Solar Powered 12 Volt Backup System

Solar energy is the 'in' thing right now, so I thought I would get a tiny piece of the action. I didn't want to pay 20,000 and have solar panels all over my roof, but I did buy a small 40 watt panel so I could have a play. I can always expand it later if need be. The following describes how I built a basic solar power system to provide some security lighting at night, and back up lighting during a power cut. The ingredients required are of course a suitable photovoltaic (PV) panel, a storage battery and a means of controlling the energy into and out of the battery so that it stays in good condition. The rest is all cables, fuses, switches and stuff like that


Note: This project is a work in progress and I'm still experimenting and carrying out tests. Unfortunately winter is here and so it's difficult to achieve full power, but I will continue to play around with it and post the results here

The first task was to install the panel. The ideal place I thought was the garden shed roof as it faces south and catches the sun pretty much all day, being the last spot to end up in shadow. Now I don't like the idea of drilling holes all over my shed roof, so I came up with a secure mounting system that could be fitted by drilling just one small hole. I constructed a framework from rough sawn treated timber (19x38x1800mm), and fixed it to the underside of the panel using M6 25mm zinc plated roofing bolts, recessed into the wood (this is important). At the centre is an M8 75mm zinc plated cup sq bolt. The photo below shows how it was put together

The diagonal batten is attached with wood screws, ensuring first that the mounting bolt is central, and all the sawn ends are treated with wood preserver to give protection from the weather. When the M8 nut is tightened on the inside of the shed (see photo below) the two parallel battens are pulled against the roof by the diagonal batten. Just enough tension is applied to create friction between the parallel battens and the roof (which is rough in texture anyway) to stop the whole assembly being able to rotate. It's a good idea to use silicone sealant around the bolt to stop any water getting in. With this fixing method there is no stress on the frame of the solar panel at all as it is effectively just sitting on the wooden battens. For larger panels, I suggest using the recommended mounting hardware

I couldn't get away with just one hole in my shed though. Two more were required. One for the PV connectors to enter and another in the floor for the outbound supply cable, and it wasn't long either before the English weather did its thing. The clouds came, the rains came and the winds came, but the little panel sat there taking it all in its stride. I'm happy... so onto the next stage

The battery used is an 88ah deep cycle type, also known as a leisure battery. Any old car battery won't do, the reason being that car batteries don't like being discharged to any great extent, that's why they are constantly being charged by the alternator. They can supply vast amounts of current for short periods when needed, for example to turn the engine over, and relatively small currents for things like the radio and small lights, but will never get depleted very much in normal use. So running a car battery down is a bad thing. On the other hand, a deep cycle battery is designed to run down a bit as it powers various appliances and then be topped up again, that said, even the voltage on these batteries should not be allowed to get too low and as such, this should be monitored. 11 volts seems to be the acceptable minimum before damage occurs. Conversely, they should not be overcharged either as this can harm them too. 14.4 volts is deemed the maximum safe limit. So within these extremes, a deep cycle leisure battery should survive well. I had thought about building a circuit to automatically disconnect the battery if its voltage fell too low, but I'm hoping that the modest power demand of my system will negate the need for this, and any energy used will be replaced during the day. Even in the dreadful English weather!

An important part of any renewable energy system is the charge controller. This maintains the battery voltage at its optimum level. Commercially made units are usually processor controlled PWM (Pulse Width Modulated) types which vary in price depending on features and current handling. Here though, I've adopted the much simpler method of shunt regulation, which may not be as efficient as PWM, but is effective and reliable. In this application, a shunt regulator relies on the fact that a solar panel can be short circuited without causing any damage. This amazing property is ideal, as it can be put to good use when designing a simple controller to charge a battery to an exact preset voltage level. The schematic is shown below

The circuit uses an LM393N voltage comparator. The inverting input (pin 6) is fed from an LM317L programmable regulator, its output voltage being set by R2, which is a precision multiturn preset. The non inverting input (pin 5) samples the voltage across the battery via the potential divider formed by two 10K resistors. As these resistors are equal in value, the voltage at pin 5 will always be half that of the battery, whatever it happens to be. Set up couldn't be easier. Simply adjust the voltage at the test point (TP) to half the desired battery voltage and the circuit will do the rest. So for example, to charge the battery to 13 volts, you would set R2 for 6.5 volts at the test point.
The adjustment range of R2 is 1.25 volts to 7.5 volts. The formula for calculating the output of the LM317L is: Vout = R2 R1 + 1 x 1.25

Now if the battery voltage is below 13 volts, pin 5 will be less than 6.5 volts and so there will be no output from the comparator. As the battery charges, its voltage will rise towards 13 volts at which point pin 5 will also be 6.5 volts. Because of the extremely high gain of the IC, pin 5 only has to go an immeasurably small amount higher than pin 6 to make the output (pin 7) go high. This turns on the IRF540A MOSFET which connects the 2R2 resistor across the solar panel, effectively dumping its energy into a resistive load instead of the battery

