LEDs are extremely reliable and have a long lifetime, but can electronic LED drivers provide the required current or voltage input over the LED’s whole lifetime?
Lighting products using electronic technology appeared in South Africa in the early 2000s, with compact fluorescents lamps (CFLs) and electronic control gear (ECG) as main carriers.
The market learned some specifics of electronic lighting, like the sensitivity for high temperatures and voltage peaks, while also appreciating its advantages, including energy-efficiency.
The many benefits of LEDs include their reliability and long lifetimes. However, as LEDs are powered by drivers based on power electronics, the question is whether these drivers are able to provide the required current/voltage over the whole lifetime of the LED system, or whether LED drivers are becoming the weak link in the chain, seen from a reliability or lifetime perspective.
Outdoor lighting is one of the applications where reliability and lifetime are very important in outdoor lighting.
Extreme climate conditions
Outdoor fixtures must withstand extreme conditions such as fluctuating temperature; moisture; aggressive exhaust gases; vibration and even lightning. These fixtures are expected to function in these challenging environments and to produce the correct light levels at the right times, over long periods of time, all in locations which are often difficult to access.
Outdoor application
In terms of the installation, there are many factors which will determine the eventual lifetime. A luminaire operating in the tropics with night temperatures above 30°C will suffer more than a similar luminaire in milder climes with an average night temperature of 15°C.
The number of switchings will also influence lifetime because of thermal shock, specifically to the soldering. This should be taken into account where a presence sensor or a controller is used.
It is, therefore, not recommended to switch the lamp off completely, but to keep it burning in dimmed mode.
The next biggest reason for premature failures is probably electrostatic discharge or surge currents caused by high voltages induced by lightning.
The system should also be protected against surges present on the AC input line. These differential mode voltages are typically generated by high-power switching in the installation. The lifetime of the fixture will be determined finally by mechanical stress caused by heavy traffic or stormy weather, for example.
Definitions related to reliability and lifetime
Reliability experts often describe the reliability of a population of electronic products by means of a representation known as the “bathtub curve” (see Fig. 1).
Fig. 1: The bathtub curve.
During the initial period of use, drivers will typically fail because of weak components (“infant mortality”). This is followed by the normal life of the product, with a low and relatively constant failure rate. Following this is the final period of the product’s lifetime where “wear-out” mechanisms of the product kick in and the failure rates increase.
The bathtub curve does not depict the failure rate of a single item, but describes the relative failure rate of an entire population of products over time. Electronic devices will fail during the infant mortality period; others will last until the wear-out period while a few units will fail during the normal life.
Reliability deals with random failures in a population of products and is expressed in terms of rates, such as Failures in Time (FIT) or Mean Time to Failure (MTTF). MTTF is the theoretical accumulation of random statistical failures of all components in the product, expressing the “constant failure rate” over lifetime.
Lifetime, on the other hand, refers to the length of time that a single product may be expected to function properly before a known wear-out mechanism renders the product unfit for use. Lifetime is typically expressed in hours and normally indicates the duration of time with a minimum survival rate of 90% (obtained from the MTTF calculations). For instance, a lifetime of 100 000 hours implies that, under normal conditions in a typical installation (population), 90% of the products installed would be expected to last 100 000 hours before failure.
MTTF predictions
While the lifetime of the LED driver depends on the component most likely to fail, the failure rate of the driver depends on all the components within the driver. The MIL-HDBK-217F reliability model is used to predict the theoretical failure rate of Xitanium LED drivers, for instance.
As an illustration, for a typical 150 W outdoor Philips Xitanium LED driver operating at a temperature of about 50ºC, a theoretical failure rate of 500 PPM/1000 hours, an MTTF value of approximately 2-million hours is obtained. Note that, for the MTTF calculation, worst case electrical stresses are assumed to obtain a conservative estimate of the LED driver’s MTTF.
Higher MTTF values are expected where more realistic values are assumed. These calculations also assume a typical operating temperature. If the operating temperatures were higher, the stress levels on the driver components would increase, leading to increased failure rates.
Note also that the MTTF data is based on theoretical calculations only and can by no means substitute actual field data. Experience has shown that this theoretical prediction is much more conservative than the actual field data.
Designing for lifetime, reliability
Key factors to take into account when developing the most reliable product are:
Topology selection
For LED drivers, the first step is to select the most robust power conversion topology, given the constraints of power, size, cost and others.
System efficiency
System efficiency (or power loss) has a direct and significant impact on the reliability and lifetime of the LED driver because all the lost power is dissipated as heat within the driver, leading to an increase in temperature. The components within the driver operate at a higher temperature if the power dissipated in the driver is high. The reliability of components declines as their operating temperature increases.
Additional protection
The output current is reduced if the driver’s case temperature exceeds a certain value due to abnormal operating conditions. This, in turn, reduces power dissipation and ensures that the temperature of the driver’s internal components does not rise above a certain threshold.
Component selection
Suppliers should be selected with care and long-term relationships with the supplier ensures that only the best components are used. From a design point-of-view, careful analysis of component stresses and adequate derating of the components ensures a highly reliable LED driver capable of achieving excellent lifetimes.
Lifetime calculations
Having selected the components, it is important to determine which are most likely to fail. Based on the knowledge of HID and LED drivers, the components most likely to fail are taken into account, especially when the driver is operating at relatively high temperatures.
