pancake winding is a coil geometry in which conductor is wound as a flat spiral lying in a single plane, and the finished magnet is assembled by stacking many such discs and connecting them in series.
The name describes the shape. Its alternative is layer winding, in which conductor is wound helically along the coil axis, layer over layer. Pancakes suit tape: a flat ribbon bends readily about its thin dimension but resists bending about its width, so a flat spiral is the only comfortable geometry for 📝HTS tape. The construction also suits manufacturing at the scale 📝fusion magnets demand, because each pancake can be wound, inspected, and electrically tested on its own, and a bad one discarded before it is built into a stack that would be far more expensive to unmake.
The distinction between the two variants is electrical rather than aesthetic. A single pancake spirals from its inner turn to its outer turn, which strands one terminal at the small radius where it is awkward to bring a lead out. A double pancake is wound from the midpoint of a conductor length outward in both directions, producing two spirals back to back joined at the inner turn, so both terminals emerge at the outer radius. That convenience is why double pancakes dominate solenoid practice and why 📝ITER winds its 📝toroidal field coil conductor in double pancakes. Magnets that solder tape into machined radial plates can accept single pancakes instead, since plate-to-plate joints are engineered regardless: the 📝Toroidal Field Model Coil (TFMC) winding pack was a stack of sixteen single pancakes, 256 turns in total.
Pancake winding is a geometry, not an insulation scheme, and the two choices are independent. A 📝no-insulation coil can be pancake-wound or not; the TFMC was both, stacking no-insulation pancakes into a magnet that reached 20 📝tesla in 2021.