Resonant LLC converter power stage design: the intuitive approach

An intuitive approach to the design of the power elements of a resonant LLC converter

[yeshasrgowda]

I never expected to learn this concept in such a easy manner, thanks a lot professor.

[angshumansharma]

Thank you Dr. Yaakov for this simple yet effective explanation. At 16.12, Amax/Amin = Vin, max/Vin, min = Vac, max/Vac, min.

[guidodiiorio]

Dr Yaakov thanks for the amazing help to give to all of us interested in power conversion. I took inspiration from this video to start designing an llc dcdc converter and I found the power section with magnetic components and switching part rather simple to implement. However I got stuck with the control section when it comes to phase gain loop simulation. I used to do this in frequency domain with conventional pwm loops but the VCO I used to modulate the working frequency of the llc works only in time domain. I am sure there is something to solve this problem in an elegant way rather than do it point by point with different frequency settings. Thanks for your priceless information.

[inico]

This teaching is pure gold, thank you very much.

[ArifKhan-bpzx]

Its amazing, now I can calculate a lot of things for power supplies, now i know why we have specific value of components in Resonant tank circuit, and Helped me to understood the design of Corsair AX1200i, Coolermaster V850V2 PSUs .. Great video

[bvc]

Thank you so much for this simple and effective explanation. I wonder of you are planning to make a lecture on the closed loop control design of the LLC converter

[zeroinfinityx]

Thank you for such a clear and concise explanation. Can you please review vac min expression at 19:50? I think it should be 2/pi, not 4/pi.

[DavidGillooly]

A very enjoyable presentation. I particularly liked the way you used the normalized resonant graphd.

[willwork]

What a great video. thank you so much professor. I have a question that can we put res tank in 2nd side of the LLC converter for a step-up design when Np:Ns is < 1, pri current will be much greater than 2nd side. Is it possible to equivalent to pri side?

[frajbenarab]

Dr Ben Yaakov thank you for your help and your patience while you were explained this video, i was wondering if you could explain in 18 "minute, you said that we should not choose the curves desoite they are above the minimum gain, by the meaning we are working in the capacitive region, why we should not operate in this region and what it can causes?.thank you

[audiokees]

Hi There Sam, All things oké there?. I was away for a time because of private problems and helping a friend with cancer who has died 3 months ago. Now I am started again with some electronics like the LLC. I had to see if for the amp a LCC or LLC is best, with good design the LLC do oke, special for the tubes who draw a constant current from the smps who is oké. I had trouble with calculation of the transformer, who now is kind of solved but still I did get fault outcomes with the voltages, now I see that when the smps do work with a 310 volt input the voltage on the transformer is higher because of the resonance working, slingering the voltage higher depending on Q. Now I have use the voltage over the transformer as input for the equations and now it fit like a couple of shoes. the voltage do have the right value. Is this something you do recognize? A lot of things I seems to wacht for when build such a converter. For a amplifier the LLC can do fine, without regulation, with it the LLC do not keep track the dynamic of the amplifier who draw complicated currents because of music. I use a LLC in open loop where I set the frequentie very close to resonance, and yes it keeps the voltages quite stable, like a normal supply 50 Hz with elco, s but without the bad 100 Hz ripple. It seems a LCC in open loop is even more stable when using around resonance. have a fine day or evening.

[fabriziocortigiani]

Thanks a lot Dr Yaakov! One Q about slide 21: Should it be Amax/Amin = Vmax/Vmin instead? Where am I mistaking? The content looks clear anyway, thanks again!

[Penguins]

I have gone through this video many times, I have a question that why Lm is not being used on Qmax and fr calculations??..Thank You

[nabilkerbila]

thank you for this valuable explanation, I have a question for you, is this topology adequate for high power, low output voltages ("The output current variation would be greater than for a PSFB ? Or no ? ")

[BS]

Thank you for the great explanations for analog circuits. I have a question please: how do you obtain the voltage gain curves like the ones shown at min 11 or 13 for example?

