Specifying Variable CCT Lighting Fixtures - Part 3
Issue 71 Feb / Mar 2013
In the third part of his series on colour tunability in LED fixtures, Dr Geoff Archenhold explains the different methods for creating variable CCT luminaires and suggests some questions specifiers can ask to ensure they get the right product.
In the previous articles on colour tuning fixtures I introduced how colour and CCT were defined and calculated along with potential errors, but all of this science and long equations don’t help the lighting designer to specify variable CCT lighting fixtures. So this article will aid lighting designers in understanding the different methods of creating variable CCT fixtures and give them the heads-up on various questions to ask lighting manufacturers.
Firstly, I would like to say Happy New Year to all of you in the lighting design community and what a great year it will be too, with lots of new technologies making an entrance into the mainstream and ever increasing efficiency of LED-based products helping to make our industry ever more environmentally friendly. Of course it’s not all going to be plain sailing in 2013 as economic winds still blow against the building and construction sector in the UK and many parts of Europe, and the battle of differentiating quality from poor quality products is even more challenging.
Secondly, I was surprised that one of my roundup predictions was confirmed so early in 2013 with Cree recently announcing the production of both the XM-L2 single-die LED and the four-die XLamp MK-R LED. The new XM-L2 LED (figure 1) is a 5 x 5mm package that delivers up to 186 lumens-per-watt at 350mA, 25°C and 5000K with up to an incredible 1198 lumens at 3A of drive current. The second generation products offer 10% greater lumen output at a 5700K CCT and 17% better lumen output at 3000K. The products are binned at 85°C junction temperature and 700mA drive current, and at 3000K deliver 224 lumens and 112 lm/W.
The Cree XLamp MK-R was launched only two years after Cree broke the 200lm/W R&D efficacy barrier in their Labs. The MK-R is a single 7mm x 7mm package which holds up to 4 LED die to deliver up to 200 lumens-per-watt (at 1W, 25°C). The 6mm optical source is able to deliver up to 1769 lumens at 15W, 85°C. Characterised at 85°C, the MK-R component is available in 2700K to 7000K colour temperatures and offers minimum CRI options of 70, 80 and 90 (at selected colour temperatures). The MK-R has a low thermal resistance of just 1.7ºC/W and has a maximum forward current of 1.25A with a maximum junction temperature allowed up to 150ºC. Having such a low thermal resistance means that with good thermal management design the relative luminous flux will only reduce to 80% even at a junction temperature of 150ºC.
I have always stated that R&D lab breakthroughs take about two years to filter through into production products, so based on announcements in 2012 we should see 250lm/W LED emitters by 2015 and between 150 and 200lm/W fixtures by 2015 (depending on application, of course!)
Table 1 A basic overview of each variable CCT system solution type.
How will manufacturers create variable CCT fixtures?
The majority of variable CCT fixtures will most likely be created by one of the
1. Multiple single white emitters.
2. Single emitter with multiple white phosphors.
3. Multiple single colour emitters (including white).
4. Single emitter with multiple colour emitters (including white).
5. Variations of the above themes.
In addition each variant of the vCCT fixture can be controlled and maintained using:
• Open loop feedback (OLF)
• Closed-loop feedback (CLF)
Each of the above variations obviously have advantages and disadvantages which I will try and highlight in as simplistic way as possible through table 1 (opposite).
Multiple single white emitters
This solution is relatively easy to implement as it uses individual LEDs to create a variable CCT output by, for example, having two controllable channels: one channel with a certain colour temperature, say 2700K, and the second channel containing LEDs with a CCT of, say, 5700K. The fixture can then have its CCT changed by controlling the relative intensity amplitude of each channel (or banks of LED CCTs) independently. For example, if Channel 1 was switched on to full but Channel 2 was switched off, the fixture would provide a CCT output of 2700K. However if Channel 1 was switched off and Channel 2 was fully on, the fixture would output a CCT of 5700K and any ratio of output in between would create a CCT between 2700K and 5700K, thus making the fixture variable CCT.
