Revisiting this video after 2 months. I see it in a new light. This so far, is indeed the best (read simplest) approach to design of multilevel resonant converters
Thanks Professor! Very clear explanation. Two small things (forgive me if I seem picky): 1) In slide #3 the FBR seems not well connected. I think the DC output should be connected to where the transformer is connected. 2) In slide #23 in the last formula for Re I believe Vin_max should be squared.
Prof, 15:39 the Re equation Re=Vinmax^2/P, here Vin_max should be the Vin_rms value instead of DC value. as in the gain curve simulation and other place, we used ac Vin as the input and Re is the ac rms equivalent resistance.
Thanks Professor. Excellent presentation. I followed along and even did the LTSpice circuit you showed in your presentation at 8:45 But I was perplexed for a while that i could not get the same shape curves you reproduced. Eventually I thought lets try NL=5 so that Lm=5*Lr and suddenly the curves were the same. Haha.
This videolesson is a little different than the others? This way of teaching around an example i mean. It is abnormally helpfull for making a circuit almost the same as this work for real! Or at least with this lesson it is much more easy, in fact it became real easy after watching it a couple of times, ánd calculate along with this lesson! But as always it is also again filled to the top with those so great insightgiving aha-sort of moments and that's what makes all of these videolessons so abnormal worthfull(if that is a real english word?)for me, makes each of them so great to watch!
Thanks Professor for such an easy to understand approach. May I ask if we can use it for CLLC as well ? The FHA deviates widely from time-domain in case of CLLC, but for a starting point do you recommend it for CLLC converter design ?
Thank you professor! Would you consider a follow up video discussing a practical implementation of this - air gaps, handling transformer leakage inductance, etc.? There must be differences in selection between these resonant converters and more traditional buck/boost...?
Hello! I understand (after day of learning and "researching") that the essential parameter is Re. When you know it - you can easily modelling converter on a simulator like LTSpice. Maybe I can be not right - it is the second LLC converter that I am developing.
Thanks a lot professor, I have seeing all your videos about LLC converter are great, do you have any recommendations on how to optimize design the LLC converter for MPPT operation ?
Good day Sir. Would you like to say a couple of words about an another architecture of an LLC converter, when C is moved to the secondary side pls? If you take a look into a transformer driver UCC25800 datasheet, you'll find that it has an advantage as "the magnetizing inductor no longer affects the converter gain. Therefore, the converter is less sensitive to the switching frequency and resonant component tolerances". Besides they offer to incorporate the C into a wienna rectifier, and so remove one of the secondary coils (which can be a big deal for a planar transformer due to a high current and a copper layer thickness). Many applications are good with an open loop gain, and such tolerance would be a benefit
Thanks for the video! When you are calculating Re using output power, you are using Vinmax. Am I correct that theoretically we should use ((4/pi)*Vinmax)/sqrt(2)? This rms value is 0.9 times voltage which is quite close 1 so I understand if you have just left that "away"...
Great video. Although I quite like the LLC topology, I was trying to understand the pros/cons of LLC vs Asymmetric Half Bridge converters. I would love to see one (or more) videos on Asymmetric half bridge converters. :)
This might be a little off the scope of your video but, since control in LLC is by keeping PWM constant approximately 50%duty then varying the switching frequency. Is there a way to keep the switching frequency constant (near or at resonant) and control output voltage by changing PWM duty cycle?
In 0:34 you show a different schematic than the one you simulate in 8:19. In 0:34 the inductors are closer to the HIGH and LOW side MOSFET, while in 8:19 the capacitor is closest. I assume it doesn't affect the basic operation of the LLC converter?
Thanks for this video professor. Regarding the Lm value, how would be your approach in case ZVS is not ensured on the primary side due to lack of magnetic energy?
Although everything seems ok, I think that somewhere there is an error in the proposed method. Why am I saying this? Because in the given numerical example, with the values obtained for Lr and Cr, the equivalent quality factor is 1.7, a value too high to ensure the proposed gain equal to 2. In addition, the working frequency range is very limited because below 96.4kHz ZVS is lost and ZCS enters, not a very good regime for transistors in LLC. According to my calculations, the ok values for Lr and Cr would be 8uH, respectively 330nF (for m=6) or 11uH with 220nF (m=3).
