The circuit in Figure 1 allows you to light any type of LED
from a single cell whose voltage ranges from 1 to 1.5V. This
range accommodates alkaline, carbon-zinc, NiCd, or NiMH
single cells. The circuit's principal application is in LED-based
flashlights, such as a red LED in an astronomer's flashlight,
which doesn't interfere with night vision. White LEDs make
handy general-purpose flashlights. You can use the circuit in
Figure 1 with LEDs ranging from infrared (1.2V) to blue or
white (3.5V). The circuit is tolerant of the varying LED voltage
requirements and delivers relatively constant power. It provides
compensation for varying battery voltage. The circuit is an
open-loop, discontinuous, flyback boost converter. Q2 is the
main switch, which charges L1 with the energy to deliver to the
LED. When Q2 turns off, it allows L1 to dump the stored
energy into the LED during flyback.
Q1, an inverting amplifier, drives Q2, an inverting switch. R4,
R5, and R2 provide feedback around the circuit. Two
inversions around the loop equal noninversion, so regeneration
(positive feedback) exists. If you replace L1 with a resistor, the
circuit would form a classic bistable flip-flop. L1 blocks dc
feedback and allows it only at ac. Thus, the circuit is astable,
meaning it oscillates. Q2's on-time is a function of the time it
takes L1's current to ramp up to the point at which Q2 can no
longer stay in saturation. At this point, the circuit flips to the off
state for the duration of the energy dump into the LED, and the
process repeats. Because inductors maintain current flow, they
are essentially current sources as long as their stored energy
lasts. An inductor assumes any voltage necessary to maintain
its constant-current flow. This property allows the circuit in
Figure 1 to comply with the LED's voltage requirement.
Constant-voltage devices, such as LEDs, are happiest when
they receive their drive from current sources. The LED in
Figure 1 receives pulses at a rapid rate. The inductor size is
relatively unimportant, because it determines only the
oscillation frequency. If, in the unlikely case the inductor value
is too large, the LED flashes too slowly, resulting in a
perceivable flicker. If the inductor value is too small, switching
losses predominate, and efficiency suffers. The value in Figure
1 produces oscillation in the 50-kHz neighborhood, a
reasonable compromise. D1 provides compensation for varying
cell voltage. By the voltage-division action at Node 4, D1
provides a variable-clipping operation. The higher the supply
voltage, the higher the clipping level, and the result is
correspondingly less feedback. Q1 inverts this clipping level to
reduce the turn-on bias to Q2 at higher cell voltages. We chose
2N3904s, but any small-signal npn works. Q2 runs at high
current at the end of the charging ramp. Internal resistance
causes its base-voltage requirement to rise. The R2-R1 divider
at Q1's base raises the collector voltage to match that
requirement and thus controls Q2's final current.
The LED's drive current is a triangular pulse of approximately
120 mA peak, for an average of approximately 30 mA to a red
LED and 15 mA to a white one. These levels give a reasonable
brightness to a flashlight without unduly stressing the LED. The
supply current for the circuit is approximately 40 mA. A
1600-mAhr NiMH AA cell lasts approximately four hours. L1
must be able to handle the peak current without saturating. The
total cost of the circuit in Figure 1 is less than that of a white
LED. You can use higher current devices and larger cells to run
multiple LEDs. In this case, you can connect the LEDs in
series. If you connect them in parallel, the LEDs need
swamping (ballast) resistors. You can also rectify and filter the
circuit's output to provide a convenient, albeit uncontrolled, dc
supply for other uses.