A Solar Patio Lantern Teardown

Some years ago, we acquired a bunch of solar patio lanterns, like this one.

They all eventually failed, so I thought I'd look into their construction, and see if I could revive the things. At the very least, I wanted to see how they work, and what the failure mechanism(s) is/are.

The upper cap of a lantern has all the works in it. It just twists off.

On the top of it, there's the solar cell that soaks up daylight and charges the unit's Ni-Cd (nickel-cadmium) cell, and the face of a photoresistor that triggers the lantern's circuitry to turn it on in darkness.

Underneath, remove the three screws and you have access to the unit's innards.

And there we have an AA 1.2 volt Ni-Cd cell, a white LED (light emitting diode) and a tiny PCA (printed circuit assembly) with a four-pin IC (integrated circuit) on it. That IC (type No. ANA608) is the key to the lantern's operation. A little background is in order.

LED Forward Voltage

All semiconductor diodes exhibit a slight voltage drop (V_F) in their forward-biased state. All LEDs operate with forward bias in order to light. The applied voltage must exceed the V_F figure in order for an LED to light. Following is a list I drew up of forward voltage drops, as displayed by the diode test function of a common digital multimeter, for a selection of diodes. The coloured LEDs are just miscellaneous ones that I had on hand. The white LED is one from a patio lantern.

1N34A Germanium Signal Diode: V_F = 0.4 VDC
1N4001 Silicon Rectifier Diode: V_F = 0.6 VDC
Red LED: V_F = 1.78 VDC
Yellow LED: V_F = 1.85 VDC
Green LED: V_F = 1.86 VDC
Blue LED: V_F = 2.5 VDC
White LED: V_F = 2.6 VDC

So, a 1.2 V Ni-Cd cell has insufficient voltage to overcome the V_F  of an LED and light it, yet the lantern works. A clever bit of circuitry known as a 'joule thief' is responsible for that.

The Joule Thief

See this Wikipedia entry for a brief introduction to the joule thief circuit. The four-pin IC (ANA608), in conjunction with a 220 microhenry inductor, constitute a joule thief that converts the Ni-Cd cell's 1.2 VDC output to a 3V peak-to-peak pulse train that can light the LED. Here's a view of the pulse train at the white LED's anode.

The pulse train's period is about 6 microseconds, which works out to roughly 167 kHz. The LED is actually 'flickering' at that rate, which of course is way beyond the persistence of human vision, so the LED appears to be constantly lit.

The ANA608 IC

The only data sheet I could find for the ANA608 is mostly in Chinese.

Anyway, I drew up a crude schematic of the lantern's circuitry.

So there's what makes solar patio lanterns light up.

As for reviving dead ones, good luck. Here's what I've observed about failure mechanisms from a sample of five units.

• The Ni-Cd cells are long lasting and not prone to failure. The cell holder terminals stay remarkably free of oxidation. I suspect that battery failure, be it real or apparent, is mostly brought on by solar cell degradation/failure. Insufficient or non-existent Ni-Cd cell charging leads to lantern failure from what appears to be a bad Ni-Cd cell. It's possible to check solar cell output with a voltmeter. A good solar cell will put out about 3V under full, bright light.
• One case of water incursion past the solar cell's edges. The water got onto the PCA and fouled its operation. When dried out, the lantern worked again. I sealed the edges of the solar cell and the photoresistor with transparent auto sealer from Canadian Tire. That worked for a while, then the photoresistor went flaky -- its dark resistance wouldn't go high enough to enable the light to switch on.
• Broken off wiring connections. These tend to be self-inflicted when one fiddles with the innards of a lantern; the stranded wiring material used in the lanterns won't tolerate much flexing without breaking at solder joints. Some breakages will be repairable. Some will be buried in hot-melt glue and won't be repairable.
• Several solar cell failures. The solar cells used in these things are the cheapest possible, and the solder connections to them are iffy at best. There's no repairing such failures. The solar cells are not non-destructively removable.

* * *

A Candy Tin Solar Patio Lantern

Since I had some salvageable components left over, I thought I'd see if I could construct some serviceable patio lanterns from scratch. The solar cells and photoresistors from the old lanterns couldn't be salvaged, so I ordered up replacements from Amazon.

A candy tin could serve as a weatherproof housing.

And here we are with a breadboarded lantern.

A Complication

The photoresistors from Amazon wouldn't work as direct replacements for the original photoresistors. I had to add a PNP transistor and a 2.7 kohm bias resistor per the following schematic.

Without the schematic of the ANA608's innards, it's not easy to see how that bit of circuitry works, but it does.

Anyway, here's a finished candy tin patio lantern out in the dusk.

It won't win any design awards, but it works quite nicely.

* * *

Enhancements -- MONDAY, APRIL 16, 2018

There's a way to have the LED switch on and off as it should without the use of a photoresistor. If the IC's 'enable' pin, pin 2, is tied directly to the positive terminal of the solar cell, then the solar cell's output will control the LED's switching. That strikes me as a good way to go, since it eliminates a component or components that can fail.

Another enhancement is the addition of a rectifier and a filter capacitor to the output of the joule thief. A slight (very slight) gain in LED brightness is obtained from that.

Here's a revised schematic showing the installation of both of the above mentioned features.

Problems So Far

• One of the photoresistors from Amazon failed. It developed an almost steady resistance of about 80 ohms, and so would not switch on the LED in darkness. I got the lantern going again by tying the IC's 'enable' pin directly to the solar cell's positive terminal.
• Water Incursion. We've been awash in rain, freezing rain and ice pellets here lately. The slip-fit lids of the candy tins aren't weatherproof; they wick up water and allow it to enter the tins. I'll have to dry them out thoroughly and tape the lids' perimeters with electrical tape.

- - -

Update -- SATURDAY, MAY 5, 2018

Two things:

• Another photoresistor failure. Another one of my photoresistor-equipped lanterns wouldn't come on in the dark. I again tied the IC's 'enable' pin directly to the solar cell's positive terminal, and the lantern is working again. I see nothing to be gained by having a photoresistor control the circuitry. Just use the solar cell's output level as the 'enable' control, and the lanterns work fine.
• The candy tin lanterns make good little night lights in a pinch. We had a prolonged power outage here that began in the late afternoon, and went well into the night. A couple of candy tin lanterns lit up a bathroom adequately for tooth brushing and the like.

 # # #

FEEDBACK

# # #