Excellent presentation. Design method is realy user friendly. One question regarding loop design: Which crossover frequency in relation to fsw would you recommend ?
this is great. i think... i think i get it now. trying for an induction heater. sort of running 270nf Cr just cus i got em as pulse rated... meh, i can alter that. from some playing with MPS LLC tool, i got something like 18uHLr, a factor of 6 for for about 74uH on Lm, and roughly 71khz. have to play around a bit more with it, but its interactive graphs do allow one to get a grasp of how the variables affect things. very handy. i havent understood the relation ship of the work coil to Lm, though now... i think i simply DO shunt it, exactly as the diagrams suggest... despite it being only a few turns of copper pipe. for an induction heater, its going to be a 1 turn secondary. i could say my RL is say, the resistance of a cylindrical section of iron (copper?) tube of say 1/3rd radius wall thickness for skin effect. the inductance of this work coil doesnt matter. vo doesnt matter other than its a function of the current being induced through that resistance. what does matter is keeping the turns ratio correct so the reflected load, Re, is approximating the desired value calculated from the power supply side of things so as to deliver the maximum power to either a slug of copper/gold to being unloaded with minimal load, without blowing up... i think that from here... i can work this out now. i recall the industrial unit i used to use ran a fixed frequency, with a coil tap for load matching. this does explain it now. as the work coil may have any number of turns, the matching transformer is simply there to maintain Re as the secondary, the lump of stuff being heated, is always considered a 1 turn coil. though RL will change as the material changes... nevertheless, as long as i calculate for the worst case scenario, a near virtual dead short, someone just stuck a superconductor in the workcoil... i should stay in the safe region...
Thank you very much professor. It is a very good material! Could you tell me please why Lr and Cr are 0.16? I run the same circuit with the same commands in LTSpice and I got gain over 2 also for Rn equal to 3.
Hello Professor, Thank you so much for your well-explained presentation. I am trying to replicate your AC analysis. I have provided all the same values but I am getting negative gains as well.
My graph starts from zero and then dips down to -3.2 Volts and after that it increases same like yours. Can you let me know what I am doing wrong. I am using Octave .ac oct 600 100m 10 command.
Nice presentation, I have play a lot with this stuff the past time, I have learn the regulation with low load"s,did go not so well except when circulation currents are high destroing the beautifull character of this systems, who is not what I want. I need, because I have Idle currents in amplifiers, a open loop supply and use analog regulators for output voltages of filament audio valve, I need to find the best spot for the resonance, I did read that I need to use the resonance point or just left in capacitive region of it for safety. I am right? Thank you very much.
@@sambenyaakov For transformer, I have also here done a sim with ac plot and sweep. I see that when I use primary 1mH or just 470uH that does not do much, all is dependent of Lr Cs and Cp. The reflected power it what matters. I can even do 5mH and sec 581uH is N 8.6 do match the voltages but then switch current do rise again, so I come closer to understanding it, needed before I go test and build it. I think start with 50 volts in system to be safe, later on I can go further. For the transformer I did read a paper who says ; The matching transformer Tr is shown in fig.1 together with its simplified equivalent circuit under the condition that the magnetizing current of the transformer is negligible with respect to the current in the resonant circuit. Then this transformer comprises both the full leakage inductance L S and the natural capacity of the windings С 0 , reduced to the primary winding, as well as an ideal transformer with its transformation ratio equal to k. AI says transformer primary have to be close to Lr resonance coil, but I think he is meaning LLC for this. I have seen a example of LCC for 19 volts 5 amps laptop charger who has a transformer of 5850uH who is quite high, for transformer protecting to not get into saturation to quick inductance needs a high as what will fit on transformer, as I do with the ETD39 what I have do 470uH it get just little windings, that is not right I think for LCC, but for LCC it is, that needs a air gap and LCC do not, here is the culprit I think I fight with.. Nice to see the current source action of LCC, when I double the secondary coil, output voltage just do not change much there, but primary it does. Nice. It is open loop in simplified ac sim. Same with inductance, set it higher or lower do not much, all what matters and give changes is N1/N2. working frequency is 0.85 x F resonance, I have to set up the N1/N2 to higher value then normally done, giving more efficiency, put primairy inductance such that it,s X is higher then the load resistor, but not much did work in sim, getting higher switch did loss more... Yes I now you have the knowledge, but also thinks "Kees", you have to learn. Well this way I do but slowly, I am 67 so brain is slower. But it keeps me busy. regards
Hello Sir , good video. But the whole average of the input voltage will appear across Lm when input is low. which will be exactly to the the required output voltage reflected when we go near to the parallel resonant point and not at series resonant point. Other wise always the average voltage across Lm = TURNS RATION X VOUT, So until we have an input voltage we will get the required output and lower than that no regulation.
