Nyquist plot was something I used to shy away from. This video helped me to come out of it. The controller design part was also excellent. Thank you so much
The best part for me was "Nyquist plot is not user friendly". For the first time I heard a professor saying so. Thank you professor for such an elaborate presentation.
Sir, you have such command in your presentation. The points of interest are succinct without unnecessary rabbit trails except keys to watch for. I prefer this over the videos that incorporate fancy graphics that do not align with the message. Great work. Be safe.
It would be so great to see how to integrate input filter, control loop and output filters characteristics for a reliable and stable DC/DC design. Thank you for your wonderful videos!
Hello Prof Yaakov, I have watched more than 30 videos of yours. I see that the way you present the concepts in your videos are unique (in the sense of what I have been taught in school & what I see online). They are simpler to understand & easier to grasp. Your videos are an imitation of your experience & knowledge. I thank you so much for your wonderful service in making such amazing contents.
@@sambenyaakov Hello Prof are there any videos on LLC Converter compensation or controller design? I feel it would be beneficial as there is very little content. I also understand they are a bit more complicated.
Excellent explanation. Thank you for it. I'm a lecturer myself and have to explain controller design for DC/DC converters. My good luck that I have found your videos right when I am in the process of redoing my slides. Again, thank you for this very good explanation.
Thanks Prof. Excellent introduction to controller design. Some of my wish list for your lectures 1. PCB Layout consideration for SMPS 2. digital controller design for pwm converters 3. dc/ac converter topologies & design Keep doing great work.
@@sambenyaakov I really would like some guidance or lectures on PCB design for power converters, I am a MSc student but universities have failed teaching me this.
Thanks Professor Yaakov for lectures. I really appreciate if continue these lectures. I hope you explain inverters (2levels and 3 levels), space vector, ...
Very nice lecturers, subscribed for a long time, and often come back to review a subject of interest, thanks for your effort! As a side note, it would be interesting if you could explain how to achieve different V/A characteristics in a welding inverter, i.e. how to make different slopes on V/A relations... I assume for analog feedback network it is only a feedback network parameter?
Thank you! As you might have guessed, it is the time that I need to build a power supply to run my device at its full power. Thank you for your lecture series, I will talk about them on my blog.
Thanks prof. I personally prefer the explanation where you cancel poles with zeros and zeros with poles to realize a -1 slope at the 0dB line. It just seems easier to understand.
Hi, at 25:52 (slide 21), you mention that the phase margin would be very bad, and that it would be an unstable system. I agree with the phase margin being bad, but I think the system is still stable. Theoretically, two poles cannot cause an unstable system (180 degree shift at infinite frequency). Anyways, thank you for the lectures. They helped me a lot.
Dear Sam, in case of LLC resonant dc-dc converter, do you suggest to use type II compensator or type III compensator? I read on SLVA662 T.I. paper that they say that for voltage mode we must use type III compensatori, what about your technical opinion? Thanks in advance for your reply. BR/Roberto
As far as I know that app note is for PWM. In case of LLC you need to consider the open lop transfer function. See: th-cam.com/video/amRqIgH10Rg/w-d-xo.html
Thanks for your great presentation. Here is a question about the feedback loop. At 3:17 ,the voltage divider is included in 'B', but at 6:15 , it is in 'A'. What is correct one?
Excellent explanation prof. I only have one doubt in the "Bode plane" slide (min 13) in the phase plot I think that instead of -180 this has to be -90 degrees. For the rest i think you are the best❤
Very good explanation! Thanks for your interesting videos! I'm sorry for that, but I guess there is a little mistake at 15:10 in the phase plot of Bode diagram; if it was intended to represent the same system of the Nyquist diagram, there could be an error near the phase margin point, that appears to be negative instead of positive. I mean that the phase is going down too fast in that plot, and that's not the situation represented by the Nyquist plot. To be coherent, the phase at 0dB should be about - 120°. Am I right?
Hi Professor, Thank you for making this video. It was an eye opener for me. I need some clarity about on of the points shared in the video. By drawing a bode plot for a closed loop system, why are we plotting AB and not A/(1 + AB) which is nothing but ~1/B ? Effectively my questions is, why are we doing analysis on just loop gain (AB) and not the whole gain of the system (A/(1+AB))?
Sir, you were saying that using rate of closure of A & (1/B) Plots, we can find the phase margin. During -20db rollover phase is -90deg, phi_m =90deg and at corner frequency phase is -135deg, phi_m =45deg. But sir, these phase margins at those points are corresponding to only plot A, but you were telling that rate of closure of two plots gives these phase margins, i feel you didn't add the phase of (1/B) plot to phase of plot A to find the overall system phase margin.
