45# 🚀 2 kW PSFB Transformer designed with Ansys - How accurate is Ansys?
SuperPowers for Electrical Engineers 🦸🏻♂️
👋 Hello Supergineers! Dr. Molina here! 👨🔧
Welcome to Dr.Molina Newsletter, where I break down Magnetic designs for you every week, in 4 minutes.
Intro
Last week, I talked about the Full Bridge Phase Shift design from Infineon.
Today, I tried to replicate it using Ansys Software and looked for some conclusions (read them at the end).
I used Pexprt and Maxwell. For those who are not familiar with this software. Ansys is one of the biggest companies of physics software in the world (>$2b of revenues). They have software for designing and simulating magnetics, released in 2000 (more or less) and Maxwell 3D is the FEM, which makes the simulations for the Rac and the Llk with more accuracy. For 20 years, this software has been the best in the market.
Let’s go
I start with Pexprt. The waveforms feature of Pexprt doesn´t allow too many options. I just tried to put the closest waveforms to the originals.
The type of waveforms accepted by Pexprt is one of the biggest limitations of this solution.
I ran the simulation without any filter and these are the outputs.
PM87 is the best option. PM87 is huge.
If you check the PM87 winding suggestion, this is what you find out. Since the core is big, the window is almost empty.
To make a fair comparison, I ran the simulation filtering by the PQ40, which is the same proposed by Infineon and Frenetic. The proposed design filtering by PQ40 makes sense. Litz wire in primary and Foil in secondary.
I checked at Frenetic and there isn´t enough window. Therefore, from a manufacturing point of view, the design is not valid. Additionally, the idea of having several turns in the same layer with foil is not very convenient at this core size.
The losses are incredibly low, 2,7W. God should be here, because this is a miracle.
I’m going to run a 3D FEM simulation in Maxwell 3D to verify the losses of this first iteration.
Well, the first iteration was 2,7 W and after 40 minutes o 3D Maxwell simulation, 52 W and 373 ºC. This is the breakdown of the losses in Pexprt.
Conclusions
❌ Pexprt is limited by the inputs (waveforms, cooling, and mechanical conditions) and the materials libraries.
❌The first design suggestions are poor, you need to filter by core.
✅ The desktop software shows a lot of information about the losses.
✅ It has been (for me) the best software for magnetic design for 20 years.
❌ Applying custom winding configurations is not that easy.
❌ The accuracy of this simulator is around 40%, therefore, a Lab test is needed from the very beginning.
Do you want to work with the best Magnetic Designers in the World?
✅ Join our Training for High-Frequency Magnetics.
🚀🚀 Book now and get a 40% of discount and pay later.
✅ Boost your Professional career in Power Electronics