I've been helping put together a small solar set-up for my father, and he wanted a light connected to the regulator to give a bit of light at night outside. I thought the best thing for the job would be a white indicator LED that's used on a car or truck, that way it's already waterproof. So I ordered a cheap one off ebay and it was fine. It did the job perfectly and gave off plenty of light, but I was curious as to how it worked. The specs claim it operates from 12 or 24 volts and I wanted to know how it went about doing this.
|12/24 Volt 2 LED White Indicator Light|
|4 terminal LEDs under the diffuser|
Even though the PCB was encapsulated to protect the circuit from water, it was done with a clear resin, so although we can't probe the board directly it can at least be mapped out. At first glance it was obvious that it was going to use linear regulation. There's only a handful of components on the board, a 39 ohm resistor, capacitor, diode, and T0-92 package that was only partly encapsulated in the resin which isn't ideal, but I think it's for a reason that I'll come to later. There aren't any components you'd expect to see in a switched power supply.
|Reverse Engineered Schematic|
|LM317 Precision Current Limiter Circuit|
Vin = V(diode) + 2*V(LED) + I*R + V(LM317)
10.25 = 0.7 + 2*V(LED) + 0.033*39 + 1.7
V(LED) = 3.3 Volts
An LED forward voltage of 3.3 Volts agrees with my experience with white LEDs. As the supply voltage changes in the regulated region, the current remains constant, as does the voltage across the diode, LEDs, and resistor. This voltage is equal to:
V(diode) + 2*V(LED) + I*R = 0.7 +2*3.3 + 0.033*39 = 8.6 Volts
Therefore the voltage across the LM317 is the supply voltage less 8.6 Volts. This means that the waste power dissipated in the regulator as heat in Watts is equal to:
(V(supply) - 8.6) * 0.033
|Current Consumption of an LED Automotive Indicator Light|
Now that's all well and good, but it's a little wasteful. If the supply voltage is 12 volts it's not too bad, the power dissipation is equal to (12-8.6)*0.033 = 112 mW, but if the supply voltage is 24 volts the power dissipation is equal to (24-8.6)*0.033 = 508 mW. That's getting a little high, particularly when you consider the thermal constraints on the design.
The thermal resistance, junction to ambient of a T0-92 LM317 is 160 degrees Celsius per Watt. At 24 Volts the device dissipates 508 mW of heat, this means the junction is 0.508 * 160 = 81 degrees Celsius above ambient temperature. In Australia a device like this might be operating on a hot day mounted to a material that is hotter than the air temperature. This means that ambient temperature might be around 50 degrees Celsius, and the junction would be at 131 degrees Celsius, which is above the junction operating temperature of some versions of the LM317. It's borderline, but when you then go ahead and encapsulate everything in resin this could make things worse depending upon the thermal conductivity of the potting compound. I think that's why they left half of the LM317 package protruding from the resin.
It's fine for the 12 Volt supply I'm using, but I wouldn't expect it to have a long life at 24 Volts. It would have been nice to see something more efficient used, but for a cheap low power device you can't expect much.