Barth and Pilawa make a ripple in solar power

8/19/2013 10:04:00 AM ITI Staff

Shade almost always creeps into conversations about solar power. What happens on cloudy days when a rooftop array isn't producing enough energy for someone's solar-equipped bungalow? What about the mobile devices that might be operated in remote, wooded research stations where consistent sunlight is never available?

Written by ITI Staff

Shade almost always creeps into conversations about solar power. What happens on cloudy days when a rooftop array isn't producing enough energy for someone's solar-equipped bungalow? What about the mobile devices that might be operated in remote, wooded research stations where consistent sunlight is never available?

Beyond the obvious implications—that energy must be drawn from alternative sources, or that use is delayed—this fluctuation also affects the efficiency of the photovoltaic (PV) system. For optimum power production, the system must always operate at the right voltage and current. When the cells are shaded and the electrical current declines, voltage must change accordingly, and such adjustments must also be made depending on temperature, age of the panels, and, in the case of a solar-powered battery charger, the voltage of the battery.

Assistant Professor Robert Pilawa and Christopher Barth at their PV research station. The solar panel below is attached to the buck converter in Barth's hand. At center, the screen shows the dithered ripple.
Assistant Professor Robert Pilawa and Christopher Barth at their PV research station. The solar panel below is attached to the buck converter in Barth's hand. At center, the screen shows the dithered ripple.
Assistant Professor Robert Pilawa and Christopher Barth at their PV research station. The solar panel below is attached to the buck converter in Barth's hand. At center, the screen shows the dithered ripple.
To make these adjustments, power converters known as maximum power point trackers are used in solar PV installations. These are often expensive and consume considerable energy, but Christopher Barth, a first-year ECE graduate student, and Assistant Professor Robert Pilawa-Podgurski have developed a new and vastly improved approach for designing efficient and inexpensive tracking hardware. It really represents about a 10x improvement to the state-of-the-art, said Pilawa. Their research earned a top paper award at the IEEE Workshop on Control and Modeling for Power Electronics (COMPEL), held last June in Salt Lake City.

Essentially, the converters used in PV systems regulate voltage by rapidly turning the signal on and off. This is known as pulse width modulation (PWM). By adjusting the widths of the on-pulses, the current and voltage draw from the solar PV system can be controlled. For instance, when the sun emerges from behind a cloud and more electrical current is being generated by the solar cell, the converter's microcontroller must quickly adjust the PWM operation to ensure that maximum power is extracted from the solar panel. If the converter is not fast and accurate enough, precious energy is wasted.

As these adjustments become more precise, however, the systems become infeasible for low-cost, low-power applications. Cost and component sizes increase, as does the energy required. If cost isn't an issue at all, and the power consumed by the microcontroller isn't an issue at all, you can just run the microcontroller at a much higher frequency, Barth said, but both tend to be key considerations.

For solar-powered battery chargers used in rural, developing communities, for example, both cost and power consumption must be minimized. And there are situations, where, in addition to cost and efficiency, the large number of power converters embedded in a given system makes size a limiting factor. One such size- and weight-constrained application is solar-power unmanned aerial vehicles. You can't always afford to just buy fancier hardware, Pilawa said. Sometimes you've got to take what you have and be smarter about how you use it.

Barth and Pilawa's solution is a process known as dithering digital ripple correlation control (DDRCC). With dithering—an existing technique—the converter rapidly toggles between two PWM operation points. At the longer pulse-width, the voltage rises until it reaches a specified maximum; then a shorter pulse-width is used, sending the voltage back down. Done repeatedly, the resulting voltage, instead of being constant, resembles a ripple—rising and falling continually.

Even though dithering uses a much lower PWM frequency overall, the ripple traditionally causes a delay in maximum power point tracking and has been viewed as a deleterious side effect. But with an algorithm for ripple correlation control, Barth and Pilawa, were able to use the ripple as the basis for the maximum power point tracking, thus achieving excellent tracking speed and precise PWM regulation. Instead of trying to squash the ripple, we used it, Pilawa said. Something that was negative, we've turned around now as a tool.

Barth and Pilawa demonstrated DDRCC using a buck converter, a type of power converter that steps down the DC voltage produced by the PV panel. Their buck converter employed a microcontroller that cost around $1, already illustrating the low cost and minimal hardware needed to accomplish DDRCC, and Pilawa believes that parts costing as little as $0.25 are feasible.

This project began as Barth's undergraduate thesis; Pilawa was his research advisor. When the deadline for COMPEL paper submissions came, only three months into Barth's graduate studies, his diligence as an undergraduate paid off. He was able to submit a paper that garnered a top award among more than 80 other submissions. It was actually my first major conference to go to, Barth said. Don't underestimate the effect of the initiative and focus you demonstrate when you're doing [undergraduate research].

Pilawa agreed: I'm a firm believer in undergrad research, and this goes to show that these aren't papers that will just gather dust. There are actually some real results that come out of it; so that is certainly exciting.

The duo now plans to develop additional design guidelines so that, depending on the specific system characteristics, the best algorithm parameters can be selected, but even now, the project could have almost immediate application. This could be handed over to some company today, and they could start developing it, Barth said. In other words, this could go from a senior thesis, to a best-paper award, to commercial acceptance in almost no time.

Editor's note: media inquiries should be directed to Brad Petersen, Associate Director of External Relations, at bradp

at
at
illinois [dot] edu or (217) 244-6376.


Share this story

This story was published August 19, 2013.