The MBRF1645 is a 16 amp schottky diode that prevents the battery from being 'shunted' by the 2R2 resistor along with the solar panel. It also prevents the battery putting volts back into the panel, though most panels have diodes built into them anyway. This component is not critical as long as it's up to the job. Normal silicon diodes can also be used, but because they have a higher voltage drop across them it's best to use schottky types as we want to utilise as much energy as possible. The IRF540A MOSFET is rated at more than 20 amps so overall the circuit should quite easily handle the low amount of power demanded of it, and cope with additional panels being added later. The 2R2 resistor should be able to disipate the energy from the solar panel without getting too hot. I used a 100 watt wirewound

It's worth mentioning a bit about the LM393N. It is not an opamp. It can be thought of like an opamp as far as its inputs are concerned, but the output (pin 7) is very different. When it's off, as in when the + input is lower than the - input, the output will be internally grounded, effectively connecting the gate of the MOSFET and the lower end of the 10K resistor to earth. But when the + input goes higher than the - input, the output will 'float' open circuit, letting the 10K resistor feed the gate, turning it on. A nice feature of the LM393N is that it has a low current consumption. This is important in 'eco friendly' systems where we don't want to waste anything. The whole circuit, including the LM317L regulator, only consumes about 3.5mA. The low power part of the circuit was constructed on veroboard as shown below. The output goes to the gate of the MOSFET

The IC is a dual device, but as only one comparator is used here, the other is shown greyed out. On the actual veroboard, the unused pins were folded under the body of the IC out of the way (It's good practice to solder unused input pins together). The reason I did this is so the copper tracks can pass from one side of the board to the other unhindered. It just makes designing easier. I fed the veroboard via a small fuse in case of component failure, though I suspect veroboard tracks would make good fuses anyway!

The battery is housed in the shed along with the charge controller which is mounted on a wooden board (an old pine shelf). This also serves as a central hub where all the cables converge and connect from the solar panel, battery and various loads. See photos below


The left hand photo shows the mounting board with, from left to right, an electronic timer, battery cut off switch/fuse and the enclosure that houses the control circuit. At the top is a heatsink with the schottky diode and MOSFET bolted onto it. As yet this hasn't got very warm at all, but I should wait until summer before making further comments! Below that is the 2.2 ohm 100 watt shunt resistor. The right hand photo shows the cover removed from the control circuit. The multiturn preset is accessible through a hole on the underside and the test point for setting the battery full voltage is located at the lower end of the veroboard

The positive supply cable from the battery goes straight to the fused switch before it goes anywhere else. A normal mains type is used here which is perfectly adequate for low voltage systems, plus also, domestic fuses are cheap and available everywhere. The terminal block at the bottom connects everything together with the solar panel entering at the right and battery connected at the left. In between are the outputs, one for a shed light, the others for the security light and house. Only the security light is switched on and off by the timer. Power is carried from the shed via a large twin & earth cable. The two main conductors feed the house, while the switched supply for the security light is carried down the earth (with negative as common). The two supplies are separated inside a weatherproof junction box

The shed light uses a 12 volt CFL bulb. These are relatively expensive to buy, but are nice and bright. Note that CFLs come in various colour temperatures ranging from 2700K (warm) to 6400K (cold white). Personally I absolutely cannot stand 6400K types as they give a horrid light reminiscent of being in a nightmare! The warmer the better for me. The one shown here is 4200K (mid white) which is still not exactly what I'd like to relax under, but it's fine for shed lighting. There are high powered versions available but I tend to stick to about 11 watts to keep the current drain at a reasonable level.
I must just mention the timer. Electronic timers that work on 12 volts are pretty rare, but a good alternative is to use a battery powered programmable thermostat. I take no credit for this having discovered it on the REUK website. Basically, the thermostat is modified by replacing the thermistor with a 10K resistor so that the temperature always displays around 25˚C. Then you just set the on time at 35˚C and the off time at 5˚C (the upper and lower limits) and it will act as a timer. And as it runs off a couple of AAA batteries it doesn't drain any current from the system

Safety: I tested the integrity of the cabling by increasing the fuse rating at the battery and then deliberately short circuiting the supply at the farthest point. The fuse blew easily. It's interesting to note that the solar panel is still regulated by the charge controller and will continue to provide power even when the battery is disconnected, though the available current depends on the amount of light of course. This may not be a good thing for some types of equipment so it's something to be aware of. The farthest point is also fused for safety


It's coming up to a year now since my small solar power system has been in service and it has performed superbly. The only modification recently carried out was to add a 1u capacitor from the gate of the MOSFET to ground. This was to cure a ticking noise caused by the MOSFET switching on and off that could be heard from audio equipment connected to the system during day time charging. The charge voltage stabilisation is much improved too. Also, instead of using the test point, and with sun permitting, I've found that setup can be achieved more easily and accurately by connecting just the solar panel without the battery or load and then adjusting R2 for 13.8V across the battery supply rail (BATTERY+/- on the veroboard). This also overcomes any slight error caused by differences in the value of the two 10K resistors (which by the way are now 47K for no particular reason!)

I've created another page that describes just the charge controller (with updates) which can be found here