"John Larkin" wrote in message
...
On Sun, 15 Jan 2012 11:25:07 -0700, Jim Thompson
wrote:

Here you go...

http://analog-innovations.com/SED/Bi...ubSwitcher.pdf

I figure, if you had a mind to, you could add regulation by running
this dude in burst mode, by stopping the 555 via its RESETbar input.

Diode D7 prevents some really weird behavior from back conduction. If
there were a need to tighten up the efficiency, I'd replace it with
another P-channel, body in the opposite direction from MP1 (for lack
of finding a suitable transistor, modeled as a 0.3 Ohm device
extrapolated from my vast CMOS device library :-). Then I'd use a
comparator to make the drop like a near ideal diode as in...

http://analog-innovations.com/SED/Pe...rIsolation.pdf

My experiments with using an auto-transformer would indicate that the
current capability of the hub dynamo (at least as presently modeled...
I'm suspicious of its accuracy)) falls off rapidly, almost acting like
a (near) constant power device.

I'd ask one of the local "pansies in spandex" to make experiments with
their hub dynamos, but I'm their enemy for suggesting publicly that
they should be required to be licensed, and be fined when they go
riding three-abreast down the middle of a thoroughfare, rather than
single-file in the bicycle lane :-)

...Jim Thompson


Good grief, this is just a fixed-duty-cycle nonregulating buck
switcher. I suggested this architecture days ago.

The 555 drains the battery all the time. If it doesn't, it won't start
up. And there's probably ways to blow out the gate of the p-fet, which
would take out the n-fet as a bonus. Amateur stuff.

You spent four days on *this* ?

My thoughts are along the lines; a couple of comparators flipping a bistable
with a 3rd comparator as a UVLO, that's to prevent the buck from latchin up
in 100% MSR when it draws enough current to drag the generator voltage below
its optimum load curve.

Not put pen to paper yet, I'd rather find time to fully study the tables &
graphs from the link Marcell posted first.