We bought a number of solar light lanterns and ornaments for our garden. They were to cheer up our garden during the winter nights. The initial display they gave was to our satisfaction but then their need for constant attention made them more trouble than the pleasure they gave.
The main problem was that the battery charge energy provided by the solar panels during the short winter daylight was considerably less than that used during the hours of darkness and they remained on long after we had retired to bed and were often on when we awoke. A supply of higher capacity NiMh batteries and a suitable battery charger eased the problem but by no means was a satisfactory answer.

As time went by almost as bad were the failures. Switch and battery connections and even solar panel connections would corrode and cause failures. Sometimes the problem was with the circuit board and on rare occasions with the LED lights themselves. I would reverse engineer the circuit as best I could and try to repair them and replaced them when I couldn’t.
Most used the solar panel as a dark detector but the odd one used a light dependent resistor. Some had back up non rechargeable batteries that I did not replace. The troublesome switches that provided on, off and sometimes flash or twinkle I did not want. Circuits varied in complexity. Some were simple like that shown above and used through the hole components; others were more complex and used smaller surface mounted devices (SMD’s) soldered onto it with chips under a circular blob of resin on the board.

Similar in action to the YX8018 chip above is a QX5252F and I used it to make my own simple circuit boards. The main difference between it and the YX8018 is the pin connections and the solar panel voltage votage which is relative to ground rather than to the battery voltage. In both cases the solar panel charges the battery via on chip diodes that prevent the battery back feeding the solar panel in hours of darkness.
In neither circuit can the 1.2 volt battery voltage alone turn on the LED light because these LED lights need forward voltages of between 1.8 and 3.0 volts to turn them on. White LEDS need more voltage than coloured LEDS. The job of the chip and the external inductor are to provide a voltage that will turn on the LED(S). The oscillator on the chip is switched on and remains on during darkness hours when the solar cell voltage is low. When on it rapidly opens and closes a switch that connects one side of the coil inductor to ground.

Those of you who have read my blog on magnetism will know a current flow through an inductor delivers photon energy into its surround space (switch closed). When the current flow tries to fall (switch open) the surround energy returns to the inductor and pressures the electrons in it to try and maintain the flow. That pressure may be regarded as a voltage pulse within the inductor and in these solar circuits it adds to the battery voltage. Because the total voltage now exceeds the forward voltage of our LED it is able to send a pulse of current through it.
The solar circuits you see above when accompanied by smoothing capacitors are often used for converting a dc source to a higher dc output. They are often referred to as a “joule thief” circuit, joule being the unit of energy. Whilst they may deliver higher pulses of output energy in no way can their output energy exceed that input from the battery or other source.

The oscillator in the QX 5252F operates at about 75 kHz, turning the switch on and off 75,000 times a second. Our eyes see no flicker. Even 24 frames a second our eyes see as continuous video.
By changing the value of the inductance you can change the current taken by the LED and hence its brightness. Higher inductor values deliver less current and lower brightness but pleasingly longer battery life. I varied the inductor used to suit the LED(S), be they white or yellow. On my board shown the resistance like inductor is 370 microHenries.
Whilst the above simple circuit proved to be less troublesome and less energy consuming than the bought circuits I was still regularly changing batteries and charging them. My next blog is about a garden solar light project that has reduced my involved time considerably.
Good sound logic and reasoning. As to the solar cell, seems to me that if the new technology LiFePO4 battery may also be a solution. Its nominal cell voltage is 3.2 and is not subject to thermal runaway as are the Lithium ion batteries. LiFePO4 can function in temperatures up to about 300 degrees F but the chemical compound for this battery will allow it to deliver full current. No oscillator or coil would be needed, just use the solar panel as both a light sensor and source to turn on the battery. Only problem now is that there are not many LiFePO4 chargers available. I built my own but if you do the same, be very careful because this battery (cell) can deliver its full rated mahr rating in about 1 second and they typically rated at 400mhr. So if it can do .4 amp in a second and is rated in mahr then theoretically it can deliver 11520 amps in second. That would equate to about 36000 watts theoretically but in reality would likely explode or catch on fire or just flat out melt down with much heat..