Well don't really know what highly sub-optimal windings is - and lily
"lily painting" there is a homily about "painting the lily and gilding fine gold",
in essence talking about wasting time optimizing things that do not need optimization
and where the efforts perhaps even mar the result. It is another instantiation of
the "big stones first" principle, a.k.a., go after the tallest tent poles first (and
a gazillion other variants). So the suggestion is to determine the key factors
and sort them by magnitude. (Sometimes also discussed as finding the bottlenecks)
But the key element is sorting them by just how much they limit the desired behavior.
Then don't waste any time on any except the first one or two, and only when those are
addressed as fully as possible do you move on to others -- typically by redoing the
list based on the modified design.
Cross-coupling between two solenoids, whether end to end or side by side is fairly
poor -- thus such windings are "sub optimal", i.e., less than optimal, for power
transfer. Taking something so sub-optimal and optimizing it (with resonance) rather
than fixing the problem (poor coupling) was what I was referring to, I just used phrasing
that was an attempt to be jocular and not sound directly critical. I'm sorry if it
became obtuse as a result. But since that is the most limiting factor in the first
design idea, that is the one to address.
Going to a cored transformer with two half-cores separated by an "air-gap" (this
being the plastic, and since the mu of the plastic is expected to be mu-nought,
just as would be for air, hence use of standard "air-gap" terminology) will give
extreme improvements in primary to secondary coupling, which probably obviates
any need for resonance (at least in an attempt to improve power transfer, resonance
may still be useful for other reasons). So, use a proper geometry and a core (if
possible) and make the winding goemetry much closer to optimal as the first step,
that was the recommendation.
As for calculating -- there are times it is indispensible. You don't want to
"cut and try" when peoples' lives are at risk or when it costs multi-mega$$ just
to try it. But for small projects, the expense of time and effort to try something
and then adjust it a few times is often far lower than the time required for
calculations of sufficient accuracy to be useful. More powerful computers and
pre-written software start shifting the cross-over point, but SPICE is pretty
awful at things where parasitics are important, and you'd need a field solver
for this problem also, so "cut and try" looks better than the computational
alternatives. MHOO
I hope this is clearer now, please feel free to ask away on anything that
isn't or new issues raised. And the "lily-painting and gilding fine gold"
thing sometimes even gets referred to as "lily-gliding" -- so I would
recommend you avoid gilding any lilies ;-)