A lesson in quick-and-dirty solar charging

A lesson in quick-and-dirty solar charging

Advances in regulator circuitry have removed the “rocket-science” element from solar-based power supplies.

The Burning Man event is well-known for its tribute to alternative lifestyles and an ecological creed. But among the most notable facets of Burning Man is its Black Rock City desert locale, a hundred miles north of Reno, Nevada. It is characterized by daytime temperatures routinely exceeding 100°F with extremely low humidity and dry weather which rapidly and continually wicks moisture from the human body. At nearly 4,000 ft above sea level, the atmosphere provides much less filtering than at lower elevations.

For many, staying cool in the desert heat poses a challenge. But it is also an opportunity to demonstrate what can be done with a photovoltaic power source in an area far removed from conventional ac mains. This was the motivation for a solar-powered charging system for a marine deep-cycle battery to power lights, radios, and a water mist system consisting of a water pump feeding a mist hose with nozzles.

The resulting assembly is an interesting illustration of what can be done with circuitry consisting of not much more than a few demo boards.

The charger takes input power from two BP solar panels (BP380U), each outputting a peak power of 80 W at 17.6 V and 4.55 A. A principle task in devising a charging circuit is to accommodate the output voltage from the solar panel which varies widely, between 0 and 20 V depending on the sun position. The circuit uses a regulator that will accept this wide input range and maintain low current intake (Max output current from the panels is about 4 A each.) while regulating a fixed voltage on the output. Principle components include a Linear Technology µModule dc/dc switching regulator, the switching mode buck- boost LTM4607 power supply µModule regulator mounted on a demo board (DC1198A-B).

The LTM4607 is a LGA-packaged (15×5×2.8-mm) chip that houses all support control components of a buck-boost dc/dc switching regulator. The switch-control circuitry and power FETs are built into the regulator, making it easy to use. The result is a clean layout with just a egulator, inductor, and a few capacitors and resistors. The 4.5 to 36-V input voltage range regulated to a fixed 20-V output well suits the solar panel 0 to 20-V output. The device can load up to 5 A in boost mode and in 10 A buck mode. The circuit is 91% efficient at the solar panel peak power and actively uses the benefits of a buck-boost wide-range input.

For the purposes of this design, output is regulated to 20 V. This 20 V powers an LTC1435/LT1620 synchronous step-down switching regulator controller and current-sense amplifier which regulates average output current independent of input and output voltage variations. The result is a high-efficiency low-dropout battery charger with over 95% efficiency.

An LTC1435/LT1620 demo board (DC133A) controls charge current to a steady 4 A at a regulated 14 V. The demo board for this application is quite close to the application circuit on the first page of the LT1620 data sheet with the exception of the FB resistor (110 kΩ). It is changed to a variable potentiometer which hand-adjusts the output voltage to set the battery float voltage to 14 V.

The DC133A demo board design incurs only a 0.5-V input-to-output voltage drop at 4 A charging current. A programming current to ground sets a battery charging current (4 A) that is regulated until the battery voltage reaches the programmed float voltage (14 V in this case). As the battery reaches its fully charged state, the circuit's programming will auto-resolve into a trickle charge and slowly reduce the charge current with respect to the battery's output voltage. This prevents overcharging and thus puts less stress on the battery.

An ideal diode circuit design (LTC4414) goes in series with the output of the DC133A charge system. It serves as circuit protection and lets loads use the battery while the charge circuit is operating. When the solar panel doesn't provide enough power for the load, the circuit resolves and draws power from the battery.

A current-sensing system goes in series with the power to the battery from the panels. It uses a shunt sense-resistor to measure input charge current and output discharge current without breaking the circuit. This takes place using an LTC6103 (operating from 4 to 60 V in an 8-lead MSOP package), a dual independent current sense amplifier that monitors current through external sensing resisters. It measures and delivers a current ratio that indicates how much the battery is charging and discharging. This is a low-power-loss method of reading current which is vital to energy efficiency.

A slight modification of the LTC6103 demo board (DC1116A) provides this function. Pins 8 and 7 (+IN_A and -IN_A respectively) go in series with the flow of charging current into the battery. Pins 6 (-IN_B) and 5 (+IN_B) are swapped and connected in reverse and measure battery discharge current — Pin 5 connects to Pin 7 and Pin 6 connects to Pin 8. Resistor values are changed by a factor of 10 and adjusted so the output is 100 mV/A with a 0.1 Ω shunt sense resistor in series with the circuit.

ADC and microcontroller readout

A microcontroller and ADC read the system voltage and displays on a LCD screen. A USB controller demo board (DC590B) with a demo board for the LTC2418 8-/16-channel 24-bit ADC (DC571A) handles these chores. A PIC microcontroller samples voltages across the ADC's different channels and accurately reads voltage in the millivolt range at a reasonable resolution. The maximum range from the reference voltage was 2.5 V, reduced via voltage divider to deliver voltages in millivolts for proper measurement on the ADC.

The ADC channels tie to the individual relevant input and output voltages including the current-sense lines. The LCD gives information on the changing solar power voltage, charge circuit voltage, battery voltage, and battery charge/discharge current. Note the DC590B demo board gets power from a 5-V rail rather than the 12-V rail, necessitating use of a buck regulator to lower the battery voltage to 5 V. An LTM4601 switching regulator in the DC1041A demo board handles this.

