You've seen them when you're driving. Mailmen use them, snow plows have them. This is a schematic of the circuit inside.This was taken from a popular unit available at NAPA auto parts stores. It's amazing how simple it is for a PWM circuit.
WARNING: This circuit generates voltages and currents which will hurt you very badly and possibly KILL if you are not careful. If you don't know much about working with line (and higher) voltages or if you aren't crazy (like me) then DO NOT attempt to construct this circuit. I CANNOT BE RESPONSIBLE if you electrocute yourself to death! That said, let's have some fun!
It uses a UC3842 (here a TL3842 clone) pulse width modulator to regulate the charge on the main storage capacitor. In this respect is similar to a computer power supply. It has an interesting Zener booster circuit using an NPN transistor, shown on the left of the drawing. This keeps the supply voltage at or below 15V at all times. One great feature that this affords is a great variety of input voltages. It’s rated at 12-80V.
It's a pretty simple circuit with a MOSFET driving a coil. The flyback pulses are taken by a fast diode to the main storage capacitor. The charge is monitored by the PWM through a couple divider resistors. Since the storage cap is "grounded" to Vdd, the charge on it is dependant on the input voltage. It ends up being around 320V at a 12V input and 300V at a 32V input. The higher the input voltage, the lower the output. I haven't tested it at the full 80V, but it probably bottoms out at around 252V. Higher input voltages will give faster charge time, but lower overall charge. The charge current is also regulated by pin 3 of the PWM. This can be done because the MOSFET is in series with 2 parallel resistors to give 1V reference / .5 Ohms = 2Amps.
The trigger circuit is a self repeating relaxation oscillator using a SIDAC, which to those unfamiliar with this neat device, is similar to a DIAC. It's a silicon equivalent to a neon bulb (a negative resistance device). It has a set breakover voltage like a neon bulb or a DIAC, which in this case is around 150V. So the trigger cap charges to 150V, the SIAC breaks over into the trigger transformer to trigger the strobe. This is one of my favorite arrangements.
The IRF740 only requires a small heat sink due to the current limiting by the 3842.
You can see why the 3842 is such a popular IC for computer power supplies. It needs very few external components. Compare this with the TL494 which needs pull up/down resistors, etc. I suppose the 3842 trades in fewer components for reduced versatility.