A third QCW from the same builder, unusual at both ends. The primary is wound on five stacked 36 mm ferrite POT cores (D50×H60 mm, 13 turns of 14 AWG) and sits inside the secondary. The secondary is D95×H90 mm, 520 turns of 0.12 mm wire under a 135×35 mm toroid. His measured figures: L_sec 21.54 mH, L_pri 17.26 µH, k = 0.5, primary resonance 313 kHz against a secondary at 356 kHz on the floor and 411 kHz on a table, upper pole 480-490 kHz. The MMC is fifteen WIMA FKP1 0.022 µF 6 kV in 3P5S, 12.88 nF, hand-labelled as such on the pack. Bridge: IXGH48N60C. The driver is his own rework of ZakW’s UD Custom. ZakW confirmed it in the thread, saying it was cool to see his UD Custom underneath.
The fibre setup is the part other builders could copy. A QCW normally has two links, one for the ramp and one for the interrupt. He thought that wasteful and derives the second from the first. A Schmitt-inverter circuit takes the incoming ramp PWM and pulls the interrupt envelope off it through 470 Ω, a 1N4148 and a 500 k / 1 nF time constant. Two more diode-and-RC networks make the buck high-side and low-side drives. He published the sheet with its values and the tuning note that went with it (at his 94 kHz PWM frequency it wants 470 Ω and 200 kΩ). He also posted the simulation, showing the interrupt gate, both buck gate signals, and the bus climbing to roughly 280 V over nine milliseconds.
He compared ramp shapes on one coil: straight, curved and a raised cosine. That comparison is more often argued about than measured. He also found that his flashovers were not turn-to-turn on the secondary. Taking it apart showed no track. They were dielectric barrier discharges between primary and secondary, and potting the primary in epoxy removed them completely. Primary current peaks around 150 A at a ramp of 15-18 ms.
Schematics and photos
schematic: deriving the interrupt from the ramp PWM, plus both buck gate drives 
simulation of that circuit: interrupt gate, both buck gates, bus to ~280 V 
ramp shape: straight 
ramp shape: curved 
ramp shape: raised cosine
Sources
This is the card from the driver tree, written from what the builder published. Send corrections through the forum.