Nokero’s Shavano N220 outputs 29 lm on its high setting, 14 lm on low, and when fully charged, it will run for 3.5 hours on high, 6 hours on low. You choose between high, low, and off with a 3-way switch. The light also has a photosensor that automatically turns it off when there’s enough light to charge the battery and presumably the LED doesn’t need to be on.
The light is housed in a package reminiscent of an oversized incandescent light bulb. You can hang it outside in the sun on a line or a hook while charging from a clip that’s attached to the base of the light, which looks like the screw end of a light bulb. The light’s solar cell is the SunPower Maxeon.
The N220 has a single 1/3W LED that might be a Seoul Semiconductor Top View 5630. (For a slightly more detailed look at the LED and battery placement, you can go to the Designing with LEDs website.) The prominent ICs on the board are a 6-pin IC that looks like a switching boost regulator, and an 8-pin chip, most likely a microcontroller. Neither of the ICs have any markings (see Fig. 3).
The two 1.2V 1200 mAhr NiMH batteries are in series and provide 2.4V to the 6-pin IC – the one that I’m assuming is a constant-current boost regulator, which operates at about 1.1MHz and boosts the voltage up to 3.2V. When switched to high, the LED is powered at 3.2V and 110 mA. The specs for the N220 say it will put out 29 lm at this power, which matches with the specs in general for a 1/3W LED such as the 5630.
When switched to low, the voltage across the LED changes to a 50-50 duty cycle at about 640Hz: During the “on” portion of the cycle, the voltage is 3.2V and the LED turns on, while during the “off” portion of the cycle, the voltage falls to 2.6V, which is below the LED's Vf and the LED turns off. The overall current drops to an average of about 50mA or half-power. It looks like the 640Hz signal is generated by the 8-pin IC, which I’m assuming is a microcontroller.
This is a neat way of dimming the LED, much better than simply dropping the analog current through the LED which is a less efficient use of power and can result in a change in the LED's color temperature.
The microcontroller itself didn’t match up with any of the 8-pin versions of the usual suspect microcontrollers. It’s likely that it is one of the special-purpose uCs made in China just for controlling LEDs. The huge size of the large garden solar light industry enables a lot of ASICs for LED control that couldn’t otherwise justify a custom IC.