Wednesday, November 10, 2010

Solar Array Simulation

When designing a device with max power point tracking (MPPT) capability such as a solar inverter, micro inverter, or power optimizer you are going to need to simulate the output of a PV panel or panels to test your MPPT design.  This is necessary to verify your design and provide an accurate efficiency spec under a wide variety of weather patterns (I-V curves). Now a lot of engineer new to this type of testing decide to take a standard programmable power supply, connect it remotely to a computer (using GPIB, LAN, USB, etc), and create software with adjustable I-V curve look-up tables with the idea of turning their power supply into a PV panel simulator. This whole concept is shown in the figure (click to enlarge).

You can also find power supply companies out there that sell this exact solution. I am here to save you time, frustration, and money by telling you this method will not work or at least it won't work how you think it will work. With this type PV simulator system the I-V curve 3 dB bandwidth will be < 1 Hz. Most MPPT designs are making load changes and measurements much faster than once per second. Here is a quick breakdown  why the bandwidth to this solution is so poor:
  1. The IO latency between the computer and the power supply.
  2. The supply programming time which consists of the time it takes the supply to process a command and the time for the internal analog circuitry to move the output of the supply to where it should be on the curve.
  3. The most limiting factor is that this is a closed loop system, which leads to oscillations in the output. Now to get rid of oscillations digital filtering will need to be added to the software which leads to multiple iterations of back and forth adjustments between the computer and supply to zero in on that point on the I-V curve where the supply output should be.
Because of the overhead just discussed you cannot achieve output bandwidths better than 1 Hz using this solution to simulate the output of a PV panel. For this reason it is not an acceptable test method to verify your MPPT design or spec the efficiency of your MPPT design. Now does this PV simulation method have a place in the MPPT hardware test cycle? Because of its relative low cost it could be used in long term reliability testing where you are just interested in continually feeding power through your design over a long period of time to make sure it doesn't break down.
So what is out there to simulate the output of a PV panel? There are two main ways to do it: take a power supply and put some custom analog circuitry around it or purchase a solar array simulator (SAS). I have heard of many different approaches using custom analog circuitry, below is a link to a paper that presents one way to do it:


Of course building a solution yourself comes with a high overhead of simulation, layout, testing, and support. If you want a finished solution you could purchase an SAS. An SAS is not a standard power supply. It is more comparable to a high powered current source with a low output capacitance (< 100 nF) to give it a high output bandwidth. Of course with these more advanced capabilities it comes with a higher price tag than a standard power supply. There are not too many companies out there that make SASs. Agilent is one of the few and below is a link to Agilent's E4360A SAS:

4 comments: