The circuit in Figure 1 employs LEDs in a series-connected chain as the stabilization elements in a voltage regulator. It makes use of the fact that a green LED’s current-voltage characteristic provides a drop of about 2 V. As a result, the user can pick from the voltages shown by changing the position of switch S1 to select which of the diodes D2-D5 are used. Transistor VT2 provides the regulator short-circuit protection.
One disadvantage with this circuit is that increasing the number of LEDs in operation decreases the devices’ light output. You can avoid this problem with the addition of R4, which allows all the diodes to glow simultaneously so they can provide bias lighting of the switch’s front-panel scale (Fig. 2).
Another improvement allows greater resolution of the regulator’s output voltage. Figure 3 shows the approximate voltage levels at the switch scale and actual values, which are determined by the magnitude of the load resistance as well as the individual properties of the device elements in the figure.
Using forward-biased LEDs (or diodes) singly or in series as voltage references is not a new idea. The circuit is similar to using a Zener diode for voltage regulation. This idea cleverly combines this voltage reference and voltage indication all in one. However, designers should be aware of its limitations.
First, LEDs have widely varying forward votlage drops even within the green color. There are light greens, dark greens, and all shades (wavelengths) in between. Also, green LEDs may use aluminum indium gallium phosphide (AlInGaP) or indium gallium nitride (InGaN) material, which will create different forward voltages. Green LEDs may be manufactured as blue LEDs with a phosphor coating as well. A quick search for green LEDs on Digi-Key shows forward voltages for green LEDs varying from 1.7 to 3.5 V.
Even two LEDs with the same part number and from the same manufacturer can vary in forward voltage. For example, the Vf range in a datasheet can be 2.2 to 2.6 V. An LED’s forward voltage also varies with forward current, which in this circuit depends on input voltage. You can minimize the change to the LED’s Vf if the regulator’s input voltage is constant.
An LED’s Vf can change with operating temperature as well. Most LED datasheets don’t even specify this parameter, but by looking at the datasheets one can speculate that it’s no better than 5% over a 25°C to 75°C range. Overall, we’re looking at an output voltage accuracy of about 10% or worse.
Second, the pass transistor’s (VT1’s) load-current capability depends on its base current, which depends on the number of LEDs used in series (that is, on output voltage). So, the regulator’s load-current capability can vary based on output voltage. VT1’s base current also depends on input voltage. Maintaining a constant input voltage is the only way to eliminate this variation.
Third, because of the LED forward-voltage matching issue, each regulator constructed will have to be calibrated separately. Since there is no calibration capability in the circuit, the voltage setting dial may have to be marked differently for each regulator that uses different batches of green LEDs.
Fourth, short-circuit protection is achieved by detecting when the output voltage is too low. Transistor VT2 conducts when the difference between its base voltage (determined by D1’s forward voltage drop) and emitter voltage exceeds forward conduction voltage of its base-emitter junction (VBEmin). Assuming D1’s Vf = 2.2 V and VT2’s VBE = 0.7 V, an output voltage of less than 1.5 V (2.2 V – 0.7 V) will turn on VT2. An output voltage of 1 V (as the table in Figure 3 shows), then, is not really achievable.
Finally, the maximum current through the “Zener” biasing resistor, R2, is 12 V/750 Ω = 16 mA. This current gets divided into the base current for pass transistor VT1 and the LED current. Consequently, the maximum load current will depend on VT1’s current gain and base current. So, select VT1 accordingly.