Microinverters and power optimizers: Power conversion migrates down to the PV panel
Alternative forms of power have become increasingly popular as utilities react to rising fuel prices and government mandates. Photovoltaic cells usually receive most of the attention as solar-energy cost drivers, but the ac/dc inverter design is equally important to overall system efficiency and cost.
By Margery Conner, Technical Editor -- EDN, December 3, 2009
| AT A GLANCE |
| The move toward smaller installations has made solar radiation less uniform for each PV (photovoltaic) panel.Microinverters and panel power optimizers can be cost-competitive with large central inverters when you figure in labor and installation costs.Reliability over the 20-year lifetime for power installations is a challenge for electronics. |
The size of solar-power installations is shifting away from the multimegawatt solar farms of 10 to 20 years ago toward smaller installations of 1 MW or less. Factors driving this trend include utilities’ increasing use of residential and industrial rooftops as localized solar-power-generating stations, thus reducing the need for conventional power plants, and the emergence of solar-PPA (power-purchase-agreement) companies, which install rooftop solar panels on homes and small businesses in exchange for access to the generated power (Reference 1).
Small installations differ in some ways from massive solar farms. For example, solar-farm panels all face in the same direction and generally experience the same amount of sunlight. They typically encounter no obstacles, such as trees or utility poles, which can obstruct the sun on different panels at various times of day, causing panel-to-panel variations in power output. Small solar installations, in contrast, must accommodate a variety of roof lines, especially in residential installations, which lack uniform panel orientation and thus can yield suboptimal power generation.
Small installations do share some features in common with solar farms, however. They both require regular cleaning, for instance, and all panels age at slightly different rates, causing a variation in panel outputs. Both small and large solar-power installations also use PV (photovoltaic) solar panels, which comprise arrays of solar cells and typically have a voltage output of 25 to 30V dc. Users of these panels usually cascade them in series, forming strings with a typical output of approximately 300 to 350V dc. They can also further parallel these strings for large solar arrays. The output of these arrays feeds into a central inverter that transforms the dc voltage to ac and synchronizes the ac voltage to the grid. The power output of inverters for large arrays of solar panels can range to 6 kW and beyond; central residential inverters range from 2.5 to 3 kW.
However, not all solar panels are created equal, and inconsistencies in manufacturing, obstructing elements, dirt, aging, and other factors can cause panels to produce different amounts of power. Each panel has an optimal power point, which affects the optimal power point for the string and, ultimately, the array. If the installation draws too much power from the array, the output power drops. If it draws less power, the installation cannot make efficient use of the array.
Algorithms can find the power “sweet spot,” which ideally would occur at each panel. In a common technique, “disturb and observe,” the power-transformation circuitry attempts to draw a little more current to see whether the voltage drops. The algorithm performs MPPT (maximum-power-point tracking), during which it searches for the point at which it gets the maximum power from a module. In traditional solar installations, this process takes place at the central inverter. With a central inverter, you most likely will find a local maximum rather than an absolute maximum for the array because the performance of one poorly performing panel will dominate the algorithm. If all the panels are well-matched, the difference between the true maximum and the local maximum power points will be insignificant. You can’t count on having well-matched panels, though, because of aging differences, a passing cloud cover, or the presence of dirt. One poorly performing module dictates the power that the other modules in series with it deliver.
|
One approach to this problem is to use a dedicated microinverter for each panel that finds the maximum power point for each panel rather than use one inverter to find the maximum for all the modules. The output of the dedicated microinverter goes directly to the grid or ac power-distribution circuit. European solar companies tried this approach more than 10 years ago using ac panels, each of which produced an ac voltage and tied into the grid. The approach has economic drawbacks, however, because it costs more to use one inverter—albeit a small one—per panel than it does to use just one central inverter or even one inverter on each string of the array.
Large installations minimize panel variations with regular cleaning schedules and by avoiding or removing shading obstacles. However, with the trend toward smaller installations, architectural limitations may dictate varying panel orientations, and utility poles and trees can cause individual panels to become dirty or be in the shade at different degrees and at various times, so you would need to optimize each panel.
Installation costs
Central inverters are also large and heavy, requiring the installation of cement pads and centralized connections to the grid, increasing installation and labor costs. A 200W microinverter that must handle only 30V-dc input becomes attractive for small installations, including users wanting do-it-yourself setups. In addition, a microinverter’s module output power is only 200 to 300V ac. In contrast, a panel array’s output to a central inverter can be as much as 600V dc in the United States and 1000V dc in Europe—hazardous levels for installers, maintenance personnel, and emergency responders.
Enphase Energy, among the first companies to deliver microinverters, sells a 200W device for approximately $200, or about $1/W, compared with a 3-kW string inverter, which sells for approximately $2000, or about 67 cents/W. Enphase suggests that its products’ lower installation and investment costs compensate for the additional 33 cents/W.
However, microinverters’ reliability is just as big of a question as their cost. Common inverter topologies use electrolytic capacitors on their output filters. These topologies have a poor reputation for reliability, especially when you subject them to the elevated operating temperatures of solar installations. The likelihood of a failure due to an electrolytic capacitor increases when you go from using one central inverter to using 10 to 20 microinverters. Most solar panels have a guaranteed life of 25 to 30 years, and operators want a similar lifetime from their inverter circuits.
Enphase’s Web site includes several white papers dealing with capacitors’ reliability and lifetimes. One paper explains how the company uses more reliable electrolytics in capacitors than those that power supplies normally use (Reference 2). According to the company, higher reliability translates to a longer life in the real-world temperatures you find in solar installations. For traditional power converters, an acceptable useful life for capacitors operating in 85°C environments is as low as 2000 hours. Enphase microinverters use Nichicon capacitors that operate for 4000 to 10,000 hours at 105°C.
