1

i am designing a transformer for 2-switch forward converter. For now, primary is wound as 5 wires of 0,5mm diameter (34 turns, 3 layers), the secondary is wound as 3 wires of 1,32mm diameter (17 turns, 2 layers). The frequency of smps is 83khz. The idea is to use litz wire in primary and secondary and increase frequency to 100khz. So, the primary would be wound as litz wire (0,1mm diameter, 120 strands) in 2 layers. The secondary would be litz too (0,071mm diameter, 1076 strands) in 2 layers too. So, the skin effect must be elliminated completely. But what about a coupling? I have read a lot of books on winding interleaving technique, but all the cases if wire radius is more or less greater than the skin depth. So, it is easy to do interleaving on this litz wire transformer (1 layer primary, 1 layer secondary, 2 layer primary, 2 layer secondary). So guys, will it worth it? I hope there must be some benefits like increased coupling between windings and reduced proximity losses... But i am not sure about that. So - is it good idea to use winding interleaving on litz wire windings? Will i get better coupling between windings and decreased leakage inductance? Thanks for your response. It is easy to do interleaving, but harder than just wound windings as 1P,2P,1S,2S. I will post results of my experiments most likely.

silkyre6xtenz
  • 83
  • 8
  • Also consider a bobbin that allows a bundle winding gap e.g 6 turns gap 6 turns gap... repeat to raise SRF as required. – Tony Stewart EE75 Mar 05 '18 at 05:49
  • What exactly is your design target here? Do you want to minimize cost or size or temperature or losses? Regardless of these points, interleaving has always the benefit of lower leakage inductance bought at the expense of higher coupling capacitance (main source of common mode EMI) and transformer complexity and hence price. – christoph Oct 13 '20 at 18:19

2 Answers2

2

This is a bit late, but...

For this frequency range proximity effect losses will be much higher than skin effect losses. Proximity effect loss is reduced by using Litz (also good for reducing skin effect losses) or bunched conductors (less effective at reducing skin effect losses). Loss resistance due to proximity effect is proportional to \$ number\_of\_wire\_strands \times wire\_diameter^4 \$ which is why bunched (or Litz) wire is so effective at reducing AC copper losses. An increase in wire strand diameter will cause proximity effect losses to increase - a seemingly counterintuitive outcome.

Since you're operating at a higher frequency, consider reducing the number of turns by a factor of 0.83 (83kHz/100kHz). You'll maintain the same flux density which allows you to reduce DC & AC copper losses which is proportional to the number of turns.
Reducing the number of turns also reduces leakage inductance \$ (L_l\propto N^2) \$.

To improve coupling (reducing leakage inductance) interleaved windings are your friend. It does increase the complexity of the construction which is costly. The number of interfaces between the primary and secondary winding layers is the important factor and will reduce the leakage inductance by a factor of \$ 1/M^2 \$, where \$M\$= number of interfaces. The standard physical arrangement for interleaved windings:

  • M=1: P : S
  • M=2: S/2 : P : S/2
  • M=4: P/4 : S/2 : P/2 : S/2 : P/4
  • M=6: S/6 : P/3 : S/3 : P/3 : S/3 : P/3 : S/6

Where P & S are primary and secondary number of turns (P & S are interchangeable in the above formulas, e.g., you can do P/2 : S : P/2 for M=2).

For a narrow band transformer there is an optimal wire diameter, in metres, to minimize AC + DC copper losses. $$ d_{opt} = {\sqrt[6]{128 \left({ \hat I \; \rho_c } \over {\pi \omega s \bar {\hat B} } \right)^2}} $$ Where
\$ \;\; \rho_c \$ = resistivity of the wire in \$\Omega \; m\$ (approx \$1.72 \times 10^{-8}\$ for annealed copper wire)
\$ \;\; \omega = 2\pi f\$ = angular frequency
\$ \;\; s = \$ number of wire strands
\$ \;\; \bar {\hat B} = \$ average value of peak flux density [T] over the space occupied by the winding \$ = {{\mu_o N \hat I} \over {l_e}} \$
\$ \;\; \mu_o = \$ magnetic constant \$ = 4 \pi 10^{-7} \$
\$ \;\; \hat I = \$ peak current which cancels out
\$ \;\; l_e = \$ effective magnetic path which is approx equal to the winding breadth for an ungapped core in metres

Good references for transformer design:
"Soft Ferrites", E. C. Snelling
"Ferrites for Inductors and Transformers", Snelling & Giles
These books can be found in uni libraries.

qrk
  • 7,474
  • 1
  • 5
  • 20
0

It is better to do interleave winding method to reduce the number of consecutive layers with rising MMF proximity losses towards centre of core.

Since your turns ratio is 2:1 this is what your interleave pattern should be 2P1S... repeating.

If Eddy current proximity losses can be minimized, the benefit is lower core temp rise at rated current.

There are many variables with core types , geometry and winding methods and skin depth becomes more important as the number of layers increases.

Tony Stewart EE75
  • 1
  • 3
  • 54
  • 182
  • sounds good, but the transformer has different current ratio: the primary conducts 3A of current while secondary conducts 10A. So, it is impossible to do interleaving as you said. – silkyre6xtenz Mar 05 '18 at 07:28
  • Then add insulation between all layers to reduce interwinding proximity flux – Tony Stewart EE75 Mar 05 '18 at 12:53