Figure 2 Sharp’s new variable CCT Tiger Zenigata
This technique is ideal when large areas are to be illuminated, such as for linear strips or 600mm x 600mm panels. However the main issue in the past has been that the cost of placing twice as many LEDs within a fixture has been cost prohibitive, especially when they are only ever used at 50% overall power input. The system cost issue will be mitigated as the rapid cost reduction of LEDs continues to accelerate and the LED costs become a smaller cost in the overall system. A further potential disadvantage of this technique has been changes in CRI as the CCT is changed from its lowest to highest CCT settings due to different phosphor types of each channel. Again, this effect has recently been overcome with the availability of high CRI phosphors >80 at higher CCT values >5000K.
The advantage of this technique is that only two channels are required to operate and so the driver electronics are much easier and lower cost to implement within the system. As two white LEDs are used in each channel, the power spectral graphs are continuous, so the quality of light spectrum is very good compared to discrete colours. In addition, the efficacy will also be quite high and essentially the efficacy is dependent on the efficacy of the types of white LEDs used in both channels. As the efficacy gap between phosphor converted Warm and Cool white LEDs closes it means that a system based on this technique would have a relatively constant efficacy across any CCT value.
The fixture to fixture CCT consistency can be managed by appropriate binning strategies for each of the two channels and good current regulation tolerance between each channel of the LED driver. The LED driver doesn’t require significantly high bit resolutions in order to change the CCT or intensity compared to other techniques.
Figure 3 Tiger Zenigata showing the CCT according to the overall current through channels.
Single Emitter with multiple white emitters
Such a solution hasn’t really been widely available but it would have the majority of the advantages and disadvantages of the previous system type. Recently, Sharp has launched the Tiger Zenigata LED array (figure 2) which combines two independent channels at 2700K (96 LEDs of 12 series by 8 parallel) and 5700K (72 LED of 12 series by 8 parallel) to offer a cost effective LED solution for variable CCT fixtures.
The Tiger Zenigata electro-optical characteristics at a case temperature of 90ºC is shown in table 2 and highlights the high CRI of 90, even at the 5700K CCT range. The 3800K characteristics is based on a combination of both the 2700K and 5700K channels.
Importantly, Sharp has characterised this type of variable CCT solution by showing how the CCT can be controlled using an open loop solution just by setting each channel forward current accordingly as shown in figure 3. Here, the summation of current through both the 2700K and 5700K LED channels equal 700mA and shows that the CCT is fairly linear.
Figure 4 Tiger Zenigata luminous flux output according to CCT.
One of the disadvantages of using open loop control is the fact that the luminous flux will vary across the CCT range as shown by the data in figure 4. Although the output only varies by approximately 10% and is therefore unseen when at high light output levels, there is still a fluctuation of output as the CCT is changed and this may be more noticeable when the fixture is dimmed to low intensity outputs.
In order to ensure such issues are taken out of production systems it is possible for the driver and control system to include either a colour or intensity feedback to ensure the CCT and luminous flux curves are linear accordingly.
Table 2 The electro-optical characteristics of the Tiger Zenigata.
The quality of the output spectrum is guaranteed because the system uses phosphor converted white LEDs as shown in figure 5 which outlines the continuous 2700K, 3800K and 5700K CCT spectrums.
Multiple single colour emitters
These types of systems can be as simple as using single Red, Green, Blue and Amber/White LED emitters to produce a white colour tuneable system by adding their emission spectrum together.
This approach offers white light sources with potentially very high luminous efficiency. Theoretically dichromatic (two colour) white-light sources are most efficient, offering an efficacy of >440lm/W. The CRI of dichromatic sources is low but can be improved dramatically by increasing the number of primary-colour LEDs for a white source. However, sources with a greater number of primary-colour LEDs require more complex LED drivers and control systems and usually have a lower luminous efficacy. Trichromatic (three colour) LED-based white-light sources achieve a good balance between CRI and efficacy with CRI exceeding 85 and efficacy exceeding 300lm/W if the optimum combination of wavelengths is achieved.
One disadvantage of such solutions occur due to different LED die materials used within the lighting system and the output and efficiency of the LEDs change with temperature.
