The method is described briefly in the article posted on my web site.

If someone gives you the self and mutual impedances of two elements and
the lengths of two feedlines connecting them to a common point, it's a
fairly simple matter to calculate the resulting element currents. What
the program does is work the same problem backward. That is, given the
self and mutual impedances of the elements (or in the case of the new
program, the feedpoint impedances when properly fed), it finds the
feedline lengths which will produce the desired current ratio *in those
two impedances*. Years ago I got interested in the problem and figured
out how to work it in that direction. It involves a bunch of variable
transformations to keep the equation sizes reasonable, and results in a
nice, closed-form solution. Some time after I came up with the method, I
spotted an article in Ham Radio magazine which described a solution
using an iterative approach -- it solved the problem in the easy
direction, looked at the result, modified the feedline lengths, tried
again, and so forth. That works (I had done it that way before figuring
out the closed form solution), but can be time consuming -- at least it
could with the computers of the time --, and you have to be careful with
the algorithm to keep the process stable.

I wrote to the author of the Ham Radio article telling him of my method,
and he suggested that I write it up. I had just gone through a lengthy
period of very unpleasant dealings with magazines and editors
(particularly Rich Rosen at Ham Radio), and had no interest in having
more of it. So he ended up writing the paper describing the method in
step-by-step fashion. You can find it in "A Voltage-Matching Method for
Feeding Two-Tower Arrays" by A. Christman, in IEEE Trans. on
Broadcasting, June 1987. Or if you're a masochist, you can try
untangling the GWBASIC spaghetti code in the original DOS programs.

Roy Lewallen, W7EL

Cecil Moore wrote:
Roy Lewallen wrote:

These phasings also provide a higher feedpoint impedance than the
W8JK, which results in decreased conductor loss and easier matching.


Where your simpfeed method really makes sense is with beams
where the elements are more than 1/2WL long. For instance,
attached is the 33 ft. long, two-driven-elements spaced at
ten feet used on 17m. The impedances are high enough to use
300 ohm twinlead for the phasing and the feedpoint impedance
at the phasing harness is around 25 ohms.

Roy, what is the secret to balancing those currents in the
two elements? With the same Z0 phasing lines, the SWR circles
don't intersect so the impedances looking into those phasing
lines is never equal. Does my question make sense?
--
73, Cecil, W5DXP