I would like to build a flyback-converter that is presented in a research paper. The following information about the transformer have been provided:

Switching frequency: 50 kHz
Number of primary turns: 9
Turn ratio: 10
Magnetizing inductance: 14uH
Leakage inductance: 0,4uH
Core: ETD 44
Input voltage: 35 V

The converter is rated for 250 W. There are 2 outputs with the center connected, so I would need 9:90:90 windings if I understand it correctly.

The transformer will be operating in DCM/BCM hybrid mode. The output will be a half-wave of a 50 Hz sine, in the low voltage regions it will operate in DCM.

From some oscilloscope measurements in the paper I can see, that the peak current through the primary winding is at about 40 A. The conduction loss of teh transformer has been calculated as 0.7 W, and Irms and Rl have been measured to calculate it (but aren't stated in the paper).

Unfortunately I don't have experience with winding transformers, so I don't know what core I need, if the core needs to have an air gap, what diameter wire I need and so on. How can I determine that, so the transformer I will build matches or at least comes super close to the values specified above?

Also is ETD 44 the "best" core for this or are there other, more efficient cores which could be used here?

  • \$\begingroup\$ ETD44 is just the core size. What's more important is the core material (some kind of ferrite, but which one?). Given the material and magnetizing inductance, you can calculate the required air gap. Do you have that information? If so, you should add it to your question. \$\endgroup\$ Commented Jul 8, 2023 at 20:43
  • \$\begingroup\$ @JonathanS. Unfortunately it is just described as a "ETD 44 ferrite core". The only additional information is the core loss, which was calculated as 1.85W. I believe it is possible from the core loss and size to figure out the material? \$\endgroup\$
    – Jenny
    Commented Jul 8, 2023 at 21:03
  • \$\begingroup\$ 250 W with DCM/BCD sounds like it would need a very large core. ETD44 is very large. Have you considered other options? \$\endgroup\$
    – winny
    Commented Jul 8, 2023 at 21:03
  • 1
    \$\begingroup\$ For 50 kHz, I would start with N87 material. Try to wind it yourself first. \$\endgroup\$
    – winny
    Commented Jul 8, 2023 at 21:04
  • \$\begingroup\$ @winny I want a high efficiency design and this thing is working in zero current switching and zero voltage switching and is the most efficient design I found. I did consider other options, especially since I want to build it as an interleaved design so the total peak output is 500 W, but this was the best design I found from an efficiency standpoint. \$\endgroup\$
    – Jenny
    Commented Jul 8, 2023 at 21:05

1 Answer 1


14uH for 9 turns means an Al of about 173.

On an N87 ETD 44 core that's about 1.5mm gap (one half gapped, one half not). If you can't procure a gapped core you can get similar results with two ungapped cores and a 0.75mm non-metallic spacer between the two halves.

SMPS transformers often use techniques such as interleaving (sandwiching the secondary between halves of the primary). Maybe if you have a DOI for the paper there might be clues in the publication that you could have missed.

Left as an exercise to scale dimensions from the photo, but you can see the air gap, so it appears they used ungapped core halves.

enter image description here

  • \$\begingroup\$ Sure, the DOI is doi.org/10.1049/iet-rpg.2019.0307 I also tried contacting the author for additional information but unfortunately never got a response. \$\endgroup\$
    – Jenny
    Commented Jul 8, 2023 at 21:58
  • \$\begingroup\$ I saw that aswell with the ungapped cores after you mentioned it :) They seem to use litz wire apparently? And have added some insulating tape, I assume I can leave that away for better thermal performance? I still need to figure out wire diameter (or is that just "as big as possible", limited by the core size?) and the core material and how I should wind it (primary in between the 2 secondary windings sounds like a good idea to me?) \$\endgroup\$
    – Jenny
    Commented Jul 8, 2023 at 22:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.