimming range and curve or efficacy or the introduction of flicker and we talked briefly about power quality. But with phase-cut dimmers there's a whole other set of things that can happen or issues that really are categorized as compatibility issues. And as I mentioned previously with phase-cuts, the performance of the dimmer light source solution is really dependent upon what the driver is capable of achieving and how compatible it is with the phase-cut dimmer you've paired it with.

So to reiterate, the behavior of an LED source controlled by a phase-cut dimmer is a function of at least three things, the characteristics of the source, the number and type of light sources on the circuit and the characteristics of the dimmer. And from the previous list you saw that there are many types of behavior variation and many sources of behavior variation. This behavior variation spans compatibility, performance, and interoperability and it can be significant in magnitude. Today unfortunately it is by in large only predictable via circuit level testing because there are no standard definitions or test procedures for evaluating dimming behavior in general let alone phase-cut dimming behavior.

So let's talk about, briefly about what these compatibility issues are or how you might recognize them when trying to dim an LED source typically again with the phase-cut dimmer. There can be dead travel where you adjust the dimmer setting without a corresponding change in light level. Pop on is where the – it's a situation where the dimmer setting needs to be raised above its existing level in order to get light output at turn on. So say you dimmed the light down to 20%, you turned it off, now you want to turn it back on, some light sources may not turn on at that low setting. You may have to raise the setting to get the light source to come on. Drop out can occur when there's no light output at the bottom of the dimming range and that was probably evident from some of the dimming curves I showed you previously.

Popcorn is a term that some people use to talk about or refer to when different turn on – when there are different turn on times for different light sources on a dim circuit. So when you turn the whole circuit on you see different light sources popping on at different times. Flashing and ghosting occur when the light source is either intermittently on when it should be off or is just on at a steady low level when it should again be off, when you expect when you turn the control on the dimmer to be off, to the off position, that the light source is off. But in some cases with LED sources one of these two phenomena occur.

And additionally there can be audible noise, general inoperability, and premature failure of the dimmer. So why do these things happen? And again we really don't have time to go into them in great detail but we can talk about them at a high level by taking a brief look about how a phase-cut dimmer is constructed. So here I'm showing a kind of a high level block diagram that is relevant for both two wire and three wire dimmers. So you see here with the presence of a neutral wire but the dotted lines indicate that the neutral may or may not be connected.

So a phase-cut dimmer typically has RFI elements to control the amount of RF emissions. There's a switching element that affects when the phase is – or when the input voltage is connected to the output or disconnected to create the phase chop. There's a timing element which controls the switch effectively and then there's a box here for all the other advanced features that more and more modern dimmers have, things like night lights and radios and presets, etc. So again what's really important here is to note that some dimmers have a neutral and the presence of a neutral allows you to create a stable input voltage to all of these blocks, in particular the timing element or the advanced features. If you don't have a neutral wire however, accessible to the dimmer, than the operating voltage for these blocks simply has to be generated from the input hot and the dimmed hot. And when of course those two are connected, you know when the phase is not being chopped, that input voltage is near zero so it's difficult for those functions or it's impossible for those functions to operate during those time periods.

Now if you're not already aware it's important to note that two wire dimmers, i.e. without a neutral wire are by far the most commonly deployed dimmers in existing infrastructure. Unfortunately here again the presence of a neutral, because it provides a stable input voltage for these other features, greatly reduces the possibility for many of the compatibility challenges we already discussed.

So what are the sources of some of these issues? So here are a few that we'll just go through real briefly. In some cases the LED load may not be able to measure the Vrms or conduction angle presented by the dimmer because it may not have been expecting it. Again, this can vary from dimmer to dimmer and some tradeoffs may have been made in the design of the LED load that restricted its ability to determine or measure either of these values and therefore it can't tell what the control signal input is.

In some cases an LED load may not draw enough current to keep the dimmer switching elements closed, leading to erratic behavior. Similarly an LED load in a two wire dimmer is definitely in series with the dimmer and it creates a series impedance which in some cases can disrupt the dimmer timing element, again leading to erratic behavior, things like flicker, ghosting, flashing.

Another potential issue is when the LED load in the off state does not pass dimmer current in the manner which keeps the dimmer advanced features functioning. Now this again is a challenge that happens in a two wire dimmer. Because in order to keep those boxes, other boxes functioning, an input voltage needs to be supplied but also current needs to flow and when there's no neutral wire, the only way, only place current can flow is through the light source or through the load. And some LEDs again are not designed to handle different types or amounts of current in the off state. And this can again lead to erratic behavior.