Great lecture thanks, dear professor in case of bidirectional if the range of voltages and power keep the same for both directions, the design parameters would be the same? Or is there anything else we should take into account? 🙏🙏
In slide 14 gain consdiered vin max /vin min . Do we need to consider gain is voutput /vinput? Or the considered vinmax /vin min. Please explain this . Thankyou
Looking the gain graph it seems that if we have fixed nominal input voltage all the valid loads we will have cross resonant frequency at gain 1. In real design project I face a problem which is: switching freq is above resonance ( as I can see my resonant current waveform ) and switching frequency is decreasing close to resonance with increasing the load ( or decreasing Q ). I have followed all the steps and I believe everything is ok with my calculations. Pout = 500W; Vo = 12V; n=16; Vin = 390V; Lr = 30uH; Lm = 220uH; Cr = 120nF; fr = 80kHz. Question is: From where could come the difference between theory and practical design. Possible things: transformer ratio is not valid ( but I checked that ) or leakage inductance is dropping with respect to current ( something I don't have the tools to measure ) , but anyways If so i think the circuit should operate at the new resonant frequency to follow the approximations in the predesign but that is not happening. This is something I don't understand.
in 20:07 min you said you use Vin_max for calculate Re value.. so is it the Vin_max = Vcc_max/2 ? am i right or wrong? here i mentioned maximum limit of supply voltage which gives to the half bridge terminal as Vcc_max
Revisiting this video after 2 months. I see it in a new light.
This so far, is indeed the best (read simplest) approach to design of multilevel resonant converters
Thanks
Very elegant! Captures the real essence of the problem in such a visual and easy to follow way. Thank you Professor!
Thanks for comment.
One of the best explanation I have ever heart. Short…on the point…simple. Thanks Professor.
Thanks for warm note.
Thanks proffesor, nice design approach! You made it simple and clear.
Hi Raanan, I am glad you liked it. Hope you are doing well.
thanks for this much-simplified explanation, sir. Its really treat to hear you
Thanks😊
Thanks Professor! Very clear explanation.
Two small things (forgive me if I seem picky):
1) In slide #3 the FBR seems not well connected. I think the DC output should be connected to where the transformer is connected.
2) In slide #23 in the last formula for Re I believe Vin_max should be squared.
You are correct. 3 The bridge is reversed. An alertness test to watchers. 23. Right again. Whoever does not err is probably doing nothing😊
Prof, 15:39 the Re equation Re=Vinmax^2/P, here Vin_max should be the Vin_rms value instead of DC value. as in the gain curve simulation and other place, we used ac Vin as the input and Re is the ac rms equivalent resistance.
Yes there is a mistake here. Re should be R equivalent as in slide 8. So first Vdc^2/P and the multiply by 8/Pi^2. Thanks for pointing this out.
Thanks Professor. Excellent presentation. I followed along and even did the LTSpice circuit you showed in your presentation at 8:45 But I was perplexed for a while that i could not get the same shape curves you reproduced. Eventually I thought lets try NL=5 so that Lm=5*Lr and suddenly the curves were the same. Haha.
I am happy you passed the test😀
@@sambenyaakov Haha. Have a blessed day.
Thank you Professor. The explanation is very clear and concise.
Thanks
This videolesson is a little different than the others? This way of teaching around an example i mean. It is abnormally helpfull for making a circuit almost the same as this work for real! Or at least with this lesson it is much more easy, in fact it became real easy after watching it a couple of times, ánd calculate along with this lesson!
But as always it is also again filled to the top with those so great insightgiving aha-sort of moments and that's what makes all of these videolessons so abnormal worthfull(if that is a real english word?)for me, makes each of them so great to watch!
Thanks Rob for taking the time to write the extended comment
Thanks Professor for such an easy to understand approach.
May I ask if we can use it for CLLC as well ?
