Divide breakdown voltage of the mag wire by your volts per turn, and
come up with a maximum turns per layer number. I would actually cut that
number back by at least 15%.
We used to wind 800 turn secondaries, and the max turns per layer we
would use was around 60.
Another important thing in multi-layer secondary winding is to refrain
from "scatter winding", which means do not wind a layer chaotically. One
should wind evenly across the face of the bobbin (for more than a few
reasons), or previous layer, then place the transformer tape over that
layer. It should appear flat. Do NOT stretch the tape. Also, keep the
start of the layer turns and the finish INBOARD on the winding face at
least half a mm. This keeps the start or finish turn of one layer
becoming proximal to an underlying layer or one above it.
Yes. There is no reason to reverse turns direction on any winding,
unless you want to degrade performance, and that should never be a goal.
Correct. The only benefit I see from doing it as Adam
was taught would be that the maximum voltage difference
between the winding layers would be the voltage developed
on a single layer, rather than twice that, like the
conventional winding scheme.
On a scatter wind where all turns are on the same layer, the final turn
is many volts away from the first, so a breach can occur between them.
This is the whole reason a good xfmr shop winds high turns count
secondaries (or primaries) on multiple layers, each of which are
segregated by insulator media.
This may be an issue for high-
voltage coils - I've made coils for up to 25kV, but have
found it more practical to add a layer of Kapton tape,
rather than suffer the lump in the winding for the return
wire.
There is no lump if the method used is right. The tape is needed to
keep turns away from each other. One winds across and back until the max
turns per layer of the designer are met, and then a turn of tape is added
AS the wire continues turning about the bobbin. Then the next layer's
turns get stacked right on the underlying layer, and there was no
cross-over or bump of any kind. Turns can start at the "bottom" or "top"
of a bobbin segment. It doesn't matter. The only thing that does is the
direction of the turns.
Also, a simple halving of the maximum layer-layer
voltage is often insufficient, and tape is preferred.
Tape doesn't seal off layer edges and should not be relied upon to do
so. It only separates layer faces. Keep turns inboard a bit to increase
creepage distances, and always vacuum impregnate an HV design in varnish
or at the very least, potting compound. That is, of course, unless it is
an oil bath design.
In
fact, sometimes even tape can't easily handle the voltages
encountered, and I've found a banked winding approach to
be effective.
If it is too high, a segmented bobbin is required. I posted a picture in
a.b.s.e a couple weeks back that one troll jumped all over, but it does
perfectly show a proper EHV xfmr (miniature low power) that has about 5
bobbin segments, each with nearly 300 turns, which is pushing the limits
of the mag wire. I'll post it again.
I see from Terman's 1943 Radio Engineer's
Handbook that this method was known from way back when.
A friend makes coils to 50kV and he uses a type of banked
winding approach, with each coil portion wound in its own
section of a custom-machined bobbin.
Yes. We had many customs made, and another easy way to do it is to
have G10 or FR4 bobbin parts made and epoxy them up yourself.
Start with off the shelf G10/FR4 tube, and get discs made for the bobbin
segment separators out of non-clad flat sheet stock It can be milled
just like a PCB design, and popped out from little perfs.
We even took an off the shelf bobbin, cut off one face (end), and added
discs to it, and then replaced the end we cut off. It really does work
well. I actually have some pics on one of these old hard drives here
somewhere.