And finally, LED loads can draw currents which create stresses on dimmers above and beyond what it's rated, again, incandescent wattage indicates. Again phase-cut dimmers were designed for incandescent sources which are much more consistent, much more simple electrically and therefore a rating, a dimmer rating is gonna be valid for all incandescent sources but LED sources vary and the rating is typically not going to be the same and it can vary.

So you saw the word load used in many of those potential issues and here one of the most significant ramifications of this is when trying to dim LEDs with phase-cut dimmers today in the real world is that the loading rules have changed and here's an example what I mentioned at the beginning that what you think you may know is not true anymore. So if you're used to dimming again incandescent source on an incandescent dimmer, so you want to put 60 watt sources on a 600 watt dimmer, you can do some simple math and know that you can put one to 10 lamps on that dimmer. And similarly if you want to put a 50 watt halogen on an ELV 600 watt ELV dimmer you'll know that you can put one to 12 of those lamps on the dimmer. However the same math does not work for LEDs and here in this particular example you see that the minimum number of lamps and the maximum number of lamps can be different from what you might expect from doing the simple math using the power consumption or energy consumption.

So trying to wrap up here, well, what's the big deal again? Let's review. We've seen how dimming LED sources can be challenging, particularly with phase-cut dimmers and this is again, due to the wide variation in LED source and dimmer characteristics. Little can be assumed as historical practices are unreliable. I've already mentioned how a 600 watt maximum load on a dimmer may not necessarily be useful guidance when trying to figure out how many LED sources you can put on that dimmer. On the other hand you know we can be familiar for certain light sources, familiar with using guidance either direct or indirectly such as it works with all ELV dimmers the same type of categorization as typically not true for LED sources. And I've already mentioned that claims aren't equal in part because of a lack of standard criteria and results can be difficult to predict given the lack of standard test procedures.

So what do you need to know? Again LEDs are inherently dimmable but they need a driver typically and dimming an LED source can change the behavior of this driver. We've talked about a number of reasons why. Probably the most key point is that dimming performance is determined by the driver capability and compatibility with the dimming equipment and compatibility is mostly an issue with phase-cut dimmers. Multiple compatibility issues are rooted in circuit level interactions between the LED driver and dimmer and we just talked about a few of those real quickly but to talk about them in more detail would take significantly more time.

So what are the recommendations for dimming LEDs today? Well first of all you should know your options. Are you trying to dim for example a lamp or a luminaire? Obviously a lamp is gonna be typically used in the retrofit application installed in a standard base and it has an integral nonreplaceable driver. If you're going into existing infrastructure, you're probably constrained to phase control. And lamps in general with standard bases are typically constrained to phase control. However luminaires often have driver options because the driver can be changed or can be specified. And driver options can lead control options so you may have the ability to decide whether you wanna use phase control or one of the other control technologies.

It may be worth your while to consider control technologies which separate AC power from the control signal because this reduces many of the compatibility issues we've discussed. If you must use a phase-cut dimming, if you can install or if you have access to a phase-cut dimmer with a neutral, again many of the compatibility issues can be reduced or eliminated.

And finally it's really becoming better and better advice if you will to consider using dimming controls designed for LED sources or take a look at new dimming technologies and we'll talk about a few examples of them in a minute. So next recommendation, take advantage of available information and guidance. Don't make assumptions, you can't make assumptions about LED sources if you're trying to dim one, you should try to determine what was this source designed to do? What does the manufacturer tell me? What does the cut sheet tell me? It should give me the dimming range, how low can this source go. Again it's going to vary with LEDs. You can't assume sub 1% for example like you can with incandescent sources. And there may be assumptions or requirements that are tied to the specification.

Next, is there a recommended dimming control selection guidance? If so, definitely, definitely use it. More and more manufacturers are providing such guidance. The guidance however should be specific. It should communicate specific makes and models of dimmer or control types and again, traditional or type of category guidance, I just use a forward or phase-cut dimmer, reverse phase-cut dimmer is likely not sufficient or likely not gonna yield uniform results.