The FHA deviates widely from time-domain in case of CLLC, but for a starting point do you recommend it for CLLC converter design ?
Thanks. I guess it could. Never done it. If you try, please share the results.
Thank you professor! Would you consider a follow up video discussing a practical implementation of this - air gaps, handling transformer leakage inductance, etc.? There must be differences in selection between these resonant converters and more traditional buck/boost...?
Thanks for comment. Integrated magnetic is indeed a great topic. I will consider developing a video on that.
Hello! I understand (after day of learning and "researching") that the essential parameter is Re. When you know it - you can easily modelling converter on a simulator like LTSpice. Maybe I can be not right - it is the second LLC converter that I am developing.
Indeed!
Thanks a lot professor, I have seeing all your videos about LLC converter are great, do you have any recommendations on how to optimize design the LLC converter for MPPT operation ?
Thanks. Did not have a chanced to work on that.
Awesome explanation. Planning to take a template print out of normalized gain curves for m=6 for fast hand calculation.
👍🙏
Good day Sir. Would you like to say a couple of words about an another architecture of an LLC converter, when C is moved to the secondary side pls? If you take a look into a transformer driver UCC25800 datasheet, you'll find that it has an advantage as "the magnetizing inductor no longer affects the converter gain. Therefore, the converter is less sensitive to the switching frequency and resonant component tolerances". Besides they offer to incorporate the C into a wienna rectifier, and so remove one of the secondary coils (which can be a big deal for a planar transformer due to a high current and a copper layer thickness). Many applications are good with an open loop gain, and such tolerance would be a benefit
Thanks for the video! When you are calculating Re using output power, you are using Vinmax. Am I correct that theoretically we should use ((4/pi)*Vinmax)/sqrt(2)? This rms value is 0.9 times voltage which is quite close 1 so I understand if you have just left that "away"...
Indeed. Thanks for pointing this out.
Outstanding Job.
Thanks for all🙂
Great video. Although I quite like the LLC topology, I was trying to understand the pros/cons of LLC vs Asymmetric Half Bridge converters. I would love to see one (or more) videos on Asymmetric half bridge converters. :)
You mean asymetrical PWM not resonance?
@@sambenyaakov yes. That's right
This might be a little off the scope of your video but, since control in LLC is by keeping PWM constant approximately 50%duty then varying the switching frequency.
Is there a way to keep the switching frequency constant (near or at resonant) and control output voltage by changing PWM duty cycle?
Thanks for the conversation. Yes, duty cycle control can be implemented to reduce the frequency span.
what is switching frequency select sir.
can you make video on Cockcroft-Walton multiplier with this resonant converter for p power
1. Please indicate the minute or slide number you ae referring to.
2. Good subject. Will try
In 0:34 you show a different schematic than the one you simulate in 8:19.
In 0:34 the inductors are closer to the HIGH and LOW side MOSFET, while in 8:19 the capacitor is closest.
I assume it doesn't affect the basic operation of the LLC converter?
Lr and C are in series so it does not matter where they are along the line. BTW I guess you noticed the wrong connation of the diodes?
Thanks for this video professor. Regarding the Lm value, how would be your approach in case ZVS is not ensured on the primary side due to lack of magnetic energy?
One way is to used packet control.
@@sambenyaakov Is this similar to burst mode?
@@GokhanSen-k8o exactly
Thank you sir
Although everything seems ok, I think that somewhere there is an error in the proposed method.
Why am I saying this?
Because in the given numerical example, with the values obtained for Lr and Cr, the equivalent quality factor is 1.7, a value too high to ensure the proposed gain equal to 2. In addition, the working frequency range is very limited because below 96.4kHz ZVS is lost and ZCS enters, not a very good regime for transistors in LLC.
According to my calculations, the ok values for Lr and Cr would be 8uH, respectively 330nF (for m=6) or 11uH with 220nF (m=3).
Thanks for sharing your thoughts. But how can I respond to "according to my calculations"? You seem to overlooked the "basic" in the title.
Hii sir, thank you for nice explanation
I want to know that is there any effect of filter capacitor of outside in this resonances converter.
You mean in parallel to the load? It has no effect on the resonant if large enough.
Excellent presentation. Design method is realy user friendly.
One question regarding loop design: Which crossover frequency in relation to fsw would you recommend ?