The -20db/dec is between the A and 1/B. This means that if you will plot it like in th-cam.com/video/L3MRnEYdi8g/w-d-xo.html the crossing will be at -20db/dec, becuse the different(which is BA) is -20db/dec.
Dear professor sam, Thanks for the great video. I am writing a report and love your practical way of designing a control loop. Do you have any litterature that i can refrence in my report :-)
I came up with this method many years ago, I have seen it somewhere, don't remember where :-). Have a look at this three part tutorial th-cam.com/video/DfCuNjHDRHE/w-d-xo.html th-cam.com/video/bGj1huz2Log/w-d-xo.html th-cam.com/video/Bn7QPJaj-vU/w-d-xo.html
Thanks Professor Yaakov for the great explanation. I have one question. When analyzing the current sense method whereby you measure the voltage across the inductor DCR, are you assuming a sinusoidal waveform? Otherwise the impedance values would not be correct. If you are, then you are using a first harmonic analysis?
Hi Chuck, sinusoidal waveform is not assumed, The output is correct for any (within practical limitations) frequency and combination thereof (a triangular ripple ). See th-cam.com/video/sqQzBxfECj4/w-d-xo.html
Sir in nyquist criteria you were saying stability effected due to phase lag but while looking in to plot the encircling also mainly depends on radius(|LG|) of the plot. But you didn't speak about magnitude of |LG| .
Thanks Dr Sam you are wonderful , it is possible Dr sam to talk about ,how we design power factor inductor in SMPS and also choose the right TVS and MOV's.
@@sambenyaakov I am grateful,if you could show how to implement a practical approach of z-parameter to design smps like your previous feedback loop step by step design video.
hi, Prof Ben-Yaakov Thank you for your lecture! One question, given the same cross-over frequency, is higher phase margin always better? Assume 80 deg vs 70 deg PM. Both are stable for sure. Could you shed some light on this? Thank you!
@@sambenyaakov Thank you. Do you mean a load step or a input step? For SMPS, I assume it is often times as a load step. For this reason, I think you're saying that given a load step, the 70deg PM will have less rise time than 80 deg. Correct? Thank a lot
Sir,I have a doubt regarding bode plot method,usually we use bode plot or nyquist technique for linear systems right?but here we are determining stability for power modulator which is actually a non linear device,how it could be possible to find out the compensation using this bode plot?Please clear this doubt sir as I am working on one project and I need to publish one paper regarding non linearity of power electronic devices.Thank You
Hi, the classical method of switch mode systems control design (which are indeed non linear) is to use averaging to obtain a linear representation of the system, or to use average simulation models which I am showing.
Sir, I didn't understand what you were trying to say, were you saying that you are approximating the nonlinear device into the linear device?And I want to ask one more question.Are time domain techniques really useful to find out the stability of nonlinear devices or is it better to go with the state space techniques for nonlinear devices?
Thank you very much , there is something i don't understand in DC and in AC. In DC for exemple the compensation network do not have effect when V(-)=V(+) we have output of the amplifier should be 0V . But as to produce the PWM signal we compared it with SAWTOOTH. The result of this comparaison will give d.c~100% and that make no sense , because when V(-)=V(+) we should have Vout=D.C*Vout .
@@sambenyaakov thank you for your quick answer. I understood my problem. Because at school we told us that Vout=D.c*Vin, but, and that make no sense when Errors is equal to zero because your duty cycle is zero. finally it same like we care about that relationship when we do not behave Fb. With FB, we focus on Vref and Vout hence it makes sense to have a duty cycle of zero when vref=vout and Vin is there only to give the sufficient needed power. Finally if i check D.c with a oscilloscope is never constant but permanent changes. This was my confusion
@@sambenyaakov the Pin Vref(V+) of the EA is the desired output and the Vout mesured( V-) is input Voltage . Hence when the Vout mesured( V-) change , the dutycycle adjust in order to have the good Vref. Vout=D.C*Vin it's finally used only for design of components once everything is turning , we can not observe directly since duty cycle is set, thank to vref of EA and vout. Am i right?
Pretty good. IN proper designs (e.g. PI control) the gain of the error amplifier at DC is very large so it takes only a minute deviation from Vref to build up the voltage required as input to modulator . See th-cam.com/video/04gJ1t2cQdA/w-d-xo.html
Dear Professor, I really appreciate your amazing lectures. I hope you be healthy for a long time to teach more and more generations.
Thanks for warm words. Comments like your keep me going.