Guidance should also include dimmer loading requirements because as we've seen, the min and max number of LED sources per control can vary and be specific to the LED source and control that you're interested in using. And even when you see guidance, beware expectations of exactly the same performance from any and all guidance combinations. Again because there's no standards, different manufacturers will use different or have different tolerances for the expectations of their guidance.

So here are a couple examples of manufacturer guidance provided here in this case from a dimmer manufacturer so you can see a report card for a particular LED luminaire communicating what the recommended dimmers are and again all the key information, how many in this case, fixtures can you put on these different dimmers, what's the light output range, are there any caveats, restrictions, etc. And on the right you can see a light bulb compatibility chart also. Lamp manufacturers also are providing guidance. Here's an example of two dimming compatibility tables for some replacement lamps where again the make and model of the recommended dimmers are listed. And again the dimming level in the top and whether in this case it even communicates whether the lamp flickers and how flicker, the amount of flicker – not the amount but the presence of flicker may vary with different amounts of lamps. Again because these performance criteria aren't standardized, one manufacturer's discernment of whether a lamp is flickering or not may vary from another manufacturer but you should always look for guidance to help weed out good and bad cases.

One potential recommendation is to ask for standard dimming guidance. You've seen in these examples on the previous two slides that different manufacturers are providing different types of guidance and again maybe using different criteria to evaluate products that they're recommending to you. There is I think room for improvement in the industry communicating this guidance and the LED Lighting Facts program has at least created the opportunity for a similar communication format. So here you see the required format for presenting dimming information and if you want your product to be listed in Lighting Facts as having dimming information. So again for users if you were to request manufacturers that you work with to include their dimming information in Lighting Facts you would have one source to go to for obtaining dimming information and a format for information would be consistent.

Okay, next recommendation, learn how to ask the right questions. And I think we've tried to talk about a number of these questions, a number of these issues that may not appear on a data sheet or may not be used to thinking about, things like what are the dimming transfer functions. Does the LED driver implement CCR or PWM to dim? If PWM, what's the dimming frequency knowing that higher frequencies produce typically less objectionable flicker? Is it a low voltage source? Well let's talk compatibility again because not only does my source need to be compatible with the dimmer, now I need step down transformer guidance also.

I've already mentioned, does the source have a universal or multiple input voltage? In this case I might want to ask what the dimming performance varies at different voltages. Or was the LED source evaluated for flicker over the dimming range? And that all input voltages for sources that can operate from different input voltages, and similarly for power quality. There are things that can be important to know about the dimmer in an installed application. Does the dimmer require a neutral or have access to a neutral? If it's a new dimmer does it have a trim adjustment? We'll talk about that in a second.

And finally, weigh the trade-offs. This is not a new topic but I think it's important here to weight your application needs versus wants. For example in my particular application, how much does flicker matter? It may matter less if I'm washing a wall than if I'm lighting the workspace in a classroom. How much does power quality matter? What other loads are in the building? This again is a complicated topic that deserves much more time by itself.

And in fact, or in reality I should say, you may end up having to pick between one of two options when you're weighing tradeoffs. One may be to only use sources and phase-cut dimming controls with defined compatibility and performance. This means when the manufacturer has already done some testing and can give you guidance or where the performance is addressed by some standard. And again some of the existing standards that separate the control from the AC input can provide more reliable performance.

Option two may require you to mock up the installation if you really need to know how it's going to perform and manufacturer guidance or standards aren't available to you. And what's important to know here is if you're going to mock up to determine whether the solution you're looking at will work for you; you need to mock up all the sources and all the dimmers. This means all source combinations. This means full circuits. You can't just look at one LED source or two on a given dimmer. If you're gonna use that source on different dimmers you need to try it out on the different dimmers. If you're going to put three sources on one circuit, five on another, you need to test out both of them. If you're gonna mix sources or mix LED's with other light source technologies, you need to try those things out because all those variations can affect your results. And of course this means beware of source or dimmer substitutions between the time you specify and go to build, because any substitution or model update can change your results.

Now when you're doing a risk analysis and trying to determine which approach to take, it may be relevant to think about well, what happens if things go wrong? What fixes are available? Well, you can if necessary switch out the LED source, at least if it's a replacement lamp, maybe the driver if it's a luminaire, maybe the dimming control. These are things most people are gonna want to avoid. Sometimes you can add and incandescent or dummy load to a circuit but again, this is not a desirable solution.

