Hi, Greg.

Found something in the ARRL Handbook about adding a capacitor across a
load to get an impedance match. Turns out they call this arrangement an
L-Network. They list 2 kinds, the 1st where the load R is less than the
source Rin and the other where R is greater than Rin, which is my case.

They state some equations for finding XC and XL with respect to the
transformation ratio between Rin and R. However, one equation falls
apart when Rin = R because it divides by zero and I forget how to get
rid of that term. More on this after I write them here:

:       L-Network
:
:    o----XL---+-----.
:              |     |
:             XC     R
:              |     |
:    o---------+-----'
:
: if Rin < R
:
: XC = R sqr[ Rin / (R - Rin)]         <--- division by 0 when R = Rin
:
: XL = (R Rin) / XC
:
: Q = XL / Rin ,  Q = R /XC
:
: Q is deterimined by the ratio of the impedance and cannot be selected
: independantly.

Accordingly with the parameters of the dynamo:

Rin =  1 ohm         <--- I forgot to measure this value
  R = 12 ohm

The transformation match occurs at a frequency of 18.947 Hz.
 Q = 0.5
XC = 24
XL = 0.4

C turns out to be 350uF but it exhibits an almost 5 db peak near 300Hz.
Also a very large current flows in and out of the capacitor which I'm
afraid may heat the windings by the same amount as the lamp since after
all it is a real maximum power transfer! i.e. the source dissipates as
much power as the load or 3 watts each for a total of 6 watts, makes
sense...

That's definitely not what I was looking to do with the bike light. My
aim was to maximize power to the light while minimizing the power lost
in the dynamo windings. I guess I was wrong after all. I am really
just looking to minimize voltage loss in Rin.

 Mike
 ****

... Dunno if we'll get that past the CSA und UL 'owever.
--- Blue Wave/DOS v2.30