Joule Thief

[MarkE]

The PNP is a flyback topology.  The NPN is a boost topology.

In each current builds in the coil while the transistor is on.

In each when the transistor turns off the coil flies back until the LED latches on carrying the current.

Because the battery continues to supply current in the NPN boost case, the inductor current does not decay as fast as the PNP flyback, and it can generally deliver more power than the PNP flyback.

[Dave45]

The PNP is a flyback topology.  The NPN is a boost topology.

In each current builds in the coil while the transistor is on.

In each when the transistor turns off the coil flies back until the LED latches on carrying the current.

Because the battery continues to supply current in the NPN boost case, the inductor current does not decay as fast as the PNP flyback, and it can generally deliver more power than the PNP flyback.

Yes the high side driver delivers more power at the expense of the battery but the low side makes better use of the electrons in a short burst because of the electron recirculation.
It reminds me of tapping a wheel to make it spin.

[MarkE]

Yes the high side driver delivers more power at the expense of the battery but the low side makes better use of the electrons in a short burst because of the electron recirculation.
It reminds me of tapping a wheel to make it spin.

The efficiency of either configuration depends on several factors.  The most important are:

The inductor resistance.
The voltage drop of the transistor while on.
The turn on and turn off times of the transistor.
Transistor base current.

All things considered equal, the boost topology is usually more efficient.  Boost or flyback can be done with PNP or NPN transistors.  Faster and higher current handling NPN transistors are easier to make than PNP transistors. 

[Dave45]

All things considered equal, the boost topology is usually more efficient.

Is it because the high side driver oscillates whereas the low side doesnt.
It would seem the low side driver would be more efficient at driving leds.
The high side would be more efficient at driving a transformer.

Its interesting I really dont have the equipment to do proper measurements, no scope, but Im still going to setup the low side setup and see if I can get it to oscillate using a tank circuit.

[MarkE]

You've got to get yourself some sort of scope so that you can see what happens.  A Hantek 6022 that plugs into a USB port on a computer is $65.  Old analog scopes on eBay are $100.

LED or transformer, it does not matter.  You can get faster, lower voltage drop NPNs than you can PNPs..