And of course I've already mentioned that in some cases replacing the dimmer with a dimmer that has a neutral wire and thereby requiring running a neutral wire to the dimmer box can reduce or eliminate many compatibility issues. But again this is a painful fix. So often there are no good sub solutions once products are installed. And at the end of the day if things go wrong you should probably think about up front and have a plan for who's responsible and who pays to fix things.

So in conclusion, will things get any easier? On one hand I told you that you can dim sources today but you need to learn some new things and do some homework and in some cases these things may appear to be painful. Will things get any easier? Well they're getting easier every day due to the development and deployment of new technology and new standards. Let's talk about some of those.

New LED drivers in some cases have embedded intelligence that can detect the characteristics of the dimmer they're attached to. And while it remains to be seen how well these types of products really work, they are starting to appear on the market and they at least hold the potential to lead to near universal compatibility. And here you see an example from a particular manufacturer touting their near universal compatibility. Also new dimming controls continue to hit the market. I've already mentioned that there are some phase-cut dimmers now that are – have been redesigned if you will to work better with LED sources and they provide guidance as you see an example here of how many LED sources and even how to mix LEDs and incandescent halogens on the same circuit.

These new dimmers also typically have different load ratings for incandescent versus LED or CFL loads and that is often stamped right on the dimmer which you see in the picture on the right. So in this particular dimmer, it can handle 600 watts of incandescent or 150 watts of CFL's or LEDs. Some dimmers also have a low end adjust trim or adjustment dial which raises the minimum dimming level. So you lose some dimming range but in making this adjustment you can reduce the chance of drop out or pop up.

Also new control technologies, this is continued to be developed and hit the market and this is really something that's taken off in the last year or so. There are power line carrier and wireless approaches and also system level solutions where the power supply or LED driver or a set of luminaires has kind of been centralized instead of being distributed in each different luminaire. In the right hand images there you see some examples of recently commercially deployed wireless solutions where of course everything has changed. The communication of the control signal is now wireless. The input device may be a cell phone or a remote control and the traditional dimmer installed in the wall is nowhere to be seen.

NEMA SSL-7 is a standards effort that has been underway for over a year that has the goal of reducing the variation if you will between phase-cut dimmers and LED sources such that many of the compatibility challenges I've discussed are eliminated and you can actually predict some performance. The standard development is broken into two pieces, SSL-7A which is focused on compatibility and is primed to be out in early 2013 and B which is more focused on performance aspects. So as designed, again, this standard will define compatibility and performance for compliance phase-cut controls and lamp luminaires. Now currently only covers forward phase-cut controls and it only covers light sources which connect to electrical grant circuits and have electronic power supplies. But by defining design specifications for both lamps, luminaires and phase-cut controls and compliance procedures or test procedures for those specifications, it holds the promise of delivering predictable performance.

Zigby Light Links is a standard that was just recently released and in fact it is the standard upon which one of the wireless solutions shown previously is based. Now because it's a different technology it requires a specific or new control and LED driver for a control/light source solution to function but in many cases this additional requirement can be low cost in particular because it can leverage other Zigby applications. Installation is of course wire free and therefore can result in very simple retrofits. Security has been addressed in the standard and it can be – and it is very easy to assign single or individual control devices to one or more light sources without the added wiring. Finally a certification process ensures compliance with the standard and thereby compatibility.

So with that we'll conclude here and take some questions. You can always e-mail questions in to TINSSL@pnnl.gov and I will turn it over to Kelly.

Kelly Gordon: Okay, yes, we've had a number of questions that have been coming in throughout the presentation. We will get to as many as we can. First just a clarification of Michael of what does RMS stand for?

Michael Poplawski: It's Root Mean Square. So it's a metric if you will for characterizing the resistive power if you will delivered by a voltage wave form.

Kelly Gordon: Great, thanks. The next question is what is the persistence time of HBLED phosphors?

Michael Poplawski: Actually I don't have those numbers in front of me but I think what's key, and I mentioned on one of the early slides, is LED phosphors are extremely fast so certainly the persistence is much lower than it is for incandescent and it's actually much lower than it is for even fluorescent phosphors. So the relationship between current and light output for LEDs is almost instantaneous. Now there are some people I've heard who have been experimenting if you will or trying to develop LED phosphors that have slower or longer persistence and they would certainly be useful in mitigating flicker behavior in some sources if and when they're developed and deployed but I haven't seen that yet.

