Switching power supplies in general are thought to be likely to
explode. This is true. They are also thought to be hugely
complicated and kinda magic. This is not true. In fact,
it's possible to build a switching power supply that is
stupidly simple, and simply stupid.
I needed a cheap and easy way to convert 3.6V from
lithium-ion cellphone batteries to 5V to drive some logic
chips, and didn't have any cool
all-in-one-switching-power-supply chips from
National Semiconductor this
time. I did have leftover parts, a resistor kit,
and the kind of determination that can only be called
The resulting hack will convert about 100ma of 3.6V DC power to
about 50-60ma of current at 5V. There's no way to boost the
current much more without replacing the 555 and probably the
TIP120 with faster components.
The 555 is wired up to make very narrow negative pulses at
60KHz, which are converted to very narrow positive pulses by
the 2n2222*, which rapidly switch the big TIP120 transistor on
and off. The TIP120, when on, sucks power straight from +3.6V
to ground through the coil. This is only not a dead short
because coils take a little time to start conducting.
...And once they start letting power through, they KEEP
power moving until something makes it stop. Which means, when
the TIP120 switches off, the inductor keeps going for a bit
and shoves power through the diode into the huge capacitor on
the other side. This is where the voltage boost happens, the
inductor gives a higher voltage than was put into it -- at
less current. No free lunch.
This is stupid because it's inefficient -- the zener diode on
the output wastes power to regulate it to 5V. A smarter
design would just slow down the oscillator, but I was lazy.
A few final notes: The MBR745 is just a fancy
really-efficient diode. Any "shottky" diode work if it can
handle the power. That unlabelled resistor in front of the
TIP120 is 10K. The inductor is actually 20uH, not 1.
* Just about any small NPN transistor will work.