The steadily increasing popularity of PV systems is being driven by the opportunity for significant savings over the lifetime of the system.  PV modules themselves malfunction only rarely during that time, as has been convincingly demonstrated by the mid-1980s PV modules that continue to operate on approximately 30 homes in Gardner, MA, as well as by a handful of other still-functional PV projects of similar vintage. In the interim, PV module technology has continued to improve: the robustness of new units surpasses that of those older units, as evidenced by the standard 20-year to 25-year warranties that accompany most PV modules today.  Inverters, on the other hand, have only made modest progress over the same period of time. Manufacturers today only offer inverter warranties of 10 years to 15 years, and this means they still need to be replaced long before any other components of the PV system.  By some accounts, inverter failures have been responsible for about 80 percent of the downtime of today's PV systems, remaining the weakest link in the overall reliability chain of such systems. This is a problem because there is a national focus on driving down the lifecycle cost of PV energy, and that means maintenance costs must be reduced over the full 20-to-25-year lifetime of the system.

New developments in inverter technology, particularly on the microinverter front, are making progress towards increasing reliability. Microinverters paired with individual PV modules can improve system reliability through their distributed architecture.  If a single unit fails, that failure does not affect the rest of the system, which will continue to operate.  In contrast, centralized inverters hold all the cards; if the central inverter fails for any reason, the entire throughput of the system is lost until the unit is repaired or replaced.  As a result, microinverter-based PV systems have a higher rate of overall system uptime.  But what about the reliability of the microinverter itself?

The lifetime of a microinverter is one of the biggest factors in calculating the lifecycle cost of electricity.  Maintenance costs associated with swapping out all the microinverters in a system, partway through the system life, would be significant. While the microinverters on the market today improve overall system availability, they do not fully address the most important long-term performance criteria -- namely, inverter reliability. They share some of the same design approaches and components of traditional inverters and will likely manifest the same reliability flaws. The harsher environment that microinverters face on a rooftop -- integrated with the mounting structure or with a PV module -- and the sheer quantity of these distributed units in difficult-to-access locations, mandates the most robust design possible for true system reliability.

Under normal use, AC modules and their component microinverters experience two basic forms of temperature stress: temperature cycling and operation under high-temperature conditions.  Measurements show that the internal temperature of a microinverter can swing as much as 50 to 60 degrees Celsius, from quiescent overnight lows to mid-day temperature highs. Higher frequency, but lesser magnitude, temperature variations overlay this daily cycle, resulting from the variations in sunlight intensity owing to passing clouds.  The other form of temperature stress arises from the relatively high-temperature operating environment.  Yes, an inverter underneath a PV module is blocked from direct sunlight, but it resides in a space near the sun-baked warmth of the module's backside, which can be 25 to 30 degrees Celsius hotter than ambient air. Internal to the inverter, some amount of self-heating also occurs due to the variable energy losses inherent in the power conversion process. These environmental effects combine to cause significant thermal cycling and power cycling for a microinverter.  The challenge is designing microinverters to withstand these stresses and thrive for 25 years or more -- and that challenge is now beginning to be met.

Navigant Consulting recently reported that "[t]he various switching devices used in inverters to convert DC to AC current are another weak point for inverter reliability." For example, fans, switches and relays are devices that are incompatible with a 25-year useful lifetime.  Also in this category are electrolytic capacitors, which seem to be the Achilles heel for today's PV inverters.  Navigant further reports that "capacitors are often cited as the most severe reliability problem for inverters. They are extremely sensitive to temperature (electrolytic capacitors), and one manufacturer reports that an increase of even 10°C can halve capacitor lifetime".  Loss of electrolyte over time leads ultimately to the demise of e-caps, and when that happens, the inverter fails.  Both high operating temperatures and repeated temperature cycling exacerbate electrolyte loss, suggesting that microinverters are a questionable application for them.  Although electrolytic capacitors have improved, these devices have a known failure mode and historically have been the cause of many inverter failures.

Despite this poor track record, it appears that some good news is on its way. Next-generation microinverters are about to enter the market, and these versions address the design weaknesses of today's inverters, promising lifetimes equal to the full 25 to 30 years of PV modules. These new designs eliminate the known weak elements of traditional inverters and their failure mechanisms. Electrolytic capacitors are replaced with film capacitors; unlike electrolytic capacitors, metalized film capacitors do not have a limiting wear-out mechanism.  End-of-life for film capacitors is typically defined as a loss of 2% of initial capacitance, compared with loss of 20% capacitance for electrolytic capacitors. If the inverter is designed to function with 20% loss of capacitance, then film capacitor life can be nearly 10x the design lifetime listed on the datasheet. Film capacitors are also more rugged than electrolytic capacitors because they can withstand much higher voltage surges relative to their voltage rating, and there are self-healing properties inherent in their construction.  Reliability is further enhanced by constraining the electrical and thermal stress on components, and protecting critical components from environmental noise and contaminants.

The true measure of the improved reliability over today's products are tests and metrics that hold real meaning in terms of a product's useful lifetime. Accelerated lifetime tests, in general, try to replicate a lifetime of harsh conditions in a condensed timeframe, and, for example, can simulate the quantity and magnitude of 20 to 100 years or more of real-world temperature cycling.  In addition, a high-temperature test, with operating inverters maintained at elevated temperatures in an oven for an extended period, is intended to accelerate the chemical reactions that take place over the course of decades in a real-world environment. Yet another example is highly accelerated lifetime testing (HALT); HALT is an accelerated test process by which a product is stressed to its limits of operation. When failures occur, an effort is made to determine the root cause, to mask that root cause, and then to continue testing to determine other possible modes of failure. Unlike other tests, HALT purposely takes units to failure through combinations of temperature, temperature cycling, vibration and operating at the device's current and expected voltage limits.  Success in a full suite of accelerated stress testing can produce a compelling body of evidence, strongly indicative of a high-reliability device.

A second approach is reliability prediction using statistical analysis of an inverter's component set.  This does not involve testing, but instead relies on industry-standard methodologies such as Telcordia SR-332 Issue 2 and MIL-Std-217.  Both tools use a parts-count and parts stress analysis approach to predict the mean time before failure (MTBF).  The MTBF is often confused with useful life, but the two are not the same. For example, assume we want to limit the total number of failures of a product population to four percent over a 20-year useful lifetime, which indicates a failure rate of 0.2% per year. The MTBF associated with those statistics is 500 years! A remarkable feat, but one that may be achievable with the new microinverters.     

The combination of availability benefits from the distributed nature of microinverters as well as the reliability advantages of the next-generation products holds great promise to reduce the lifecycle costs of PV systems and to further fuel the growing demand for solar electricity.



Miles C. Russell is co-founder, CEO and President of GreenRay Inc.  He is responsible for company management and product development. He was formerly Director of Systems Development at Schott Solar. During his 32 years in the PV industry, Miles generated the concept for the SunSineTM AC module, led the design of several mounting systems, and successfully grew and managed a small business.  Mr. Russell has been involved in PV systems research, design, and development continuously since July 1977, when he joined the engineering staff at the Massachusetts Institute of Technology Lincoln Laboratory.  Miles holds an MS in Mechanical Engineering from Stanford University.