Kelly Gordon: Great. The next question is why do you say that pulsed mode gives longer light? For the same average power either in pulsed mode or continuous mode, the LED temperature will be the same and that should give the same light.

Michael Poplawski: Well I meant can result in longer life compared to full output. So certainly compared to full output it's going to be consuming reduced power but certainly a dimmed source compared to a lower wattage if you will full output source may not – as this questioner has mentioned, result in longer lifetime.

Kelly Gordon: Okay. The next question, technically DALI, 0-10V and DMX512 are not dimming technologies but rather communication protocols. It was not phrased as a question so I guess it's just do you have a comment on that?

Michael Poplawski: Okay, well they're communication protocols used to dim light sources.

Kelly Gordon: Right. Are the luminaires shown on slide 15 integrated or non-integrated? So I may have to flip to that slide.

Michael Poplawski: They were – I'd have to double check so those were measured in a evaluation of both integrated and retrofit solution so I'd have to go back. I think they're all integrated but I'd have to go back and look. But those are all troffers actually. They're LED troffers either integrated or retrofits.

Kelly Gordon: Okay, the next question. Graphs show about 60% output at 80% dimmed input suggesting less efficacy yet the efficacy is 120%. I'm not sure which one this is referring to. Does that ring a bell for you?

Michael Poplawski: No. But if the questioner maybe wants to send in an e-mail referring to which specific slide they're asking about I can answer better.

Kelly Gordon: Maybe they can clarify that with you. Let's go on to the next question. Why do some LEDs flicker while dimming, then stabilize at a set dimming level?

Michael Poplawski: Yeah, so that is another behavior variation that I really didn't even address but typically at least when I've looked at flicker and what people end up most concerned about are what's the steady state behavior if you will. But we've definitely seen some sources have transient behavior so that can either be where the flicker varies while you're dimming or can be not aperiodic or not inconsistent. So there can be some kind of random phenomena resulting in changes in the light output modulation. So the reasons for that are even more complicated than the reasons for the steady state variation and it's really gonna be specific to the particular source and dimmer combination.

Kelly Gordon: Okay, next question, is there a graph that shows the human flicker percent detection limit versus frequency?

Michael Poplawski: So you know we were – this presentation was really focused on dimming and I mentioned a few things about flicker. We could certainly go into much more detail about that. There have been studies done with users that have attempted to gauge user ability to consciously perceive flicker in a given light source as the amount of flicker has varied and or whether the user objects to the amount of flicker and in some cases whether the users say for example task behavior is changing. But as I mentioned, many, actually most of the potential impairments of flicker are population sensitive. So not everybody responds the same.

And furthermore can be even difficult to talk statistically because certain use, certain population groups have higher susceptibility. So there are, there has been a variety of research studies done. I wouldn't say there have been a lot and I wouldn't say that they have been conclusive. This is very much a research-y topic but if the questioner wants to e-mail me I can direct them both some additional presentations and reference material on flicker susceptibility or perception if you will. So short answer is there is no plot that is applicable for all users.

Kelly Gordon: Here's a related question. Is there any flicker percent or flicker index limit that LED manufacturers need to design for?

Michael Poplawski: Well you know again this is definitely very much related and certainly many manufacturers have been asking that question and the answer like I said is not necessarily simple because there can be tradeoffs for designing a light source with less flicker, especially for smaller light sources or more space constrained light sources. And there are lighting application dependencies, having more flicker in a light source is potentially acceptable in some applications as it is less acceptable in others. And finally like I said, there's the user dependencies. So it's more important to reduce flicker if you're lighting a classroom full of young people as opposed to some other applications, again washing a wall for example. So you know of course the general guidance is to try to minimize it as much as possible. There are some good references that don't require a research paper necessarily.

I think we all know and the research backs up that the amount of flicker in incandescent sources is problematic to almost nobody while anybody who remembers the flicker present in magnetically ballasted fluorescent sources, that was problematic to many people and there was a large number of research studies that documented the effects on migraine sufferers for example. So I would suggest at a minimum trying to have less than – less flicker than a magnetically ballasted fluorescent source. And again if the user wants to e-mail me I can send more information about that.

Kelly Gordon: Okay, the next one is a quite broad question but perhaps you can take a crack at it. What are the advantages of one control type over another?

