🚀 Welcome back to the Frenetic Newsletter after a short summer break! Today we analyze why Specs Reviewing is key to your Magnetics design.
View in browser

Hello everyone, Sotiris here! 👋

 

🏝️ We’re finally back with the first Frenetic Newsletter after summer vacations! I must confess that I had a really hard time figuring out which topic I should have addressed: lots of ideas, but very little time to execute them.

👀 By the way, if you missed the last Newsletter we released before the break, I suggest you give it a read! It’s one of my favorites on the PCB Layout and Measurements.

 

☝️ Iterations can be avoided

Today I’ll revisit an LLC design that I simulated in an old Newsletter of mine, where we analyzed a case study of various Power Transformer designs to prove that the 2-chamber approach can be an option up to a power level of 2kW.

 

📈 The 2 chambers allow us to control the leakage inductance precisely, which in series with the resonant capacitor oscillates near the resonant frequency. We are controlling the leakage inductance by increasing/decreasing the separation between layers.

These were the simulator output parameters.

    NL 66 - 1

    Figure 1. Transformer specs – 2D cross section

     

    🤨 You might wonder, why are we revisiting this design?

    Well, I want to review the specs of the design with you and show you the process. In my opinion, the specs of each design are the most important part of it, and that’s why I will dedicate this entire Newsletter to just specs reviewing.

     

    NL 66 - 2

    Figure 2.  Transformer specs & waveforms 

     

    🤓 Opening the project, we can see that this 900W LLC Transformer has an input voltage of 400VDC and outputs 300VDC, with a resonant frequency of 100kHz. First check is the turns ratio selected.

    NL 66 - Formula 1

    Rounding up 0.66 to 0.7 checks out in Figure 2.

    No matter if this is a center tapped or full wave rectification, the turns ratio equation is the same.

     

    ⚙️ Remember at resonance the voltage of the Transformer primary is going to be centered around zero (no DC bias at the transformer) with peak-peak value at 400V. Effectively I’m thinking like the capacitor is removing the DC component of the 0-400V pulse wave the half bridge produces, resulting in -200V to +200V voltage at the transformer, if we assume an ideal Transformer and an external series resonant Inductor. That’s why the term Vbulk/2 exists.

    Now that’s settled, I notice that I had selected 125uH for the magnetizing inductance and 25uH for the series resonant inductance.

     

    🔎 Going back to some theory here, I see that I have chosen a ratio of magnetizing to series inductance of:

    NL 66 - Formula 2

    Selecting m values from 2-6 gives is the guideline. Alright, let’s calculate now the max/min gain of the LC tank circuit:

    NL 66 - Formula 3

    📉 In the equations above you see Vblk and Vmin, which are essentially the min/max DC bus input voltage of the LLC Converter. The minimum input voltage requirement can usually be calculated easily from the necessary hold-up time need for the Converter. During this time the voltage of PFC stage (DC bus) in front of the LLC Converter starts dropping as the milliseconds go by.

     

    🖩 Some typical numbers I use for 1kW PFCs look like:

    NL 66 - Formula 4

    Assuming a 400V PFC with a +-5% output ripple and some extra tolerance, I end up with 430V as the max input voltage. Again, your needs might differ from this simple example.

     

    🎬 You can also have a look at two of our exclusive Webinars on the LLC topic:

    • LLC equations and Maxwell.
    • Design considerations and prediction models for LLC designs.
    NL 66 - 3

    Figure 3. Selection of the peak gain – quality factor of the resonant series LC

     

    💡 The LC resonant tank should be chosen such as to be able to give us peak gains higher that Mmax=1.15, which happens during a brownout/hold-up time condition. Following the green graph for m=5 (as selected earlier) and a peak gain value ~20-30% higher at 1.15*1.2=1.4 we find that we need a quality factor Qe=0.4. Choosing a 100nF series resonant capacitor in that application we verify that:

    NL 66 - Formula 5

    ⚡ Notice that the actual switching frequency varies depending on the loading and voltage of the input, as expected. That is calculated in Figure 2 automatically, for you to be able to check easily.

     

    🙌 Now we are ready to design! 

     

    😎 I hope you've enjoyed today's Newsletter. See you in the next one!

     

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

     

    ✨ Get a 25% discount on the Mag & Mad tickets with Frenetic!

    Use the code ROARING25 and get your reduced tickets before September 1st. 

     

    🚀 Save the date: September 13-14, 2023 | Madrid, Spain. 

    Check out the updated conference agenda, speakers and tickets information here!

     

     

    Reference

    1. SLUC675 Calculation tool | TI.com. Available at: ti.com/tool/download/SLUC675/01.00.00.0C (Accessed: 25 August 2023).

     

    Sotiris Zorbas, MSc 

    Power Εlectronics Εngineer 

    +34 915 29 60 07
    www.frenetic.ai/
     
    small logo

    Mag & Mad, Calle Jorge Juan, 137, Madrid, Madrid 28028, Spain

    Unsubscribe Manage preferences