Excellent video as per FesZ standard. Clearly dissecting the problem from the definition all the way to possible solutions. Just I would like to add a small comment that the current spike during the turn-on event of the High Side MOSFET is not only due to the stored energy in the MOSFET's output capacitance, but highly probably due to Reverse Recovery mechanism of the Low Side MOSFET's body diode. I don't know, however, if you would touch on that in your second part on this topic. As a final remark, although ignoring this parasitic diode effect, you didn't lose generality on the guidelines for the RC snubber design which are spot on. Keep it up!
Another superb video - I hope in your " other life " you are an educator , you have an excellent way of presenting material - a rare gift 💪, as at university one is often on the calculus treadmill and dealing with abstract concepts, so it often takes companies at least 2 years to get interns to be productive.
Very well explained, thanks. I encountered this precise situation on a buck SMPS. The product was in its final version and the EMC testing revealed a terrible mountain of noise in the 100 MHz range. We found that a combination of schottky diode parasitic few nH inductance coupled with a capacitor was the source of ringing. Changing the diode reference only changed the ringing frequency. Finally, a RC snubber was the solution, without degrading the power supply performances. The cost was a PCB re-design 😭, lost time and the commercial team very angry 😡
Nice video as usual. One note (if it hasnt been mentioned in the comments yet): your comment regarding the contribution of the low side free wheeling diode was not correct. It is actually important to consider and can be a major contributor to ringing after the low to high transition. It is the reverse recovery charge that causes a very high current spike, depending on the implementation. The already conducting high side switch needs to discharge it, before the output even can start rising. The intrinsic MOSFET diode usually has a high Qrr. Higher voltage FETs tend to get even worse. It helps to add a Schottky, but only if it manages to take over most of the current (which is peak inductor current, to make things worse). I use to choose the low side MOSFET to have an as low Qrr as possible. Your ringing waveform looks like you do not have an extra Schootky, as ringing is quite asymmetrical.
Happy 200th!🥳🎉 Excuse this late comment, but i am always backlogged in viewing my subscribed channels. At any rate, yet another excellent video with a lot of new fascinating details i need to learn. However, i have a hard time understanding why the forward voltage of the diode would cause a negative voltage (5:20). Do you have any other video that dives deeper into this weird phenomenon?
Well, the diode is placed with anode to GND, and cathode to Switching node; when it is conducting, the cathode is at a voltage potential ~0.5V lower than the anode (exact voltage will be diode type and current dependent). As long as the GND (anode) is 0V, then the cathode (switching node) will be at a negative voltage - the exact value being the forward voltage drop.
@@FesZElectronics I feel like my electronic major has missed something fundamental here... Please let me dumb it down further... If we froze time at a moment the diode is conducting, and i use a multimeter, negative lead connected to GND and positive lead to the other side of the diode, would the multimeter show a negative voltage? If your answer is yes, then i REALLY need to sue my teacher.....
Lets consider a different example, you have a 10V supply, that drives a resistor trough a diode (diode has the anode towards the positive terminal of the supply, and cathode towards the resistor A -|>|- K); this way the diode is forward biased, and you get a current trough the resistor; to keep numbers simple, lets say the forward voltage is 1V, so the resistor has 9V on it; coming back to the diode, the anode is at 10V, the cathode at 9v, the the cathode is at a lower potential than the anode by 1v; Coming back to the smps circuit, if the diode is forward biased, and the anode is at 0V, the cathode needs to be at a negative voltage; This negative point can also be seen in the simulation at 6:23 ; for short periods when no switch is conducting, the switching node goes negative - that is when the diode is conducting.
😃😄😁😆😅 5 Smile's ever had that in a older relay makes a humming noise some flybacks will help in high frequency and current not in mine unfortunately mainly for that diodes reason😁. Helps fantastic if you want 4 voltage regulator working as the material adds a block to that 5v reg and needs pulling all working to a amp at 32v great for a dynamo ....
12:05 - Here looks like we are referring to parasitics... worth mentioning just to make traces shorter from Buck IC to MOSFET's drain... which will decrease parasitics inductance..
Indeed I am talking about the parasitic inductance, but formed between the input capacitor and the 2 switches (mosfet+mosfet or mosfet+diode); this was not linked to the driving signal - if that is what you mean by traces from the buck IC...
@@FesZElectronics We are tremendously fortunate to have you covering the range of topics that you do. If you need any encouragement, I too would eagerly watch current mode vs voltage mode.