An interesting paper by Chhajed et al in 2005 demonstrated the variation of LED spectral output and the CCT within a three LED system as shown in figures 6 and 7 respectively. They observed that the chromaticity point shifts, CRI decreases (84 to 72), colour temperature increases (6500K to 7200K) and the efficacy decreases (319 to 297lm/W) as the junction temperature of the LEDs increases from 20 to 80°C.
Figure 5 Tiger Zenigata power spectral density graphs for 2700K, 3800K and 5700K CCTs.
A serious disadvantage of three colour systems occur with changes in CRI as the CCT values change particularly for warm-white light with a low blue component, this method becomes more inefficient (mixing proportion for 2700K: 43%R, 55%G, 2%B). The CCT variation issues become acutely apparent when one then tries to dim the intensity down from 100% to say 1% as the blue content is already low at 2% when the light is at 100% on, so to dim down requires very high resolution current control for the blue channel, which provides certain challenges. In reality, most variable CCT systems will not be able to obtain high resolution output stages because they use PWM type topologies and as the resolution of the PWM increases the PWM flicker frequency decreases, which is not good for visual quality or human health. One way of overcoming this issue is to use DC or analogue current methods so there is reduced flicker.
In order to improve the spectral quality it is possible to add further LED channels of different colours, e.g. white or amber LEDs to the system. Adding further LED wavelengths into the light engine will improve CCT and CRI of the system but impacts on the system efficacy and of course makes the control systems more costly and complex.
A novel solution available in the market today is the Osram Brilliant Mix LED combination. The Brilliant Mix solution utilises two off-white (white LEDs which have a known blue and green tint to their output) plus a Red/Amber or Blue LED to enable full CCT and CRI control. The advantage of the new Brilliant Mix concept is that it can provide high efficacy output (>100 lm/W) and high CRI (>90) across CCT tuning points from 2700K to 5000K using only 3 driver channels.
Figure 6 Spectral variations of a three LED (RGB) white light system with junction temperature (from Chhajed et al. 2005).
Single emitter with multiple colour emitters (including white)
This type of variable CCT solution is very similar to the previous system type. However, it is much more used for narrow beam angle fixtures such as stage or theatre projection lighting systems because the different LEDs are close together within a single emitter. Therefore, it has similar advantages and disadvantages but in addition the thermal variations are more complex because the power density and thus the thermal density will be much higher.
Open Loop feedback control systems
The advantage of open look control systems are that no feedback is necessary for the system to operate, so the driver systems is much less complex. As discussed, the issue with these types of systems that there is no (or little) CCT white point stability as there is no feedback to enable the driver system to automatically compensate for variations in LED current(s), LED junction and ambient temperatures, LED colour binning and external lighting conditions.
Therefore, it is recommended that if you want to specify variable CCT lighting fixtures you will need to avoid any system based on open loop feedback control as colour accuracy and consistency between fixtures will not be guaranteed.
Figure 7 Chromaticity/CCT variations of a three LED (RGB) white light system with junction temperature (from Chhajed et al. 2005).
Closed Loop feedback control systems
A closed loop control system is one where the control system self-compensates according to certain variables which are fedback from the systems outputs. In the case of a variable CCT system for example one could use a thermistor to measure LED temperature, an ambient light sensor to measure light intensity, a colour sensor to measure LED colour output and current sensors to measurement current through LEDs accurately. One or more of these variables would be used to feedback into a control system to regulate the CCT of the fixture to enable an accurate solution. Obviously, such a control system is more complex and requires additional components but in reality this type of solution is the only one if you require high quality, accurate and consistent CCT values between fixtures in a lighting scheme.
Caution should be taken that even with closed loop feedback control systems errors still occur so, for example, even if an optical colour sensor is deployed within a system potential errors will still be present, such as:
• The colour sensor only detects a portion of the light of an LED group due to location.
• The colour sensor sensitivity curve drifts and / or changes with temperature.
• The sensor’s deviation from linearity changes with temperature.
• A deviation between the sensor and the eye sensitivity curve.
• Limitation of the analog to digital conversion (ADC) resolution.
