Anatomy of a Chinese Solar Spotlight kit

Recently we purchased a set of Solar Spotlights from Bunnings to light a pond we added to our garden.

The kit consisted of a pair of LED spotlights, each with 9 white LEDs in the housing.
These attach via 5m long cables to a base unit that houses 3 AA NiCad batteries,
a small solar panel and a light sensor so the lamps would only light during darkness.

Bunnings: http://www.bunnings.com.au/products_product_solar-magic-spotlight-set-of-2-with-remote-solar-panel_2076.aspx

I was initially skeptical of the ability to provide useful light for any length of time, and this was soon proven in practice.

So it was then time to take the system apart to find out what makes it tick.

Spotlights

It took a little while to work out how to get into the spotlight units, but once the secret was unlocked it was dead easy.
Access was gained by twisting the front lens holder using the small spigots moulded into the face.
A short twist released the bayonet fixing, revealing a silicone rubber gasket and clear plastic cover and reflector assembly.

The rear of the reflector holds a small PCB upon which the LEDs mount.

Then the first two shonky Chinese design elements were unearthed:
                1/ all 9 LEDS were directly connected in parallel with no load sharing resistors.
                2/ a lump of cheap Chinese steel was fitted to the rear of the housing to give some “weight” to the unit.

The first aspect requires the LEDs to have very similar characteristics to evenly divide the current amongst them.
In practice this seems to be OK as all LED’s light at apparently equal brightness.
Connecting to a lab power supply confirmed the absence of limiting resistors as the current dramatically increased as the volts were raised above 3.5V, the nominal forward voltage of a white LED.

The second aspect, well perhaps some people judge quality by how much something weighs!

The Base Unit

The base unit superficially looks the part, an array of solar cells approx 125mm square cover the top, and a small photo sensor is evident in the top right corner.

The spotlights attach below the base unit through some small DC coaxial connectors.
A slide switch enables the spotlights to be disabled, but still allow charging.

Four screws removed from the bottom of the unit reveals the internals.

First obvious aspect is the battery holder could hold 6 AA batteries, but only 3 were fitted.
Perhaps this allows a larger capacity option?

The top is connected to a small PCB via wires that are “adequately” long enough.
ie you can just lay the two halves down side by side.

Base Unit Circuitry

It was then decided to trace the PCB circuit.

Visually the major components are a couple of transistors, a switch, DC connectors and a hand full of resistors.

The traced circuit is below:

Now the true evilness of Cheap Chinese “Design” was revealed.

1/ no limit to battery voltage during charging
2/ no current limiting to LEDs

No charge limiting

All that exists between the solar array and the 3 NiCad cells is a blocking diode.
The function of the blocking diode of course is to prevent battery discharge back via the solar array in darkness / poor light conditions.
The solar array uses 12 individual solar cells which means the available voltage is about 6V.
If the switch is left in the disable mode for long enough (open), I’m certain the NiCads would begin venting with such an excessive source voltage.

LED drive

Unfortunately the bad design already found in the spotlights themselves is perpetuated here.
You may recall each spotlight was found to simply be 9 LEDs in parallel with no limiting resistors.

When connected to the base unit, the final transistor’s collector load becomes simply 18 LEDs in parallel.
Other than the transistor itself, no other componentry exists to limit current from the battery.

It is obvious that the designers rely upon the voltage of 3 NiCads in series and the forward voltage of a white LED are very similar.
This also, by pure luck more than good management, prevents the batteries being discharged too deeply.
Once the battery voltage drops below the LED’s conduction voltage, current drain is minimised.

Light Sensor

Surprisingly, the light sensor circuit actually works correctly.
In daylight the Light Dependent Resistor lowers in resistance and allows base current to drive into the first NPN transistor.
This action robs current from the second transistor’s base and thus turns off the second transistor.

Upon darkness, the first transistor will not receive enough base current, allowing the second transistor to then turn on, and thus enable the LEDs.

Run Time Calculations

Let’s have a look at the operating parameters.

From the lab supply, it was determined each spotlight would draw approximately 200-300mA at 3.5-3.6V.
So the total current drain using both lamps is ~500mA.

The Nicad cells have a claimed capacity of 1000mAh, but the endpoint voltage for this capacity may be below the conduction voltage of the LEDs.
Halving the claimed capacity, and dividing by 500mA gives a run time of around an hour, which was observed in practice.

Hardly good enough for a late night soiree!

The design of the spotlights was aesthetically pleasing so I decided it is time to give this thing some balls by souping up the solar panel and battery system.

Adding Balls

First obvious thing to go was the battery and charging system, they were simply well below expectations.

It was however decided to retain the NiCad batteries and charging system, albeit adding a 3.6V zener diode across the battery to shunt excess charge current.
The original power supply would then simply drive the photo sensor circuit, leaving the final transistor as an open collector to sink an external 12V system.

It was decided to source a 500mA “BuckPuck”. 

These devices are designed especially for driving power LED’s such as the Luxeon and Cree devices.
They are a compact integrated buck regulator that provide a constant current instead of a constant voltage, perfect for LED applications.
More information can be found on the LuxDrive website: http://www.luxdrive.com/luxdrive-products/buckpuck-3021-3023-led-driver/
Recom are now also supplying functionally equivalent devices: http://www.recom-international.com/innoline-led.html

With a 12V supply, the current drain using the BuckPuck to drive the pair of spotlights was around 180mA.

Battery

I had at hand a 12V / 17Ah SLA battery from an old UPS.
A measly 180mA drain on such a beast would easily keep the LEDs lit all night long, and a few more nights to boot without ever seeing a charger.

Solar Panel

I had a 12V / 10W solar panel which could provide ~600mA in full sun with a simple direct battery connection (via a blocking diode).

Simple Shunt Regulator

A real simple shunt regulator was bashed into shape using bits to hand.

The power transistor will begin to conduct if the total voltage rises over 13.65V.
If it turns on too hard, the solar panel’s voltage will lower and thus the transistor will back off.
To enable the transistor to handle the expected 8W or so of shunted power, it is mounted to a reasonable sized heatsink.

Final Circuit

The final circuit is shown below:

The second NPN transistor was up-rated to an MJE800 after the original 2S2050 fused into a single blob of silicon on the first night’s use (the lights didn’t turn off).

The new parts are somewhat spread about.
The 10W solar panel is on the house’s roof.
The 12V battery is about 10m away from the panel, and 10m away from the base unit. The Shunt regulator is beside the battery.
The base unit PCB was hacked about and now houses the BuckPuck.

Conclusion

The hacked solution works very well, providing light from the spotlights for an entire evening.

Also the light output is increased above the original levels as the LEDs are now driven in a more linear mode than hovering about the LED’s knee / NiCad terminal voltage.

I called the original system “cheap” Chinese, but @ $50 it is not all that cheap in real world monetary value.
The internals though definitely fall into the cheap category.

Possible Future Mods

I will probably increase the size of the solar panel and battery.
No not for these spotlights especially, but I’m already using LEDs elsewhere within the house for task lighting.

Also all of my networking gear runs off a “12V” supply.
The total draw is about 1.2A @ 12V, or ~15W 24/7 so there’s a possibility to shave something of the household power bill.
Since all 3 devices (modem, router, switch) use the typical inefficient iron cored plugpack the mains power is no doubt over 20W continuous.