#62 🚀 How to design a 600W LLC series Inductor (HF AC Inductor)
SuperPowers for Electrical Engineers 🦸🏻♂️🦸🏼♀️
👋 Hello, friends! Dr. Molina here 👨🔧
Welcome to Dr.Molina Newsletter, where I break down Magnetic design for you every week, in
4 minutesthe time needed.
With this edition, I’m already completing a year since the first edition!!
This is crazy!!
Some simple numbers after one year.
62 editions
More than 1,000,000 views of my articles
I’m happy with the results. Let’s continue!
How to design an AC inductor for a 600W LLC converter
To continue with specific examples of my Inductor’s table, today I will focus on High-Frequency AC inductors. As we can see, the inductor could be designed with Ferrite or Powder. I’m following the specifications of this app note of Infineon, where they use ferrite (3C95); therefore, I will use the same material.
The electrical specifications of the power converter and the information we extract for designing our inductor are:
In the following graph, we can compare the hysteresis loop of the current in an LLC resonant inductor with the one in a DC inductor in the previous article.
INDUCTOR DESIGN USING THE ENERGY METHOD
The inductor energy is calculated as:
In this case, if we have 14 uH case, with a peak current of 5,8 A:
As we did in this previous article, using this energy as the minimum energy to be stored in our inductor, the peak current should be below the saturation current of our design.
The main difference between this design and a DC inductor design is the magnetization curve, as shown in the picture.
With the Emin of (1), I have taken two RM cores with two different gaps to be analyzed.
In this case, the maximum energy is calculated assuming a maximum Bpk=100 mT. This value selection is the main difference between a pure AC inductor and a DC inductor. In the previous article, we used 360 mT for a DC inductor at a similar frequency. This Bpk selection also depends on the frequency; for example, at 1 MHz, we will reduce to 20 mT.
The only combination that provides the required energy is RM12 with 0,5 mm of GAP.
NUMBER OF TURNS
With the core and gap selected and the desired inductance, we can calculate the number of turns needed. The value of the gap could vary to have a specific value of inductance.
With the number of turns and the adjusted gap, we can calculate the core losses.
WINDING CALCULATION
For high-frequency inductors, the only wire technology that makes sense is the Litz wire. The ideal winding would be one single layer (to reduce proximity effects between layers) and one parallel. However, for high-frequency inductors, where there is no DC Bias, the skin effect and the proximity between each conductor dominate.
A simple process to see if this is possible to get a single layer. Take the width of the coil former, subtract the margin tape, and divide the rest by the total number of turns.
The total diameter for having a single layer is 1,8 mm, including isolation. That
For the Litz wire, considering the skin effect, for a 150 kHz frequency, a 0,1mm strand diameter is fine. You can decrease the strand, but the trade-off between price and losses will be unbalanced.
For a 1,8 maximum diameter and 0,1mm of the strand, assuming the isolation tape, the resultant number of strands will be 120. This value represents a current density of 4,5A/mm2.
LOSSES BREAKDOWN
The resultant losses breakdown is as follows.
If you need this design, let me know it.
CATALOG DESIGNS
For high-frequency inductors, I haven´t found a catalog of the leading manufacturers. At Frenetic, we have created some solutions, and I plan to expand these designs to cover the whole spectrum. In this case, I will not suggest any catalog design today.
Frenetic Core Optimization
During this explanation, I have shared the procedure using the equations and the public datasheets of cores. However, if you are one of our professional users of Frenetic, you can do a deeper analysis using our Core Optimizer. The Infineon design used RM12, which is fine but running the Core Optimizer, you can see other cores shapes and combinations of gaps with better performance, lower size, and cost.
In the picture below, I have plotted PQ and RM shapes filtering for 100 mT maximum magnetic field.
For example, the PQ26/14 or PQ32 could be a smaller solution.
In the following picture, we can analyze the best combination of turns and gaps for the maximum magnetic field of 100 mT.
This analysis can be extended for critical conditions and seen in the same graph losses for the same design under different input conditions. For example, the LLC works under different switching conditions. In the following table, I can compare the losses for the same core under the minimal, nominal, and maximum switching frequencies. The three colors represent each case:
Pink - Minimum frequency 100kHz
Dark Blue - Nominal - 155 kHz
Green - Maximum Frequency - 250 kHz
We can observe that once we increase the number of turns, the differences decrease. The maximum frequency represents the worst case of core losses.
Sometimes, I don´t remember why I made a decision. With the previous graphs obtained in the simulations of Frenetic Online, you are tracking your design options. In two years, you will check the design, and the results will be there for you 🤗.
NEXT WEBINAR
In the last Webinar, we couldn´t answer all the questions. Therefore, the next webinar, on the next 20th of December, will be a Ask us Something Webinar. You can send us questions before!
As always, you need to register, and here is the link.