The occasion for this project was retrofitting an illuminated wall picture with LED backlighting. As soon as I applied LED power, the brightness blinded me — much brighter than the original T20 fluorescent tube, so I knew immediately that a dimmer control was required.
Schematic of the LED Brightness Control Circuit
How do the Chinese make any money at only $4.60 per reel (free shipping)—and the quality is good! (Note that there has been a modest price increase since…)
The LEDs come on a 5M reel with a total of 300 devices (100 series strings of 3LEDs per string). Measured current was 1A @ 12V for an input power of 12W. Since my wall wart power supply was rated for 0.8A, I pared it down to 240 devices (every 3 devices, there is a cut point indicated).
Assembly was simple. I was having a hard time determining how I was going to glue down the LED strips, but finally realized that it already had an adhesive backing.
I opted for a simple adjustable current regulator. While adjustable voltage works, the limits of voltage are not clearly defined and there can be a lot of dead band on the adjustment. The current regulator, on the other hand, has good range of adjustment, low dropout voltage and minimal dead band.
The power device is a composite PNP /NPN pair that acts as a high gain PNP transistor. This could also be done with a PNP Darlington, but the dropout (saturation) voltage would be a little higher. This function could also be done with an NPN /PNP composite pair, but I wanted the LEDs to be grounded to the negative bus for possible future enhancements.
It is essentially an emitter follower with a 0.6Ω emitter (shunt) resistor. The voltage across this resistor is determined by the base voltage at Q1. The collector (LED load) current is essentially the same as the emitter current. Since the voltage at the base of Q1 is adjustable, the load current is also adjustable. Schottky diode D1 provides a fixed 0.3V drop in order to reduce potentiometer dead band. A standard diode could prevent Q1 from turning off completely—this is important because the potentiometer is also employed as an On /Off switch. No, it does not completely remove power from the circuit, but the losses in the Off mode are far less than the magnetic and winding losses of the continuously operating wall wart.
C1 is a bypass capacitor that helps prevent oscillation—yes, a circuit like this can oscillate—I connected a scope to it to prove to myself that it was stable.
C2 filters out the AC wall wart ripple so that the load current is pure DC. I wanted pure, flicker-free light. Many LED circuits pulse the current—I wanted to avoid this.
Scaling for other currents & input voltage
Regulating the load current also regulates the output voltage, so the raw input voltage may be higher than necessary and need not be exactly matched to the load. However, additional voltage causes additional power dissipation in power transistor Q2. In my case, only a small heatsink was required and it ran comfortably warm under worst conditions. Specific circuit voltages are indicated on the schematic. For other current ratings, only R3 & 4 need be scaled (drops 0.5V @ full load)(may be combined into one resistor). R2 must be scaled for the raw input voltage at full load.
I had the 2N3906 (PNP TO-92) and D44H8 (NPN TO-220) devices on hand, so that is what I used. Transistor selection is not critical.
For the future
Light alternate rows in pseudo random fashion to make it shimmer (if possible)—original picture rotated a wavy filter to make it shimmer.