As global energy demand continues to drive oil prices higher design engineers across all application spaces investigate techniques to take advantage of “free” energy. Photovoltaic solar cells provide the most common alternative energy. Countless articles and studies have been done on Maximum Power Point Tracking (MPPT) algorithms to extract as much energy from a solar source as possible. However, these techniques are too complicated, too expensive, and frankly, require too much energy to be of much use to low power solar applications.
Consider an application that requires a 3.3V supply rail providing an average power of just a few tens of microwatts with peak power reaching into the tens of milliwatts. This could be, for example, a remote wireless sensor operating intermittently, a hotel door lock, an industrial control device, etc. This design will consider the energy harvesting device (solar panel), an energy storage device (battery), a battery charger and a voltage converter.
Many manufacturers produce low power photovoltaic devices including G24i (www.g24i.com), IXYS (www.ixys.com) and the Amorton series from Panasonic (www.anasonic.net/energy/amorton/en/) . I have some experience with the G24i products, so I’ll focus on their dye sensitized photovoltaic modules. The Indy4050 is targeted for indoor applications. Its 3050mm2 provide about 90mW at 200 lux and 465mW at 1000 lux. Remember that typical office lighting ranges from about 320 – 500 lux. For low power applications, the other critical specification is the output voltage at which maximum power occurs (VMP). For the Indy4050, VMP is 1.8V at 200 lux and 2.0V at 1000lux. If two modules are wired in series, this maximum power voltage aligns very well with a single-cell Li-ion battery.
Small (both in physical size and capacity) Li-polymer batteries can be found from many sources. I have purchased batteries from www.batteryspace.com, www.gmbattery.com and www.powerstream.com, but there are plenty of others. Consider the PGEB016144 3.7V/200mAh battery from PowerStream. The 740mWh capacity can power the average hundreds of microwatt load for thousands of hours giving the solar panel plenty of opportunity to recharge the battery. It is just 1mm × 44mm × 61mm. Furthermore, the 3.6 — 4.0V maximum power voltage of two series connected G24i solar modules coincides nicely with the charging voltage required for the battery.
This may seem like a great solution as-is, but there are some critical power management issues remaining. The open-circuit voltage of the two modules is about 5V at 1000 lux – more than enough voltage to damage the battery. Furthermore, the varying battery voltage needs to be regulated to 3.3V. Given the low power available from the solar panel and the amount of energy storage provided by the battery, it is critical that the remaining power management functions require as little quiescent current as possible.
Linear Technology’s LTC4070 Shunt Battery Charger System and LTC3388 High Efficiency Nanopower Step-Down Regulator meet these requirements superbly. See Figure 2 for the final application schematic. With a battery voltage of 3.7V and no load on the 3.3V supply, the combined quiescent current of the LTC4070 and LTC3388 is just 1.2µA! Virtually every electron generated by the photovoltaic cells finds its way into the battery for storage.
I ran this configuration without the LTC3388 for just over two days in miserable conditions to verify performance. In place of the LTC3388, I simply loaded the battery directly with a 4kΩ load (just under 1mA – close to 4mW) and left the setup outside. It rained nearly steadily for all 52 hours of the test – a few breaks in the rain, but never any sun. Even with these adverse conditions, it was clear that the battery was net charging. With more experimentation and optimization I’m sure this ~4mW constant load application could be powered with a smaller solar module and less battery capacity. Here’s the data (no fancy automated data acquisition system here, I just walked up to the setup and measured a few points during the experiment):
So why did I only run this experiment for 52 hours you ask? Well, I was the victim of grand larceny (okay, maybe it doesn’t meet the legal definition of “grand”, but it was grand in my mind). I decided to leave my setup outside over the weekend unsecured and in plain sight. By Monday morning, it had disappeared without a trace. In summary, low power photovoltaic products really can be used with extremely high efficiency (after the lousy solar to electrical efficiency hit – I can’t do anything about that one). The key is to use as few ICs as possible and ensure that those ICs do their job with as few electrons as possible. The other key is to match the solar panel with the energy storage element. Fancy MPPT algorithms are not necessary, if the energy storage element forces the solar panel to always operate at or near its maximum power voltage.