Michael Poplawski: I mean there's a – again we didn't have time to go into the details comparing and contrasting many of the different communication technologies or control approaches. You know certainly the ones that separate the control signals separate from the input AC power tend to have fewer potential issues but they also can tend to be more complicated to install and commission and more expensive. I mean there's a reason phase-cut has been so widely deployed. Not only does it dim the most common light source, incandescent, dims it quite wonderfully, it's very easy to install, but it has its own limitations too. It will dim every source on a given circuit so you can't do addressing on a given circuit.

So, control technologies vary on a number of ways. I think there are probably many resources online and again we could probably provide the questioner with some other material to describe some of the pros and cons of the different technologies in general if they were interested. But I think again with respect to the topic at hand here, dimming LED sources, the big tradeoffs are again the ease and large install base of phase-cut versus the additional compatibility challenges that present for LEDs versus the less frequent and more complex technologies such as 0-10V and DALI, etc. And there are potentially fewer issues.

Kelly Gordon: Okay. Next question, as a consumer how do we know which dimmer to pair with a certain LED replacement bulb? Are there standard labelings being adopted to inform the consumer of the dimming technology used in a dimmer switch or the driver electronics compatibility?

Michael Poplawski: So the short answer is it's a work in progress. Again in terms of which dimmer to use, many manufacturers both dimmer manufacturers and light source manufacturers are providing compatibility and guidance in the form of tables, but to date you have to go looking for them. You maybe have to go to the manufacturer website or know who to ask. And so there certainly seems to be potential for improvement by creating at least a one site clearinghouse if you will that could direct users to dimming information for many manufacturers and as I mentioned the Lighting Facts database holds the potential to do that but that requires manufacturers to put their dimming information into the database or link their information to the database. And that is not mandatory, it's optional.

So certainly something users could do is try to make requests of manufacturers to make their compatibility information more readily accessible, maybe more consistent than what it presents and maybe at the end of the day, explore options to have a one stop source for it. What was the second half of the question?

Kelly Gordon: Are there standard labelings being adopted to inform the consumer?

Michael Poplawski: I don't know of any yet but certainly groups like ENERGY STAR® are thinking about that. So I think that remains to be seen. I think again there's room for improvement in communicating compatibility with consumers and to date people are thinking about it or talking about it but there hasn't been any large scale deployments to such approaches.

Kelly Gordon: Okay, next question is, is there a difference in dimming or dimming performance when an LED product is constant current versus constant voltage?

Michael Poplawski: Not that I can think of, as a function of just that. Dimming performance certainly is a strong function of the driver design and how it can detect the dimming of the control signal and create an output but whether the driver is constant voltage or constant current, I mean typically LEDs are dimmed by adjusting their current level so a constant current driver is going to implement the dimming directly. A constant voltage driver may or may not. There may be some other circuitry in the luminaire that is regulating current and that is actually implementing the dimming.

Kelly Gordon: Okay. I think we have time for one more question. What do you think will be the most common way to dim LED drivers?

Michael Poplawski: Well if I had a crystal ball to answer that question I might not be here doing this webinar. You know I don't know. I think I look at this challenge as one like many in the energy efficiency world that requires an all approach path. So I think – and I think we're seeing that. You know SSL-7 has the promise of improving compatibility with phase-cut dimmers which have all their advantages, their ease of installation, their simplicity, etc. Of course that standard will still require the installation of a compliant dimmer. It won't say anything about existing dimmers but it might make quite usable traditional phase-cut technology.

However you know I think what's true for LED in general is we're really going to see the potential achieved for this technology as people explore new approaches and that means new approaches to you know, designing and deploying light sources, lamp and luminaire designs but also in this case dimming approaches. So it remains to be seen. The wireless approach for example has many advantages but you know it's using a new standard that may take a little while to get the bugs worked out and may require some user comfort or development of some user comfort with controlling their lights, putting their lights on a wireless network and using their iPhone instead of a wall switch. So you know a lot of potential, many advantages, but I think it remains to be seen how that's all gonna shake out.

Kelly Gordon: There will be plenty for us to continue to keep track of and I think that is going to close out our webinar for today. Thank you so much Michael for that information-packed presentation. Thank you to everyone who joined us today for the webcast brought to you by the U.S. Department of Energy. You may disconnect. Thank you.