Hand winding smps transformer

Question

I would like to hand wind a smps transformer for prototyping work. The specs of the transformer are as follows :

Transformer Parameters

Core Material: EE16, NC-2H or equivalent, gapped for ALG of 88.55 nH/t² Bobbin EE16, Horizontal, 10 pins, (5/5)

Winding Details

Shield: 15T x 3, 35 AWG

Primary: 105T, 35 AWG

1st Half Bias: 6T x 4, 30 AWG

2nd Half /Feedback: 6T x 4, 30 AWG

Secondary: 7T, 22 TIW

Winding Order

Shield (2–NC), Primary (4–1), Bias (5–3), Feedback (3–2), 5 V (10–8)

Primary Inductance 1.074 mH, ±10%

Primary Resonant Frequency 1000 kHz (minimum)

Leakage Inductance 95 μH (maximum)

I have procured EE16 core, bobin and the copper wires. Can anyone explain me how to actually go about winding this one? specifically:

1. Which direction to start each winding CW / CCW ?
2. What to do in cases where turns are very less to fill one layer?
3. I have seen another transformer design where there were parallel coils connected to same terminals, In this case do we turn both coils side by side or on different layers?

Answer 1

1) Winding direction doesn't matter as long as you keep it consistent. If you are adapting a lathe or drill as a winder, the winding direction will be determined by the machine.

2) If the small windings are at approximately the same potential they can share a layer, otherwise you must provide sufficient insulation between them. Sometimes you may have to separate layers to reduce capacitance between them, or interleave them to reduce leakage inductance, but that is too complex to discuss here.

3) Winding coils side by side is simpler and provides better voltage isolation especially when the sections are different bins on the bobbin. That is essential for high voltage supplies and makes mains voltage safety approval easier.

However it increases the leakage inductance, and that matters for high frequency applications like SMPS. The only reasonable way to know if you can do this is to wind one that way and measure the leakage inductance. If it's out of spec (or even close to it) wind another in layers and compare them.