We consider the lifetime of a capacitor as an example of a critical component. The typical equation to calculate the lifetime at a certain ambient temperature LT, is given in Eqn 1.
(1)
where:
k is a factor that depends on the ripple current flowing through the capacitor.
T is the temperature of the operating temperature.
L0 is the lifetime of the capacitor at the rated case temperature.
The equation shows that every 10ºC drop in the operating temperature of the capacitor doubles its lifetime. This is known as Arrhenius’ Law.
System performance
For simplicity, this rule can also be applied around the Tcase temperature of the driver because capacitors are often the lifetime-determining components at higher operating temperatures.
Table 1 shows a simplified overview of the survival rate, with an example of a driver with a Tcase of 75ºC.
This reiterates the need for high-efficiency LED drivers to minimise power dissipation and, therefore, lower component temperatures. The construction of the fixture and the ability to lower the temperature of both the driver and the LED module have a significant effect on lifetime. Critical here is the additional thermal stress arising from the mutual heating of different components in a system. The self-generated heat of the driver is typically 20 – 25ºC. However, when the driver is mounted very close to the LED board, the heat from the LEDs will lead to additional temperature increase in the driver.
Ambient temperatures will vary between night and day, per region and from season to season. It is important not to exceed the maximum temperatures. The ambient temperatures in Table 1 can be used for lifetime calculations. A luminaire installed in Stockholm would generally experience half the failure rate of a similar luminaire installed in Barcelona.
| Jan | Feb | Mar | Apr | May | Jun | Jul | Aug | Sep | Oct | Nov | Dec | Year | |
| Minsk | -2,7 | -1,4 | 3,3 | 11,5 | 18,3 | 21,5 | 22,4 | 22,2 | 15,9 | 9,6 | 2,6 | -1,0 | 10,1 |
| Stockholm | -0,7 | -0,6 | 3,0 | 8,6 | 15,7 | 20,7 | 21,9 | 20,4 | 15,1 | 9,9 | 4,5 | 1,0 | 10,0 |
| Budapest | 1,2 | 4,5 | 10,2 | 16,3 | 21,4 | 24,4 | 26,5 | 26,0 | 22,1 | 16,1 | 8,1 | 3,1 | 15,0 |
| London | 7,9 | 8,2 | 10,9 | 13,3 | 17,2 | 20,2 | 22,8 | 22,6 | 19,3 | 15,2 | 10,9 | 8,8 | 14,8 |
| Paris | 6,9 | 8,2 | 11,8 | 14,7 | 19,0 | 21,8 | 24,4 | 24,6 | 20,8 | 15,8 | 10,4 | 7,8 | 15,5 |
| Barcelona | 13,4 | 14,5 | 15,9 | 17,6 | 20,5 | 24,2 | 27,5 | 28,0 | 25,5 | 21,5 | 17,0 | 14,3 | 20,0 |
| Lisbon | 14,8 | 16,2 | 18,8 | 19,8 | 22,1 | 25,7 | 27,9 | 28,3 | 26,5 | 22,5 | 18,2 | 15,3 | 21,5 |
| Valencia | 16,1 | 17,2 | 18,7 | 20,2 | 22,8 | 26,2 | 29,1 | 29,6 | 27,6 | 23,6 | 19,5 | 16,8 | 22,3 |
| Athens | 12,9 | 13,6 | 16,0 | 20,3 | 25,3 | 29,8 | 32,6 | 32,3 | 28,9 | 23,1 | 18,6 | 14,7 | 22,3 |
Temperature sensors can help to protect the module from over-temperature, module temperature protection (MTP) can be programmed so that the connected module will be dimmed down at over-temperature and be protected without switching it off.
Lifetime and switching
Switching has a big impact on system lifetime. The temperature difference (shock) between a system at rest in a cold ambient environment and a running system could be in the range of -10 to 35ºC, leading to thermal shock. Frequent switching such as by means of presence detection, will shorten the lifetime of the system. It is preferable to dim the light to maximise system lifetime.
Mains voltage
Product specifications include operating parameters for input voltage. Over-voltage, which can occur during switching or load changes, can impact the lifetime of the driver negatively.
Surge protection
In addition to the normal voltage fluctuations in the power line, LED lighting systems are also subject to damage from high-voltage surges such as lightning strikes or high load switching.
Surges can occur in two modes:
Philips offers external SPDs which include so-called active surge arrestors (spark-gaps) which will clamp these voltages on the input of the luminaire (in case of a high or differential mode surge voltage) and divert this energy to earth. This device provides protection for both the driver and other devices in the luminaire. The clamping component (spark-gap) is not allowed in Class-II luminaires because of regulations.
Experience has shown that, in environments such as cities, residential areas or in street lighting, a surge capability of 4 kV will be sufficient.
In higher risk areas such as mountains where lightning has a density above four flashes/km2/year, or where poles are installed in open fields, it is advisable to use an active SPD type-3 device to clamp the high-voltage surge at the input from the luminaire, and so protect the driver and module.
Testing and qualification
Accelerated life testing, including HALT/MEOST, is also performed to ensure high driver reliability. The data from these tests is compared with data obtained from similar tests for other products which have been operating in the field for a longer duration of time and for which enough field data is available. This ensures that every new product achieves at least the same level of reliability as a previously-released product. Initial burn-in or stress tests are done on statistically relevant sample sizes to limit failures in the infant mortality period.
Contact Henk Rotman, Philips Lighting SA, Tel 011 471-5000, henk.rotman@philips.com