Nyquist plot was something I used to shy away from. This video helped me to come out of it. The controller design part was also excellent. Thank you so much
👍😊
Hard stuff, hard to digest, but cannot be ignored when designing a controller
There are more complicated ways to teach this 😊
Thanks for summarising a large part of controller design in 29 min.
👍🙂🙏
The best part for me was "Nyquist plot is not user friendly". For the first time I heard a professor saying so. Thank you professor for such an elaborate presentation.
😊.Thanks for comment
Nyquist plot vs Bode plot is very intuitive👍 Recently I find that intuition only comes from deep understanding. Thank you professor!
Thanks
This was the cleanest explanation of phase margin I've ever seen....... Thank you very much Prof. Ben-Yaakov!
Thanks. 😊
Sir, you have such command in your presentation. The points of interest are succinct without unnecessary rabbit trails except keys to watch for. I prefer this over the videos that incorporate fancy graphics that do not align with the message. Great work.
Be safe.
Thanks for kind words. Comments like yours keep me going.
Your videos are the best. Thank you for posting them.
Dear sir every time I watch a lecture from you. I learn something new. Be Healthy and take care.
Thanks. Comments like yours keep me going.
It would be so great to see how to integrate input filter, control loop and output filters characteristics for a reliable and stable DC/DC design. Thank you for your wonderful videos!
Great suggestion! Wii see.
Thank you Sam! Fantastic video. Very easy to follow and I liked the controller design approach.
Thanks
Hello Prof Yaakov,
I have watched more than 30 videos of yours. I see that the way you present the concepts in your videos are unique (in the sense of what I have been taught in school & what I see online). They are simpler to understand & easier to grasp. Your videos are an imitation of your experience & knowledge. I thank you so much for your wonderful service in making such amazing contents.
Thank you Lakshmi for the kind note. Comments like yours keep me going. Welcome to join
www.linkedin.com/groups/13606756
@@sambenyaakov Hello Prof are there any videos on LLC Converter compensation or controller design? I feel it would be beneficial as there is very little content. I also understand they are a bit more complicated.
Really thank you so much sir. Now control systems’ mysterious looking sides makes a lot sense in real applications.
Thanks
Your Videos have the best in-depth technical content with easy to understand language. Thank you Professor
Thanks
Excellent lecture, Mr. Ben-Yaakov, as usual!
Thanks
excellent excellent presentation and very well explained. Thanks for that. I will recommend everyone to follow your Courses. you are great.
Amazing class!
Thanks
Excellent explanation. Thank you for it. I'm a lecturer myself and have to explain controller design for DC/DC converters. My good luck that I have found your videos right when I am in the process of redoing my slides. Again, thank you for this very good explanation.
Thanks for comment. I am happy the video is useful .
excellent explanation ! Instead of losing me in algebra ( which is meaningless for me !) Bode plot unveils the feedback problem.
Thanks
Thanks Prof. Excellent introduction to controller design.
Some of my wish list for your lectures
1. PCB Layout consideration for SMPS
2. digital controller design for pwm converters
3. dc/ac converter topologies & design
Keep doing great work.
Will try. Thanks.
@@sambenyaakov I really would like some guidance or lectures on PCB design for power converters, I am a MSc student but universities have failed teaching me this.
Thank you Professor for the intuitive explanation.
👍
Very thankful to you prof
Thanks
very good explanation. and it is very useful to design a power supply system
Thanks for comment.
Thanks Professor Yaakov for lectures. I really appreciate if continue these lectures. I hope you explain inverters (2levels and 3 levels), space vector, ...
Thanks for comment. Will try.
Thanks so much DR Ben, your lectures help me a lot ,please continue post your lectures
Thanks for continued encouragement and warm words.
Perfect 👍👏🙏🌿🌹 Thank you
So nice of you
Great work Prof.
Thanks
Very nice lecturers, subscribed for a long time, and often come back to review a subject of interest, thanks for your effort! As a side note, it would be interesting if you could explain how to achieve different V/A characteristics in a welding inverter, i.e. how to make different slopes on V/A relations... I assume for analog feedback network it is only a feedback network parameter?
Thanks for comment. I do have experience in electronic welding machines. Perhaps.
Very intuitive. Thanks you so much.
Thank you Professor!
Hi Dick, thanks for comment. I just looked at you :A Linear permanent magnet synchronous motor for Needle-free Jet Injection
Impressive.
Thank you! As you might have guessed, it is the time that I need to build a power supply to run my device at its full power. Thank you for your lecture series, I will talk about them on my blog.
Thanks Prof. for nice explanation. Waiting for more videos.
Thanks for comment.
Thanks prof. I personally prefer the explanation where you cancel poles with zeros and zeros with poles to realize a -1 slope at the 0dB line. It just seems easier to understand.