Capacitor lifetime is sensitive to temperature and follows the Arrhenius equation, which states that useful life doubles for every 10°C temperature drop. The NREL (National Renewable Energy Laboratory) solar-radiation database for the California desert town of Palm Springs in the summer lists a maximum ambient temperature of 46°C, resulting in a core temperature for the capacitor of 65°C, or 40°C lower than the 10,000-hour rating at 105°C, yielding a 160,000-hour operating life at this temperature (Reference 3). Enphase claims to have designed its microinverters “for a service life of 20 years,” but the Web site guarantees its products for only 15 years.
Inverter reliability
Enphase sells its stand-alone microinverters separately from the panels. The inverters can work with a range of solar panels, require no central inverter, and have almost 95% efficiency—compared with central inverters’ 98% efficiency figure. They perform MPPT at the panel, increasing the efficiency of each module.
However, according to Kevin Kayser, marketing manager for National Semiconductor’s SolarMagic renewable-energy business unit, electrolytic capacitors aren’t solely responsible for failure mechanisms in solar installations. “Talk to any system integrator out there, and they’ll tell you the most unreliable component is the inverter,” he says. “[The microinverter topology] multiplies the inverter throughout the array.” The company’s SolarMagic modules exemplify another panel-based power-management scheme that works with—rather than replaces—a central inverter. National Semiconductor coined the term “power optimizers” for dc/dc converters that optimize the power at each solar panel, adjusting current and voltage so that each panel outputs its maximum dc power. SolarMagic panel installations use a string topology and require a central inverter. Each panel outputs maximum power, optimizing the string’s performance. If one of the panel’s voltage drops due to, say, the presence of dirt, SolarMagic algorithms adjust the module’s current to arrive at the optimal power output.
“[SolarMagic’s] dc/dc-conversion approach is less complex,” says Kayser. “Microinverters boost a 28V panel to 350V dc and convert it to ac, touching the grid at every module. Because of the 60-Hz frequency of the grid, the capacitor must handle that frequency, requiring less-reliable electrolytic or film capacitors. [SolarMagic’s] dc/dc approach uses ceramic capacitors.”
National Semiconductor is not the only company in this market. SolarEdge has announced a similar dc/dc power-management approach at the panel level, but its technology requires a proprietary central inverter. The company embeds the panel-control electronics in the panel, rather than selling them as separate modules, and performs MPPT for each panel. SolarEdge then cascades the panels in strings with a fixed voltage output for each string and feeds them into a central inverter, which performs only dc-to-ac conversion because MPPT has already taken place at the panel level.
For designers who choose to develop their own integrated panel converters, STMicroelectronics offers the SPV1020 PWM (pulse-width-modulation)-mode dc/dc boost-converter IC, which can maximize the power that PV panels generate independently of panel temperature or the amount of incident sunlight. The converter implements hardware algorithms to calculate the MPPT for PV cells within the solar panel. Because individual solar cells can begin to fail and disrupt the total panel’s power output, panels typically have bypass diodes that switch poorly performing cells out of the internal PV-panel array of cells. The SPV1020 resides in the connection box of a panel and replaces the bypass diodes. The chip also integrates power MOSFETs that perform dc/dc switching and synchronous rectification. Like SolarMagic’s technology, because the dc/dc conversion occurs at a relatively low voltage, the power converter can use ceramic capacitors rather than less reliable electrolytic capacitors.
Microinverter companies are focusing on squeezing every last bit of efficiency from their topologies, in part because of the need to deliver as much power as possible to their installation’s customers and in part because they are competing with the high conversion efficiency of central inverters. The technology benefits from the increased efficiency of low-voltage switching elements. The voltage output of a solar panel ranges from 25 to 30V, so the first regulation stage in a microinverter typically uses 40V. These MOSFETs have over the last 10 years improved their on-state resistance by an order of magnitude. For example, Infineon’s 40V Optimos-3 series has an on-resistance as low as 1.1 mΩ; 10 years ago, it was more than 10 mΩ. In addition, low-voltage MOSFETs’ use in high-volume consumer products, such as onboard power regulation for desktop and laptop motherboards, has decreased their price. Microinverters also use high-voltage FETs for the final output-voltage phase-matching stage, which requires 600 or 800V MOSFETs. Infineon’s CoolMOS technology has been on the market for more than 10 years and has during that time reduced the on-resistance to 20% or less.
| References |
|
Enphase Energy www.enphasenergy.com
Infineon www.infineon.com
National Renewable Energy Laboratory www.nrel.gov
National Semiconductor www.national.com
Nichicon www.nichicon.co.jp
SolarEdge www.solaredge.com
STMicroelectronics www.st.com
-
It seems SPV1020 is not a part of ST Microelectronics.
Requesting to please elaborate that part and provide the datasheet or any document on that solution.
Ankit - 2010-19-4 01:05:00 PDT -
Are you sure that you correctly quoted the ST micro SPV1020 part number?
ST don't seem to have heard of it! Perhaps it's someone else'e product.
Kevin Krause - 2009-4-12 02:13:00 PST -
The assumed lifetime of a typical Si PV array is ~20 years. While the lifetime of the PV diode itself should be much greater, it's the PVA encapsulation that ages, becomes opaque, and causes decreased array output andeventual failure.
But if a good silicone encapsulant were used, I imagine 50+ years life could be expected. Where then would this put the MTBF of 15 years for micro-inverters?
It's all "good enough" quality via de facto built-in-obsolescence and aging mechanisms that shortchanges the potential for deployed PV systems.
Elmer Fudd - 2009-3-12 15:15:00 PST
