Sepic

The diode marked in blue shouldn't be there. It's wasting power.

http://everycircuit.com/circuit/5612498957107200

Yeah, I took that into account when I built this but never had the chance to delve into fixing that, I work a crazy job now, on call 24/7/365 so my circuits might be slow on getting updated.

I was going to do a flywheel instead like you did, and this may sound stupid, but what is the 220 ohm and cap for, smoothing?

These circuits tend to induce a hell of a lot of RF noise and EMI. Just check my circuit. Remove the RC filter, and check the pulsed voltage when it settles at 2us/s scale. You'll see very well defined ringing. The coil forms a tuned circuit with some stray capacitance in the circuit, and starts ringing. These harmonics are very EMI rich and need to be scrapped off. This is what the RC filter does. Add it again to the circuit and make the same experiment. No sign of ringing. Also I do recommend you use a galvanic separation such as a small RF transformer instead of an inductor, because the ground rail (if not properly designed) will become very noisy and will probably interact with the circuit you wish to power with this device. With a transformer you can separate the RF ground plane from the ground rail for the output. I'll post an example.

For instance check this out. A galvanic separation allows you to separate the two ground rails of the output, from the switching circuit. This is very beneficial when you have less experience with PCB's and even an experienced designer often chooses this method. EC doesn't work with no connection to ground, so I had to ground one part of the output through a small value capacitor to ground. The capacitor doesn't conduct any current, so essentially the two ground rails aren't connected in any way. http://everycircuit.com/circuit/5612498957107200