1/5 and below
@@sambenyaakov Thank you, Professor
this is great. i think... i think i get it now.
trying for an induction heater. sort of running 270nf Cr just cus i got em as pulse rated... meh, i can alter that. from some playing with MPS LLC tool, i got something like 18uHLr, a factor of 6 for for about 74uH on Lm, and roughly 71khz. have to play around a bit more with it, but its interactive graphs do allow one to get a grasp of how the variables affect things. very handy.
i havent understood the relation ship of the work coil to Lm, though now... i think i simply DO shunt it, exactly as the diagrams suggest... despite it being only a few turns of copper pipe.
for an induction heater, its going to be a 1 turn secondary.
i could say my RL is say, the resistance of a cylindrical section of iron (copper?) tube of say 1/3rd radius wall thickness for skin effect.
the inductance of this work coil doesnt matter. vo doesnt matter other than its a function of the current being induced through that resistance.
what does matter is keeping the turns ratio correct so the reflected load, Re, is approximating the desired value calculated from the power supply side of things so as to deliver the maximum power to either a slug of copper/gold to being unloaded with minimal load, without blowing up...
i think that from here... i can work this out now.
i recall the industrial unit i used to use ran a fixed frequency, with a coil tap for load matching. this does explain it now. as the work coil may have any number of turns, the matching transformer is simply there to maintain Re as the secondary, the lump of stuff being heated, is always considered a 1 turn coil. though RL will change as the material changes... nevertheless, as long as i calculate for the worst case scenario, a near virtual dead short, someone just stuck a superconductor in the workcoil... i should stay in the safe region...
Thanks for sharing
Thank you very much professor. It is a very good material! Could you tell me please why Lr and Cr are 0.16? I run the same circuit with the same commands in LTSpice and I got gain over 2 also for Rn equal to 3.
0.16 is 1/2Pi to get a frequency of 1Hz. You ran the simulation with m(NL) =1 I ran it with NL=6
Thank you sir!
Hello Professor, Thank you so much for your well-explained presentation. I am trying to replicate your AC analysis. I have provided all the same values but I am getting negative gains as well.
My graph starts from zero and then dips down to -3.2 Volts and after that it increases same like yours. Can you let me know what I am doing wrong. I am using Octave .ac oct 600 100m 10 command.
I am not sure I follow. Perhps negative in db?
Hello Sam, do you have any suggestions on minimum load regulation for LLC converters. Using SiC mosfets and SiC diodes.
I would try burst mode for very low load.
Nice presentation, I have play a lot with this stuff the past time, I have learn the regulation with low load"s,did go not so well except when circulation currents are high destroing the beautifull character of this systems, who is not what I want. I need, because I have Idle currents in amplifiers, a open loop supply and use analog regulators for output voltages of filament audio valve, I need to find the best spot for the resonance, I did read that I need to use the resonance point or just left in capacitive region of it for safety. I am right? Thank you very much.
Thanks for sharing
@@audiokees4045 Great. Thanks for sharing
@@sambenyaakov For transformer, I have also here done a sim with ac plot and sweep. I see that when I use primary 1mH or just 470uH that does not do much, all is dependent of Lr Cs and Cp. The reflected power it what matters. I can even do 5mH and sec 581uH is N 8.6 do match the voltages but then switch current do rise again, so I come closer to understanding it, needed before I go test and build it. I think start with 50 volts in system to be safe, later on I can go further.
For the transformer I did read a paper who says ;
The matching transformer Tr is shown in fig.1 together with its simplified equivalent circuit under the condition that the magnetizing current of the transformer is negligible with respect to the current in the resonant circuit. Then this transformer comprises both the full leakage inductance L S and the natural capacity of the windings С 0 , reduced to the primary winding, as well as an ideal transformer with its transformation ratio equal to k. AI says transformer primary have to be close to Lr resonance coil, but I think he is meaning LLC for this. I have seen a example of LCC for 19 volts 5 amps laptop charger who has a transformer of 5850uH who is quite high, for transformer protecting to not get into saturation to quick inductance needs a high as what will fit on transformer, as I do with the ETD39 what I have do 470uH it get just little windings, that is not right I think for LCC, but for LCC it is, that needs a air gap and LCC do not, here is the culprit I think I fight with..