The LTM4601 is an LGA-packaged 15×15×2.8-mm µModule dc/dc switching regulator with a 4.5 to 20-V input to 0.6 to 5-V output at a maximum 12-A load current. The design of the LTM4601 makes it easy to get a regulated 5-V output from the 12-V battery. In this system, the efficiency is approximately 90%. The output voltage is set by a resistor to 5 V and is easily changed on the demo board with a jumper to get such voltage rails as 3.3, 2.5, 1.8, 1.5, and 1.2 V.

Making mist

The misting assembly mainly consists of a bilge water pump, 15-ft hose, and misting nozzle system. The hose screws onto an adapter plug attached to the pump with hose clamps. It connects the pump to a system of five mist nozzles. The bilge pump runs from the 12-V source. Reducing the voltage on the pump changes the water pressure and the amount of mist delivered.

For flexibility, a voltage regulator takes in 12 V from the battery and outputs a variable 12 V to the pump. This setup uses the buck/boost properties of the LTM4607 design. A feedback resistor on the stock board controls output voltage. In this system, a 50 kΩ variable-knob potentiometer takes its place, giving easy control of the output from 0.8 to 12 V. A 5.62 kΩ resistor is also in series as an output voltage limit to keep output below 15 V. The design enables control of the water pressure with the twist of a knob.

The pump uses about 6 A from the battery at maximum level. This means it uses about 4 A of current from each solar panel at peak output. The ability to control the pump speed and pressure makes it possible to lower the settings enough to let the pump run entirely from the solar panels to conserve battery power. At Burning Man, the mist system could be set to run all day with no concern of discharging the battery.

LED illumination was the other main load on the Burning Man solar charging system. Making this possible is the recent development of LEDs that can put out over 100 Lumens at 1 A. A power source for such devices is the LTC3475 (16-lead TSSOP thermally enhanced package) constant current LED driver which comes on a DC923A demo board. It is designed to drive two 1.5-A channels from a wide (4 to 30 V) input voltage range.

The 12-V battery connects directly to the input of this demo board, powering three LED lights in series for each channel to give a total of six LEDs with both channels on. These LEDs are amazingly bright and were sufficient to light up our entire camp when wrapped with a diffuser cloth. Of course, the discharge current during the evening was entirely on the battery. With a 2-A total discharge current, the lights could run all night. By late morning or early afternoon the battery was fully recharged, ready to power the mist system in the heat of the day.

Other loads can draw power from a system like this. Communication radios are an example — cell phone reception is nonexistent at Burning Man. Radios today commonly include an adapter plug for charging through a car cigarette lighter. For quick charging, adding a 12-V cigarette lighter female adapter to the battery proved useful in that radio batteries last only a few hours with constant use.

Debugging and pitfalls

Anyone designing a solar-based charger should allow for limited solar panel use: As the voltage from the panel varies, so too does the current. Real-world factors such as shade falling on the solar panels or lack of sunlight can create unforeseen conditions in charging circuitry. In one extreme case, waving my hand over the panels locked up the system and latched it in current-limit mode. This is because the current-mode architecture forces the system to draw more current from solar panels when diminishing sunlight drops the panel output voltage. The solution is to have a system generating power that varies with changes in input voltage and current, depending on the amount of sunlight on the solar panels.

Because the panels could only provide 4 A, insufficient current at the input latched up the system. It remained latched until manually reset. The simple solution was to reset the RUN pin on the LTM4607 µModule regulator when input voltage dropped below a certain threshold. This could be done using a comparator with a set reference voltage to trigger when the panel voltage dropped.

Unfortunately this caused the set-up to either turn on or off depending on the amount of sunlight on the panels. A better solution is to adjust the charge current with respect to the solar panel voltage so that charge current varies with the level of sunlight on the solar panels. In other words, when the solar panel output voltage drops, so should the charge current.

One approach to synthesizing the behavior is to use an op amp (LT1006) and MOSFET to control the PROG pin on the LTC1435 (DC133A). This pin controls the level of charge current output in a ratio to the amount of current sunk from the pin. The op amp tracks the solar panel voltage, adjusts the Rds-on of the MOSFET and controls the current, depending on the voltage level of the panels.

When the panel voltage is at its maximum, the op amp turns the MOSFET on, letting maximum current sink out and deliver 4 A of charge current. When panel voltage is low, the charge current should also be low to maintain proper power output. A potentiometer can adjust MOSFET current biasing such that it cuts charge current in half when the solar panel voltage reaches its midway point, 10 V. At maximum, 20 V, charge current should be 4 A. The approach in this case was to set the potentiometer for maximum current when testing at a solar panel voltage of 20 V and to reduce the voltage to 10 V while adjusting the potentiometer slowly to a 2 A charge current without latching up.


Linear Technology Corp.

Thanks to Graham Freeman for lending the BP solar panels for this project. Simon Lim performed the initial prototype design for the solar power charger and helped throughout the entire design. Mark Thoren provided source code on the DC590B and ongoing software support. Fran Hoffart and Jay Lin helped with debugging. Thanks to Doug Coker, Mike Fahmie, and Dev Gopalkrishnan for conceiving of and inspiring me with the original idea of a mist system and for providing contact to Graham Freeman.

TAGS: Solar
Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.