[power-max]

Hi Sam, I got back around to finishing my LLC converter design. I found a very helpful article by Infineon (URL below). and I outline my work below. My design uses a 48V battery (3.0Vmin, 3.7Vnom, 4.2Vmax, 13S) to 2KW at 175V. The realized design in LTspice that really overdelivers! It can push 15A, 175V, 2.6KW under full load with minimum input voltage, and at no load, max input voltage, It under-delivers, providing 170V. That's good it means I can keep it in regulation with a tighter frequency range.

The values I calculated are as follows:

Cr = 8.5uF
Lr = 1.2uH
Lm = 7.2uH
Ls = 95uH

The *PROBLEM* is the 200A peak currents (guessing around 130A rms, not quite sinusoidal at the max load min input voltage). The primary voltage also resonates up to 130Vpk. This design "works" on paper but does not sound reasonable to me. Where did I go wrong?


I picked 3.7V to be the nominal voltage for this design.
* Amax = 3.7/3.0 = 1.23 (max load and min input voltage edge case)
* Amin = 3.7/4.2 = 0.88 (no load and maximum input voltage edge case)
* Fmax = 5 Hz normalized. (arbitrarily chosen to avoid a huge frequency range)
* f = 50kHz (arbitrarily chosen so fmax = 250kHz and fmin = 24kHz)

*Step 1: find suitable K value.*
With no load, Q -> ∞; Amin of 0.88 or less, frequency is limited to 5, 'K' to be no higher than 6. As I understand K should be as high as possible.
With K = 6, Amin is 0.86, which is acceptable.

*Step 2 and 3: find Qmax and Fmin*
The highest Q factor that supports Amax = 1.23 was found to be about 0.4. Q=0.4 allows Amax to reach as high as 1.28, at this peak which occurs at 0.422 Hz (normalized)

*Step 4: find Rac_min*
Rac_min = 8/π^2 * (Np/Ns)^2 * Vout^2/Pmax
Rac_min = 8/π^2 * (48.1V/175V)^2 * 175^2/2000 = 0.9376

*Step 5, lots of arithmetic!*
Lr/Cr = Qmax * Rac_min
Lr*Cr = (2πf)^-2
Lm=K*Lr

Rounded, I end up with the following values:
Cr = 8.5uF
Lr = 1.2uH
Lm = 7.2uH
Ls = 95uH

Is this correct, and what can be iterated to make this design realizable?

[salahalatai]

how to calculate or know the range of switching frequency Fsmin, Fsmax that the converter can work on ?

[samelias]

Hi Professor, I have a question, please. You marked A_max and A_min on the plot. How did you get them? I'm asking for two reasons 1-I have a V_in of 35-75(v) and V-out of 2(v), and so V_out/V_in is very small while your gain region is around 1. 2- I also saw somewhere that A should be (n).(V_out/V-_n), but in your lecture, we have the gain first, but we calculate n towrds the end. Thank you

[alibughneda]

I would like to thank you for the useful information.
Second, I have some questions that I hope to get clear answers to help me design my circuit:

1-Could I use an LLC Resonant tank with a multilevel inverter Or it is limited to use on half-bridge and full-bridge only. If I use it with a multilevel, does it affect the waveform of the output of the inverter?

2-What is the output voltage waveform of the resonant tank (VLm)? Is it the same as the waveform of the output voltage of the inverter (Vinv) or different?

[power-max]

Hi Sam, I built a spreadsheet which computes all component values provided design requirements and operating frequency range! Now the next step is to close the loop. Do you have videos that walk through making it stable?

I've noticed that at F_res, the output impedance is very low. But at low or high input voltage requiring boost or buck operation, the output impedance increases due to one of the reactive components dominating, which changes the dymamics of the plant. How is this stabilized? Can I use the same methods and assumptions used for PWM controllers?