The final transformer is now made of 0.4mm enammeled wire.
The secondary is wound directly on the motor. The motor MUST be made of plastic, at least for the rotor. If transformer is wound on conductive part, it wont work.
The secondary is made of 4 layers of 20-19-20-19 turns (78 total).
The gap between primary and secondary must smaller as possible, but must have enough headroom to avoid touching between two parts. I used a thin cardboard, approximately 0.5mm thick.
The primary is then wound on this cardboard. It is splitted in two halves accordingly to push-pull drive stage. Boths halves are made of 0.4mm wire, two layers of 23-22 turns for each one (90 turns total).
The transformer diameter is around 32mm for secondary and 34mm for the primary.
This gives an inductance of around 102µH for each primary.
With this inductance, powered with 9Vdc, a switching frequency of 50kHz is correct and gives a maximum current of di=U*dt/L
di = 9*10µ/102µ = 900mA peak. (each primary is powered for 1/2 period = 1/(2*50k))
Primary switching frequency will define how much power is available on secondary. Decreasing frequency will increase peak current, and so on for output power. On my design it works great, but I guess it wouldn't be able to power all leds at full power for a full revolution of propeller. However, on my display, the leds are never all lit at the same time and big capacitors help to provide energy during this periods.'