Of course, the number of feedback mechanisms employed and the quality/resolution of the feedback signals determine the overall accuracy of a lighting system, but this needs to be traded off against system complexity and cost.
The following questions are not exhaustive as I haven’t covered aspects, such as what patents and IPR do you have to enable you to sell variable CCT fixtures etc, but it gives you a clear steer as to the quality of the solutions you will specify and avoid post-installation problems such as colour shimmer, which was highlighted to me by several lighting designers during Light + Building last year.
If you have any further questions on this matter or any LED lighting technology, I am always at your disposal.
10 questions to ask when specifying variable CCT fixtures.
Q.1 Is the control system open-loop or closed-loop?
If open loop I would avoid if possible.
If closed loop then ask the supplier what they measure, e.g. temperature, output colour etc, and how many times per second the measurements occur.
The faster the control system the better the performance is when dimming, for example.
Q.2 What is the CCT and CRI colour accuracy of the control system?
This enables you to determine how good the control system is and the errors in the system.
Q.3 What is the CCT and CRI error between a group of 20 similar fixtures?
This enables you to determine if you have CCT and CRI variation issues between fixtures. It doesn’t necessarily fit that Q.2 means that consistency between fixtures is the same accuracy and hence why you must ask Q.3.
Q.4 What current control method do you use (e.g. PWM, PAM or DC)?
If the answer is PWM, PAM or other pulsing control methods then you need to follow on with additional questions below.
Q.5 If the control system uses a pulsing technique, what is the pulse frequency?
You know the ripple current will be 100% if pulsing, but the frequency is important as it determines the current resolution, which will highlights how good the dimming performance is.
If the frequency is low then you may see stroboscopic effects, but more importantly you will see a colour shimmering effect across a scheme, especially if the pulsing is not synchronised between driver channels and fixtures in the installation.
The higher the frequency the better, ie 1000-5000 Hz would be good.
Q.6 What is the current and feedback resolutions used in the system as this determines the accuracy of individual fixtures and the performance during dimming?
The higher the current resolution and feedback resolution the better the accuracy.
In reality it would be very good to have 16-bit resolution. However, in practice most systems will be 8-bit or 10-bit at most.
A word of warning: because a 10-bit system actually needs 16-bits to store ie; 8-bits + 2 bits so some manufacturers will state they use 16-bits but in reality it is only 10-bits although they use 16-bits to store the 10-bit number!
It is important to have as high a feedback resolution as possible because this defines the accuracy of the overall system.
Q.7 What control protocol do you use to control the CCT between fixtures?
This is important as you don’t really want to tie yourself into proprietary manufacturer standards if you can help it.
There are several standards that are now available that can help control variable CCT lighting systems, including DALI colour and RDM, so opt for one of these if you can.
Q.8 How is the fixture control system affected by ambient light or other lighting systems?
This is very important because some control systems and fixtures will start to act inconsistently if ambient light affects the control system and light sensor feedback.
A really good test of this would be to point one or more variable CCT lighting fixtures directly at each other and see if they become unstable, or to point one of the fixtures outside through a window and see if it remains stable with no colour shifts or flickering.
Q.9 Can you show me the CCT and CRI graphs against dimming level and manufacturing batch variation between fixtures?
This will show you how the variable CCT system actually performs when dimmed down. It will also highlight the variation between fixtures within the manufacturing process.
Q.10 What type of variable CCT system do you employ?
This will enable you to determine the type of issues you may see as highlighted within this article. When specifying variable CCT systems you should always ask for demonstrations of four similar fixtures to be done at the same time. The reason being that if just one fixture is used for a demonstration, the human eye isn’t very good at seeing colour or intensity inconsistencies and so even an open loop system could be demonstrated changing its CCT and would look good.
However, if you have multiple variable CCT fixtures your eye will be able to detect subtle variations in colour and intensity (when dimmed to below 50%) quite easily and so you can better judge how the system performs when multiple fixtures are used together within open plan spaces.
Dr. Geoff Archenhold is an active investor in LED driver and fixture manufacturers and a lighting energy consultant.
Parts 1 and 2 of this variable CCT series of articles are available to read online at