While the explanation might be simpler, the application involves a more tedious trial and error procedure to get the desired phase margin.
Ok, thanks for that. I've never had a problem but I'm sure you're correct and will re-look at your method.
Hi, at 25:52 (slide 21), you mention that the phase margin would be very bad, and that it would be an unstable system. I agree with the phase margin being bad, but I think the system is still stable. Theoretically, two poles cannot cause an unstable system (180 degree shift at infinite frequency).
Anyways, thank you for the lectures. They helped me a lot.
Under damped even if not oscillation is not good. Fur her, the circuit always include parasitic capacitances that cause further phase delay.
Dear Sam,
in case of LLC resonant dc-dc converter, do you suggest to use type II compensator or type III compensator? I read on SLVA662 T.I. paper that they say that for voltage mode we must use type III compensatori, what about your technical opinion?
Thanks in advance for your reply.
BR/Roberto
As far as I know that app note is for PWM. In case of LLC you need to consider the open lop transfer function. See: th-cam.com/video/amRqIgH10Rg/w-d-xo.html
Fantastic work. Thank you.
Very nice explanation
Thank you! There is a way to plot bode/Nyquist of an DC/DC control loop convert with LTspice? maybe you have a video about that already?
There is a scrpt to do that try gooling. I prefer to apply average models. There are a number of videos on that in my TH-cam channel.
great
🙏🙂
Thanks for your great presentation. Here is a question about the feedback loop. At 3:17 ,the voltage divider is included in 'B', but at 6:15 , it is in 'A'. What is correct one?
Does not matter. The loopgain is the gain around the loop.
Excellent explanation prof. I only have one doubt in the "Bode plane" slide (min 13) in the phase plot I think that instead of -180 this has to be -90 degrees. For the rest i think you are the best❤
No.it is -180. The drop in slope is -40db/dec. This is a case of unstable system.
Very good explanation! Thanks for your interesting videos!
I'm sorry for that, but I guess there is a little mistake at 15:10 in the phase plot of Bode diagram; if it was intended to represent the same system of the Nyquist diagram, there could be an error near the phase margin point, that appears to be negative instead of positive. I mean that the phase is going down too fast in that plot, and that's not the situation represented by the Nyquist plot. To be coherent, the phase at 0dB should be about - 120°. Am I right?
Thanks for note. Yes, there is an error, already spotted by some viewers in the past.
Hi Professor, Thank you for making this video. It was an eye opener for me.
I need some clarity about on of the points shared in the video. By drawing a bode plot for a closed loop system, why are we plotting AB and not A/(1 + AB) which is nothing but ~1/B ? Effectively my questions is, why are we doing analysis on just loop gain (AB) and not the whole gain of the system (A/(1+AB))?
When AP>>! the gain is i/B. But AB determines when AB>>! and if the system is stable.
Seems the Bode Plot has a mistake. It shows the phase has past 180 degrees already, whereas it has only past 90 degrees in the complex plane.
Thanks for comment. Please indicate the minute in video.
I have confusion, @16:31: Nyquist plot is showing system is stable, but bode plot is showing system is unstable, whether my understanding is correct?
Hi Imtiaz, thank you for note, Yes, there is an error there already pointed out by other viewers. I am adding a not on that in the page of video.
Sir, you were saying that using rate of closure of A & (1/B) Plots, we can find the phase margin. During -20db rollover phase is -90deg, phi_m =90deg and at corner frequency phase is -135deg, phi_m =45deg. But sir, these phase margins at those points are corresponding to only plot A, but you were telling that rate of closure of two plots gives these phase margins, i feel you didn't add the phase of (1/B) plot to phase of plot A to find the overall system phase margin.
Please indicate the minute in video you are referring to.
@@sambenyaakov at 25:15
The -20db/dec is between the A and 1/B. This means that if you will plot it like in th-cam.com/video/L3MRnEYdi8g/w-d-xo.html the crossing will be at -20db/dec, becuse the different(which is BA) is -20db/dec.
Well explained.. Thanks
Thanks for comment
Really nice! Thank you.
Thank for comment and taking the time to write it.
Dear professor sam, Thanks for the great video. I am writing a report and love your practical way of designing a control loop. Do you have any litterature that i can refrence in my report :-)
i need some litterature for designing a PID controller based on the inverse plant response
I came up with this method many years ago, I have seen it somewhere, don't remember where :-). Have a look at this three part tutorial
th-cam.com/video/DfCuNjHDRHE/w-d-xo.html
th-cam.com/video/bGj1huz2Log/w-d-xo.html
th-cam.com/video/Bn7QPJaj-vU/w-d-xo.html
Great video!
