Hello, Sotiris here! 👋

🚀 Today’s newsletter is about creating a new PSFB transformer in a different way than before. This time I’ll have a specific design goal in mind. That is cost and ease of manufacturing. 🚀 

I’ve also asked my colleague Lucas for some info on the manufacturing process since he has an extensive hands-on experience in this area! You’ll see his tips during the design process described below.

This time I wanted to spice things up, by keeping track of time ⏱️ during all creation stages.

Let’s start the design!

⏱️ 3:22pm

👉   I created a copy of my original ETD54 project and called it PSFB_1kW_E55. That was designed in newsletters “Frenetic#8”- “Frenetic#11”, located here.

👉   Then I waited at the “List of Designs” tab, where the tool makes complete transformer suggestions fully specced. 2 minutes have elapsed for the tool to make its suggestions.

⭐️   Talking with Lucas, he pointed out that the E core is the cheapest and easiest to find, in general. Of course, the type of ferrite material plays a role as well. A 3C95 material is much more common and cheap than a high frequency 3F46 material. Having already built PQ and ETD samples I went with the E core suggestion.

✅   The goal of low cost is met!

I just loaded the suggested project and started making changes to the core and windings tabs.

Core selection:

👉 I changed the suggested core to a E55/28/21 that was available in the lab.

👉 The selected core is bigger thus the bobbin area is going to be larger, allowing for more freedom in the winding’s selection, as one can easily imagine. For the same winding construction, a larger bobbin area will decrease the filling factor. As discussed, many times a low filling factor of 60-80% is the sweet spot for an easy winding process.

👉 The magnetizing inductance with a 50um center leg gap is around 3200uH which is fine for the test. We have seen by now that when designing a “zero” gap core, in reality there is a gap because of the surface finish of the core set. The larger the gap, the lower the AL value and the inductance of a winding is simply:

L = N² A_L 

Where:

• L is the inductance of the winding in µH
• N is the number of turns
• A_L is the µH/N²

Measuring many ungapped samples primary inductance, using our tool one can easily find the gap needed to achieve this result. The gap ranges between 0.02-0.05mm. Thus predicting a 0.05mm as a worst case scenario(I care for the lowest primary inductance in A PSFB -> biggest gap in the range) is good enough for the purposes of this test.

👉 Note the BpkT (peak flux density) marked in the photo, is at a safe 52mT. Looking at the N27 material datasheet I observed that the saturation limit at 100°C (the worst case) is rated at 410mT. We are far for this number, so no pushing our lack close to the saturation limit. Operating near the saturation limit brings about high core losses and a high possibility of instabilities and decreased reliability in a design. It’s a nice topic to explore in more detail in a subsequent newsletter 🧐.

👉 Core losses are at 1.33W, noted. No comments yet, we will take a look at the total losses of the transformer, when the windings are finalized in the next stage.

Time? ⏱️ 3:27pm

Windings selection:

I’ve made some changes in the windings based on what Litz wires ware readily available in the lab. I also strived for a 5A/mm² rule of thumb I go with every time. That rule isn’t written in stone but is the suggestion that many manufacturers and books recommend. I’ve seen good results with it, so I follow it.

⭐️ Lucas pointed out to me that the PSP topology produces similar losses with the PSPS in most cases and the big advantage is the simpler winding process. I’ve checked Luca’s tip and it does check out! By the way, the pinout problems and issues I talked about in Frenetic#13, are less troublesome in this configuration. The lower layer count results in a simpler construction.

Noticing:

• Filling factor at a nice 68%. Below 85% is fine for a fast and easy winding process.

✅ The ease of manufacturing goal is met.

• Total losses at 5.56W. At a 1000W power rating that’s a transformer efficiency of 99.44%. I like it as well, it’s lower than my other PSFB attempts, so no complaints.

Ok I think I’m finished! Looking at the clock, I got: ⏱️ 3:45pm

🚀 🚀  That’s 23 minutes of design time. I doublechecked I’m not wrong…! 🚀 🚀 

I want to stress at this point that this fast iteration wouldn’t be possible without the prior experience designing transformers specifically for this topology.

⚡   The point is that the software is really allowing you to make quick designs.

⚡   Success is not guaranteed of course, but the ability to make a couple of iterations and order samples to test in the lab goes a long way towards have a successful design in the shortest amount of time

Thinking about how much time and effort a single iteration took me before I had access to Frenetic online tool, feels like going back to the dark ages 😂

If you are curious to know more about Frenetic Online, just click here to book a demo.👈

What's coming next week?

👉Next week I’ll make a couple more design iterations for 2 more PSFB transformers with different design goals in mind!

👉As always, if you have a question or an idea, let me know!

 👋  Until next week 👋 

Sotiris Zorbas, MSc 

Power Εlectronics Εngineer