Editor's Note: Scientific American's George Musser will be chronicling his experiences installing solar panels and taking other steps to save energy in 60-Second Solar. Read his introduction here and see all posts here.

When people talk about improving the efficiency of solar energy production, they usually talk about the panels themselves. What fraction of sunlight do they convert into electricity? Most solar cells today are made of crystalline silicon, but could cleverer designs or advanced materials such as thin films, organic polymers, layered semiconductors, and phosphorescent dyes do better? Probably, but that’s only half the story. The auxiliary equipment that connects the panels to your household wiring or the electrical grid is just as important.

A Lawrence Berkeley Labs study I cited in an earlier post found that solar has gotten cheaper over the past decade largely because of better auxiliary equipment rather than better panels. To find out what further steps engineers can take, I talked to Guy Sella, the co-founder and CEO of SolarEdge, an American-Israeli manufacturer of such equipment.

The need for this equipment arises from how a solar photovoltaic cell works. Light shining on the cell knocks electrons off the silicon atoms, and an electrical voltage built into the semiconductor material pulls the electrons in one direction, creating an electrical current. What happens then depends on what you connect to the cell.

If you don’t connect anything and just leave the wires dangling, the current has nowhere to go, electrons pile up on one side of the cell, and the voltage across the cell increases until it reaches the built-in voltage -- typically 0.6 volts for silicon. The BP SX3400b panels that are going up on my house each consist of 50 cells connected in electrical series, for about 30 volts if you don’t connect an electrical load. Twelve of these panels are strung together for a total of about 360 volts.

When you attach a load and start to draw power from the cell, the voltage drops -- gradually at first, then precipitously as the electrons flow out too quickly for a voltage to develop across the cell. This behavior is captured in a graph known as the current-voltage, or I-V, curve. When the voltage reaches zero, the cell delivers its maximum current -- which is about 9 amps for my BP panel in full-on sunlight and less when it’s twilight or overcast. Because the cells in a panel and the panels in a string are wired in series, the amperage of one determines the amperage of all. If you need more current, you have to wire strings of panels in parallel. My solar array consists of two 12-panel strings, doubling the current.

Because power equals volts times amps, a panel doesn’t do a whole lot of good if it generates 30 volts at 0 amps or 9 amps at 0 volts. In between these extremes, it produces useful power, and there’s a sweet spot in the middle where the power is maximized -- for my panels, 8.16 amps at 24.5 volts, giving 200 watts of power. If you hit this sweet spot and point this panel straight at the sun, it will convert 16 percent of the incoming solar energy to electricity. When most people talk about efficiency, this is the number they’re referring to, but it presumes you've hit the sweet spot, and that's easier said than done.

The job of optimizing the electrical performance of the panels typically falls to a piece of equipment called the inverter. Its main function is to convert the direct current produced by the cell into the alternating current used by the electrical grid -- a process known as “inversion” because it reverses the more common function of converting AC to DC (as battery chargers, for example, do). But a modern inverter does more than invert. It also adjusts how much current it draws in order to maximize the panels’ power output. As Sella explained, it’s tricky for many reasons:

  • Electrical mismatches. Because of the vagaries of manufacturing, different panels have slightly different I-V curves. The inverter responds only to the average I-V curve. Consequently, it draws too little current for some panels and too much for others, reducing their power output by several percent.
  • Partial shading. If the shadow of a tree branch or another solar panel falls on the panel (as in the above photo) and diminishes the sunlight hitting it by, say, a percent, you might innocently think it would diminish the power output by a percent. Actually, even a small shadow can completely zero out the power. Because the cells are wired in series, knocking out one can knock out all, just as a single blown Christmas tree bulb can black out a whole string of bulbs. Even when uneven illumination doesn't choke off all the power, it worsens the electrical mismatches. In a typical setup, Sella said the power output declines as much as 25 percent.
  • Temperature fluctuations. As the temperature increases, electrons flow through the semiconductor material of a solar cell more readily and the built-in voltage decreases. For my BP panels, the peak voltage drops by about 0.1 volt per degree Celsius. The trouble is that the inverter can handle only a limited range of voltages -- my SMA America SB4000US unit works from 220 to 480 volts. During extreme temperature swings, the voltage will fall outside this range and the energy will be lost. Depending on your climate, up to 15 percent of your annual energy production goes to waste.
  • Inability to optimize. Because of the above problems, the overall array I-V curve might have multiple sweet spots, some sweeter than others. The inverter will lock onto one, even if a better choice lies elsewhere. And whenever the sun’s brightness changes because of cloud cover or the time of day, the inverter needs to find the new optimum. In fickle weather, it may not be able to keep pace. Between these two problems, you give up 10 percent or so of the panels’ potential output.
  • Incomplete use of available space. Even if I had room on my roof for a 25th panel, I couldn’t install it. It would mean that one string would have 13 panels and the other 12, yet the strings must be of equal length. I couldn't subdivide my array into five strings of five panels each, since the length of the strings is dictated by the voltage that the inverter can handle. Because of the need to keep the number of panels numerically balanced, Sella said the typical commercial solar installation can utilize only about three-quarters of its roof.
  • Damage or theft. If a panel breaks or gets stolen (it happens), the whole array can fail. What's worse, you can’t just replace the lost panel with the latest model; you have to use the exact same model as the original, or else you’ll create an electrical mismatch. Thus a photovoltaic system installed in 2009 is locked into 2009 technology for its 25-year lifetime.

To get around these problems, Sella said that SolarEdge has developed a small box that you can attach to each panel (see photo at top). This box optimizes the electrical performance of each panel individually. He said the company has tested its technology on 17 houses in the U.S., Europe, Israel, and Japan and found it wrung another 10 to 20 percent of power out of the arrays at no extra cost. In fact, by simplifying the wiring or allowing more flexible use of roof space, the SolarEdge box can cut the installation cost. Sella said it will come out in October.

Over the past 30 years, solar power has gone from 40 times as expensive as fossil fuels to just a few times. At the rate the technology continues to improve, it won’t be long before it’s competitive even without government subsidies.

SolarEdge’s PowerBox units on the backside of solar panels in Germany (first image). Partial shading of rows of solar panels in Spain (second image). Courtesy of SolarEdge.