Hello everyone, 

🤓 I hope you’ve all had a wonderful weekend!

Ready for some Power Electronics insights?

💻 Part of my job at Frenetic involves building demonstration Transformers and Inductors for potential clients, using our online platform. Therefore, being up to date on the latest updates, potential issues and new features is fundamental for me. This week I want to explore an important improvement in the “Suggest Wire” feature in Frenetic Online, released a few weeks ago. But before sharing all the secrets about the new functionality and its parameters, let me focus on a few key points.

⚠️ The bottleneck

My first question is: “How much time does a designer spend on each step of the Magnetics design?”. In our platform, the main steps of the process would be:

1. Waveform tab
2. Core Optimizer/Core tabs
3. Windings

Discussing with experienced users on the tricky parts of design, and adding my personal experience with Magnetic design, I’d argue that 30-60% of the time is allocated to the Windings tab. Here, you have a lot of parameters to think about, such as Wire type, Winding losses, Leakage effects, Materials in stock, Manufacturability of the Windings setup and Isolation concerns.

⚡The solution

The simplest thing a designer can opt for after selecting the Core, Turns and Core material, is to move to the Windings tab and hit the “Suggest Wire” option at the bottom:

Figure 1. Suggest Wire Option

🤔 You might wonder now, what’s the big news?

Well, before our latest update, the user had no control over the suggestion. But, good news: that has changed! The user can now control multiple things:

Figure 2.  Suggest Wire chart flow

🤖 Functionality and the reasoning behind it

The selection of the Windings is tricky and requires experience from the designer. We have tried to put in the “Suggest Wire” option as much expertise as possible in order to cover multiple scenarios. Starting with the type of wire, which is not so obvious, respecting the insulation provided from the user, trying to get current densities to safe values, accounting for manufacturability aspects, such as filling factors, parallel windings, and on top of that, figuring out the best approach as far as winding losses go.

👀 There are more things that are considered on the background, but I think the best way to judge is to try it!

🔎 From words to action

Let’s revisit an older project of mine, in which I remember spending quite some time on the Windings section, trying to get the best performance. In total, I simulated 3 versions of that 1.25kW Transformer, playing with the Windings each time.

📃 Here the basic PSFB specs that I had:

Figure 3. Example 1.25kW transformer specs  

Figure 4. Versions and simulation results  

Here is the summary:

🙅‍♂️ The first try was V0 and, if you notice, the turns number is high. Besides that, building V0 is nearly impossible and certainly not good for mass production. Why is that?

First of all, there is no space for the primary windings (in blue) to escape the bobbin and get connected to the pins. And then, 0.04mm litz wire strand thickness is not the cheap option.

🔥 After V0, I reviewed that design and tried a lot more carefully to manage spacing, margins and total losses. By the way, I changed the turns, keeping the same turns ratio, and went with cheaper options as far a litz wires go. 0.05 and 0.07mm for the litz wires is a good design choice to balance cost-efficiency.

🧑‍🏭 Used a bifilar winding for the primary and a x3 parallel windings for the secondary. Again, manufacturability-wise I don’t like it much... There is just not enough room for the wire to come out on the sides.

By now I have spent a couple hours going back and forth with my design choices and still don’t have something that hasn’t got a major issue.

✨ The “Suggest Wire” option result

Finally, a design with enough room for the primary and secondary wires! 0.1mm litz wire strand diameter for both windings is the cheapest option that makes sense in that frequency range/this type of design. Losses are higher, but nothing quite dramatic. And the best part is that the “Suggest Wire” option took literally 33 seconds to produce this design.  

☝️ Can we do even better based on the suggested design?

My idea would be to keep the same design as the suggestion indicated, and maybe try to use 0.07mm thick strands, keeping the same litz OD, to bring the losses a bit down and keep all the benefits of that design, without harming our budget too much.

✏️ What I got using 185x0.07mm for the primary and 250x0.07mm for the secondary keeping the geometry and dimension nearly identical, was 4.87W of total losses and 92°C hotspot temperature. That is a very good design as well as the suggestion.

✅ Learnings

At the beginning of this project, I struggled to fit windings, used more parallel wires than I intended to, and ended up with V1 design that looks good on paper, but not so much in reality. After our updated feature was released, I revised that design and made a new version in only 33 seconds!

The result is a much better design that is manufacturable without issues. Then I saw an opportunity to make this design a bit better keeping the same geometry but adapting the strand diameter to reduce those skin/proximity effects some more.

🚀 In the end, I got a design that outperforms my previous attempts! The point is not that I modified the suggestion, but the fact that in less than 1min I was in a comfortable place tweaking the suggested design to make it even better if possible. 

🎬 If you're curious to discover more on the updated feature, you can check out the instructional video made by my colleague Lucas here!

😎 I hope you've enjoyed the read! Stay tuned for the next one. 

-------------------------------------------

🏆 Explore the latest trend in Frenetic's Magnetic Notes: "Magnetic Design for a Dual Active Bridge"

🔎 Dive into our comprehensive articles covering critical DAB aspects, including Bidirectional Operation, Optimal Transformer Design, and Hardware Validation.

🗞️ Read the full article here.

Sotiris Zorbas, MSc 

Power Εlectronics Εngineer