Nice to see the current source action of LCC, when I double the secondary coil, output voltage just do not change much there, but primary it does. Nice. It is open loop in simplified ac sim. Same with inductance, set it higher or lower do not much, all what matters and give changes is N1/N2. working frequency is 0.85 x F resonance, I have to set up the N1/N2 to higher value then normally done, giving more efficiency, put primairy inductance such that it,s X is higher then the load resistor, but not much did work in sim, getting higher switch did loss more...
Yes I now you have the knowledge, but also thinks "Kees", you have to learn. Well this way I do but slowly, I am 67 so brain is slower. But it keeps me busy.
regards
Hello Sir , good video. But the whole average of the input voltage will appear across Lm when input is low. which will be exactly to the the required output voltage reflected when we go near to the parallel resonant point and not at series resonant point. Other wise always the average voltage across Lm = TURNS RATION X VOUT, So until we have an input voltage we will get the required output and lower than that no regulation.
Looking at it as or series or parallel resonant does not tell the full story. This is a multiresonant converter.
Great lecture thanks, dear professor in case of bidirectional if the range of voltages and power keep the same for both directions, the design parameters would be the same? Or is there anything else we should take into account?
🙏🙏
You need to examine the situation from both sides.
@@sambenyaakov yes I got it, thanks 🙏🌺
Lr and Cr is calculated for resoannt frequency or switching frequency. What is the difference between those.
For resonant, operation is at the switching frequency
@@sambenyaakov 👍
In slide 14 gain consdiered vin max /vin min . Do we need to consider gain is voutput /vinput? Or the considered vinmax /vin min.
Please explain this . Thankyou
Looking the gain graph it seems that if we have fixed nominal input voltage all the valid loads we will have cross resonant frequency at gain 1.
In real design project I face a problem which is: switching freq is above resonance ( as I can see my resonant current waveform ) and switching frequency is decreasing close to resonance with increasing the load ( or decreasing Q ). I have followed all the steps and I believe everything is ok with my calculations. Pout = 500W; Vo = 12V; n=16; Vin = 390V; Lr = 30uH; Lm = 220uH; Cr = 120nF; fr = 80kHz.
Question is: From where could come the difference between theory and practical design.
Possible things: transformer ratio is not valid ( but I checked that ) or leakage inductance is dropping with respect to current ( something I don't have the tools to measure ) , but anyways If so i think the circuit should operate at the new resonant frequency to follow the approximations in the predesign but that is not happening. This is something I don't understand.
The design is based on the first harmonic approximation. In reality the time domain behavior is different.
in 20:07 min you said you use Vin_max for calculate Re value.. so is it the
Vin_max = Vcc_max/2 ?
am i right or wrong?
here i mentioned maximum limit of supply voltage which gives to the half bridge terminal as Vcc_max
Approximately ( square wave- sinus) in a half bridge
Hii, what happens with input source current waveforms and its peaks values? If it high then who to reduce it?
This will be like nay over current protection.
Do you have any references either research articles or book that talks about the use of normalized resistance while assuming Z=1?
Sorry no. This is original with me.
@@sambenyaakov it's ok. I really learnt a lot from the presentation
An app note from TI named slup263 describes something similar.
Does the math work the same for an "inverter", ie AC output, not DC.
For an inverter, we'd lose the rectifier. But would we lose Cout?
No, the analysis assumes a DC clamping at the output, that is the voltage on the transformer is a square wave.
@@sambenyaakov Which video do you recommend for an LLC inverter, with sinusoidal AC output?
@@johnaweiss This would be a resonant converter look up my video on that.
@@sambenyaakov An inverter is a converter?
@@johnaweiss Strictly speaking ' a converter is DC-DC an inverter DC-AC, but sometime they are mixed
Excellent, thanks.
Thanks
I need a simplified version of this video. 🙂
🤔 I think it is as simple as it can get. Perhps LC or just C?
@@sambenyaakov I didn't mean a simpler circuit, just a simpler recipe for picking the LLC values.
🙏🙏🙏
🙏😊
can you attach the simulation file please?
Sorry, too time consuming to locate.
how to calculate Rn if the Vin max and min 50-150
Look for Rn that will give you a gain of >3
@@sambenyaakov do not get it sorry
IMHO at 0:33 dio at secondary side conected not correctly )
Yes this is a typo, already mentioned by many
Mark