To deliver the maximum amount of power from solar panels or arrays, standalone inverters and grid-tie inverters use a sophisticated strategy to find the maximum power point, or MPP, for given conditions. They do this by varying the load on the solar panel or array until it finds the point on the panel’s or array’s I-V curve that will deliver the most power. The inverter then converts this DC power to AC power, which is then used locally or fed into the power grid.
Determining how well an inverter accomplishes this task is extremely important. The feasibility of a project may rely on an inverter being as efficient as the manufacturer or distributor says it is. A discrepancy of even a few percent could render a project uneconomical, and at worst, could lead to lawsuits.
To effectively test the efficiency of photovoltaic inverters, you need a solar array simulator that both accurately simulates the output of a solar panel or array and that can supply the output power required for a particular application. Using actual solar panels or arrays is just not practical because it is not feasible to control their output to the degree required to simulate all conditions.
Modern Simulators Are Up to the Challenge
To simulate a solar panel or array, inverter designers and manufacturers use modern, digitally-controlled power supplies. When coupled with sophisticated control software, these systems can simulate solar panel arrays up to 1 MW.
Accurately simulating a solar array to test inverters can be quite a challenge. As noted earlier, inverters continually change their output impedance, searching for that maximum power point. The simulator must respond to those load changes as a solar array would. Not only must the simulator maintain its power output, it must track the I-V curve of the solar panel or array it is simulating.
To complicate matters, many solar inverters generate AC ripple on their DC input, which is connected to the photovoltaic array. For single phase inverters, the frequency of this ripple is twice the line frequency (120 Hz for U.S. models). Normally, you would want a power supply to suppress this ripple, but a solar array simulator’s power supplies must not suppress it.
An increasing number of inverters (and virtually all micro-inverters) accurately measure amplitude and phase of the ripple voltage and current to quickly determine the MPP of the array. This approach allows inverters to determine the MPP at a much higher speed when compared to conventional dithering techniques (also called perturbate-and-observe). Faster tracking to the MPP results in a much higher overall efficiency in cloudy conditions, where the irradiance is constantly changing. It is likely that all solar inverters will soon use this approach since end users are very sensitive to the overall efficiency of their solar energy installations.
Another requirement for modern solar array simulators is flexibility. There are many different types of solar panels and solar arrays on the market, each having their own particular characteristics. Modern solar array simulators must be programmable to allow them to determine how inverter designs will work with all these different types of solar arrays.
Finally, solar array simulators need to measure and log inverter’s AC output and correlate that data with the DC power input. This closes the test loop and allows you to determine how efficient your inverter design really is.
Staying on the Curve
In order to perform an accurate test, the output of a solar array simulator must faithfully follow the I-V curve of a solar array or solar panel. That is to say that it must respond just as a solar array would to the changing load conditions imposed by the inverter under test. In order to evaluate how well a simulator can do this, you need to consider three parameters: output noise current, phase error between output voltage and current, and the MPP tracking accuracy.
Excessive output noise current will make it difficult, if not impossible for an inverter to find the maximum power point, and this will cause test problems.
The point of inoperability due to the level of noise is a function of the inverter itself. The noise contributed by any PV simulator is made up primarily of its internal power switching noise and its associated harmonics. Since both the inverter and the PV simulator have control loops, there could be interaction. From the inverter design perspective, the inverter design engineer is trying to design a cost effective, high performance inverter. Since an actual solar panel doesn’t produce any noise, the engineer is less likely to add filtering and the associated cost. The level of noise is really key; 70ma versus 700ma is of course an order of magnitude and could be critical to some inverters and not so much to others. However, we have been seeing a trend in which the inverter designs are reducing nice-to-have circuitry in favor of lower cost. While this doesn’t make the inverter any less reliable, it may make performance testing a challenge with PV simulators that are not very refined.
In the real world, there is essentially zero phase difference between the output voltage and output current of a solar panel or array, even when inverters use MPP tracking strategies that change the load very quickly. To accurately simulate a solar panel or array, therefore, it is important that the phase error of the simulator be less than 15 degrees even if the load is changing that quickly. Many simulators are unable to do this, making them unsuitable for testing those inverters that use high sweep frequencies.
In an extreme case, when the phase error between the output voltage and output current approaches +/- 90 deg, the inverter under test will actually begin seeking the MPP in the wrong direction and will become unstable. With a significant phase error, even if the inverter is stable it will lock onto a curve location that is not the MPP. The amount of error is proportional to the phase error. We have observed this behavior on microinverters and residential inverters that use fast MPP tracking algorithms.
Another important specification is the maximum power point dynamic tracking accuracy. This is a measure of how much a simulator will deviate from a programmed I-V curve under dynamic conditions, where many factors can contribute to this inaccuracy.
Figure yy (above) shows a comparison of the dynamic MPP tracking accuracy of two simulators currently on the market. A test frequency of 16 Hz was used for this test to simulate the loading effect of an inverter. The red lines show the ideal response of the simulator, while the blue lines show the actual response. As you can see, there is quite a difference between the two simulators, and the difference is even higher as the dithering frequency increases.