By Rajul Randive, Crystal IS
Due in large part to their ability to deliver reliable, maintenance-free performance for up to 10 years, UV-C LEDs increasingly are being found in today’s low-flow, point-of-use (POU) water disinfection systems. It’s not surprising then that design engineers are taking UV-C LEDs to the next level by putting them to work in significantly higher-flow point-of-entry (POE) home and industrial water treatment systems.
While such systems require a greater number of LEDs, determining the projected LED population’s lifetime and reliability at the outset still are critically important. This article demonstrates the importance of taking an LED population’s reliability data into account during a POE system design phase. It also provides performance testing results for an NSF 55 Class B POE reactor at different flow rates. NSF Class B ultraviolet (UV) water treatment systems are intended for treatment of already treated water to ensure water quality. These systems are installed at point of entry (POE) in residential applications to protect water in the home from opportunistic piping pathogens or possible residual contaminants from the municipal infrastructure.
LED Performance at Higher POE Flow Rates
While mercury lamps still are the incumbent technology for POE applications, product engineers are considering new higher-flow designs that leverage the many benefits UV-C LEDs provide, including a more effective germicidal light source, significantly more design flexibility and reduced cost of ownership for the end user – a true win-win.
However, like established POU systems, understanding and designing for a POE system’s end-of-life performance still is critically important. In both cases, it’s not the first gallon of water that users need to be concerned about; it’s the very last. Understanding how LED reliability (B-value) is affected by the quantity of LEDs employed, therefore, will allow designers to create home and industrial water disinfection systems that can meet virtually any application’s performance, maintenance and cost requirements.
A typical POU system, such as a countertop or under-sink purifier, typically employs two to five LEDs at a flowrate of 0.5 to 1 gallon per minute. Whole-house disinfection systems, on the other hand, currently process anywhere from five to 40 gallons per minute. Most operate at roughly 10 to 25 gallons per minute and are designed to meet NSF 55 Class B disinfection standards.
Figure 1 demonstrates that a POE system using only 16 LEDs can meet NSF Class B performance at 10 gallons per minute, fulfilling the requirements for a standard residential system. That’s key because, from a system cost standpoint, it clearly demonstrates that designers can meet NSF 55 Class B performance at higher flow rates with only 10 to 20 LEDs – a much lower number than previously believed. This confirms, then, that UV-C LED-based systems are ideally suited for higher-flow home and industrial systems from both a performance and cost standpoint.
Designing for Reliability and Performance
However, demonstrating that an LED assembly can handle higher-flow duty at the outset is one thing, but can it perform consistently at that level over several years, or the lifetime of the product? Again, it’s not how the system performs when it’s installed; it’s about ensuring the last glass of water is treated to the same NSF Class B level.
The expected lifetime for a residential POE system often is three to six years, which – with the LED’s intermittent operation – translates to approximately 3,000 hours at end-of-life. When designing to this end-of-life, the reliability of a multiple LED assembly must be analyzed, not just a single LED. This means understanding how to interpret the lifetime (L-value) and reliability (B-value) data based on an assembly of LEDs. Product reliability refers to the percentage of the LED population operating outside specifications. For a given forward current and operating temperature, there will be a natural statistical distribution of light degradation (or lifetime). The percentage of devices in an LED assembly or population which exhibit light output below a specified L value is known as the B value.
Figure 2 illustrates the different LED behaviors based on B values of a population. The solid line shows that at 10,000 hours, under specific operating conditions, 50% of a sample population (the average) will emit 70% of the initial output, while 50% of the sample population will emit something less than 70%.
The dotted line in Figure 2 shows data for B10 of this group of devices. The B10 line shows that at 10,000 hours only 10% of devices emit less than 50% of their initial output – which is the L50B10 value. It’s important to note, however, that there are many factors that impact reliability and lifetime. For instance, humidity, current and voltage, temperature and mechanical forces can dictate how long a product will last without failure.
A System-Focused Approach
To put this in perspective, the author’s company tested a random population of 1,300 LEDs to simulate the effect of variability on treatment systems with different numbers of LEDs.
When reviewing reliability data for a POU design with a 300-hour end-of-life, designers must focus on operational data in the first few hundred hours. Similarly, for a POE system with a 3,000-hour end-of-life, the focus should be on data relevant to that timeframe – and based on the number of LEDs used in the design. Evaluating the performance based on a PCB array with only two to five LEDs – a typical design for POU – obviously is not relevant to a POE design. To illustrate this point, reliability data is interpreted based on two systems – a POU using two to five LEDs and a POE using 10 to 20 LEDs.
The data in Figure 3 shows test data for 1,300 LEDs, which were stressed and tested for over 10,000 hours. Simulations then were created to randomly select LEDs from the population and model scenarios for designs using two, five, 10, 20 and 30 LEDs for a combination of 50,000 simulated systems.
Figure 3a shows the behavior of the two LEDs after 300 hours. Of the simulated 50,000 two-LED systems, 99% of them still will have better than 70% of the initial power at 300 hours (L70B1). Figure 3b shows the behavior of the LEDs after 1,000 hours. Of the simulated 10-LED and 20-LED systems, 95% still have better than 70% of the initial power at 1,000 hours (L70B5).
These simulations clearly demonstrate that when evaluating reliability, one must consider the entire LED assembly/population rather than each individual device. If a POE system designer only considered the data for a two-LED system at 1,000 hours, it would show a much higher B-value than a 20-LED system. To compensate, the designer might unknowingly over-engineer the system – employing too many LEDs for redundancy and subsequently increasing the system’s overall cost. However, when considering the reliability data with a more appropriate 10- or 20-LED system, the designer will see the system reliability is 95%.
This design approach provides a much more accurate picture of how the system will perform over its lifetime, resulting in a system that’s both highly efficient and cost-effective at the same time.
UV-C LEDs – Taking the Next Step
Moving from mercury-based UV lamps to UV-C LEDs in water disinfection systems obviously requires a new design approach. As this shift already has taken place with POU systems, POE designers are leveraging the lessons learned with POU systems to create new, sophisticated reactor designs that deliver cost-effective NSF 55 Class B-level disinfection performance at significantly higher flow rates.
Going forward, different industries likely will require different designs. In a highly competitive industry like consumer goods, engineers may design the lowest-cost solution possible. In others, like healthcare, maintaining the highest possible level of disinfection for a set period of time will be paramount, regardless of the additional cost.
By better understanding LED reliability data and how it relates to the total number of LEDs, designers accurately can balance performance, reliability and cost – regardless of the industry or application.
Dr Rajul Randive is the director of application engineering at Crystal IS, a US-based manufacturer of UV-C LEDs. He supports customers with their product designs and prototype testing to ensure their products meet the application requirements. Dr. Randive also captures customer feedback to inform Crystal IS’ product roadmap for future UV-C LED products. He holds a Ph.D. in Organic Chemistry from Clarkson University. For more information, contact randive@cisUV-C.com.