Thanks Professor Yaakov for the great explanation. I have one question. When analyzing the current sense method whereby you measure the voltage across the inductor DCR, are you assuming a sinusoidal waveform? Otherwise the impedance values would not be correct. If you are, then you are using a first harmonic analysis?
Hi Chuck, sinusoidal waveform is
not assumed, The output is correct for any (within practical limitations)
frequency and combination thereof (a triangular ripple ). See
th-cam.com/video/sqQzBxfECj4/w-d-xo.html
Sir in nyquist criteria you were saying stability effected due to phase lag but while looking in to plot the encircling also mainly depends on radius(|LG|) of the plot. But you didn't speak about magnitude of |LG| .
Magnitute and phase will determine the cross point of the unity circle OR encircling of the -1point.
Thank you so much.
👍😀
Thanks Dr Sam you are wonderful , it is possible Dr sam to talk about ,how we design power factor inductor in SMPS and also choose the right TVS and MOV's.
Hi Tom, Have you seen
th-cam.com/video/6Mi8QDD71vE/w-d-xo.html
About thermals, good idea, will see.
thanks so much
very nice
please give some idea about z-transform for digital control with IQmath .
Planning to prepare a video on digital control.
@@sambenyaakov I am grateful,if you could show how to implement a practical approach of z-parameter to design smps like your previous feedback loop step by step design video.
hi, Prof Ben-Yaakov Thank you for your lecture! One question, given the same cross-over frequency, is higher phase margin always better? Assume 80 deg vs 70 deg PM. Both are stable for sure. Could you shed some light on this? Thank you!
Rise time to step will be shorter for 70 deg.
@@sambenyaakov Thank you. Do you mean a load step or a input step? For SMPS, I assume it is often times as a load step. For this reason, I think you're saying that given a load step, the 70deg PM will have less rise time than 80 deg. Correct? Thank a lot
Thanks again prof, helps so much to understand nuances that I glossed over back at University. Where do you teach?
www.ee.bgu.ac.il/~pel/
Sir,I have a doubt regarding bode plot method,usually we use bode plot or nyquist technique for linear systems right?but here we are determining stability for power modulator which is actually a non linear device,how it could be possible to find out the compensation using this bode plot?Please clear this doubt sir as I am working on one project and I need to publish one paper regarding non linearity of power electronic devices.Thank You
Hi, the classical method of switch mode systems control design (which are indeed non linear) is to use averaging to obtain a linear representation of the system, or to use average simulation models which I am showing.
Sir, I didn't understand what you were trying to say, were you saying that you are approximating the nonlinear device into the linear device?And I want to ask one more question.Are time domain techniques really useful to find out the stability of nonlinear devices or is it better to go with the state space techniques for nonlinear devices?
Google:
state space averaging
average simulation o PWM
sir does it mean we should never cross 0 db line?
You can't avoid it (gain will always drop eventually). The point is that at the cross over the phase margin should be OK
@@sambenyaakov sir can you tell more about control scheme
Thank you very much , there is something i don't understand in DC and in AC. In DC for exemple the compensation network do not have effect when V(-)=V(+) we have output of the amplifier should be 0V . But as to produce the PWM signal we compared it with SAWTOOTH. The result of this comparaison will give d.c~100% and that make no sense , because when V(-)=V(+) we should have Vout=D.C*Vout .
Good question. {V(-)=V(+)+epsilon} which is sufficient to produce an output for generating D
@@sambenyaakov thank you for your quick answer.
I understood my problem. Because at school we told us that Vout=D.c*Vin, but, and that make no sense when Errors is equal to zero because your duty cycle is zero. finally it same like we care about that relationship when we do not behave Fb. With FB, we focus on Vref and Vout hence it makes sense to have a duty cycle of zero when vref=vout and Vin is there only to give the sufficient needed power. Finally if i check D.c with a oscilloscope is never constant but permanent changes. This was my confusion
@@sambenyaakov
the Pin Vref(V+) of the EA is the desired output and the Vout mesured( V-) is input Voltage . Hence when the Vout mesured( V-) change , the dutycycle adjust in order to have the good Vref.
Vout=D.C*Vin it's finally used only for design of components once everything is turning , we can not observe directly since duty cycle is set, thank to vref of EA and vout.
Am i right?
Pretty good. IN proper designs (e.g. PI control) the gain of the error amplifier at DC is very large so it takes only a minute deviation from Vref to build up the voltage required as input to modulator . See th-cam.com/video/04gJ1t2cQdA/w-d-xo.html
Sorry Sam, but for this lecture was not good, I think you missed several times
Can you be specific?
very nice
😊