At 4:48 you introduced a diode across SW2 to ensure current is always flowing through the inductor. How does the diode achieve this? Its anode is at ground so the only way for it to conduct is if Vs2 < -0.6V rigbt? Why is that node swinging below ground when it wasn't before (the only change is that you introduced the delay). Is it that with the delay, the left hand side of the inductor would be floating for a short period of time, changing its voltage to try and keep current flowing, and the diode now allows current to flow from ground, and into the inductor?
The inductor current in a buck converter will always flow from the switching node towards the output; when the high side switch is on, the inductor behaves like a load, getting charged with current, so the left side is at a higher voltage potential than the right side; when the high side switch is off, current still flows either trough the diode or the low side switch, an this time the inductor behaves like a battery - it is producing a voltage which is inverted to when it was behaving like a load - so the left side is at a lower voltage potential than the right; if the low side is an ideal switch, then that side of the inductor is at 0V; if its a real switch (with non-zero voltage drop) or a diode, the switching node goes negative - this is how the diode is being polarized.
In the video you mentioned it is not easy to measure the series resistance to determine the Q, that is not accurate. A simple 2-port impedance measurement can be done to determine the Q very easily. We do this measurement all quite often. We have published multiple papers and app notes on this topic.
Maybe you can share the names of some of the App notes or papers (but not links - youtube usually hides comments with links) Anyway, my point was that the resistance is not just trace resistance, its also in the switches and the decoupling capacitor; you will need to keep one switch on, the other off, and then the other way around to get a proper measurement, as I guess you already detailed in your work; it can be done, but its not as easy as measuring the stray capacitance and inductance :D The method for measuring L/C that I will try out in part 2 is to measure the ringing frequency with and without some known added capacitance.
@@FesZElectronics Sure, you can find multiple app notes about 2-port impedance measurements on the publication page of our website. Also I would steer you to the Signal Integrity Journal where lots of our peers have publications on measuring the Q using the 2-port shunt through impedance measurement.
TH-cam has millions of videos, but i don't see a single video for a complete practical (line powered) flyback SMPS design with explanation. Why no one has the courage to make one? How difficult it is? If anyone has such video link please share here, but from my knowledge no one makes a video about proper flyback converter😢
I'm concerned that much of the ringing (at 1:40) may be caused by your probing technique. The scope ground is not referenced to the measurement point (on the board), so there is a lot of inductance in series with the scope probing point. A local (on board) ground should show much less ringing, and even better results can be obtained with alternate scope probe connections (without the standard 6-inch scope ground lead).
Maybe it was not pointed out, but I am using the short ground connection - the probe tip spring ground thingy; its visible at 1:37 . I am not using an alligator clip or anything longer...
@@FesZElectronics - Oh, I see it now if I look carefully. Have you ever made a video about the importance of this detail of measurements? Since we both know of the importance, and since it is difficult to see (especially in a small-window video), you may want to consider mentioning it at times when it is important. Otherwise, people may continue to use the scope probe tip measurements and relying on scope-grounds back through power supply wiring. Many hours have been wasted troubleshooting scop-ground induced ringing that was not really there. Keep up the good work! Thanks, Carl
How would you propose doing it differently? As another viewer pointed out, it was not clearly highlighted in the video, but I did use a ground tip spring on the probe; do you suggest a better ground is needed?
The board its self is something I made a while back, and it has some intentional issues; I did things this way to be able to better highlight possible problems and solutions. It was never supposed to be perfect :D
If real life is different from the simulation, then the simulation doesn't reflect the real life circuit. FesZ takes steps to remedy that by adding in the parasitics explicitly to the simulation. The next video will show the proof in the pudding!
I'm less than two minutes in and I see someone probing a node with lots of high frequency content with no ground lead on the scope probe .... annnnnd I'm out. That probing technique is so astoundingly bad that I doubt the ability of this person to do anything correctly.
Nice video Fesz! Keen to see the practical measurements and testing! 👌
Excellent video as per FesZ standard. Clearly dissecting the problem from the definition all the way to possible solutions. Just I would like to add a small comment that the current spike during the turn-on event of the High Side MOSFET is not only due to the stored energy in the MOSFET's output capacitance, but highly probably due to Reverse Recovery mechanism of the Low Side MOSFET's body diode. I don't know, however, if you would touch on that in your second part on this topic. As a final remark, although ignoring this parasitic diode effect, you didn't lose generality on the guidelines for the RC snubber design which are spot on. Keep it up!
Another superb video - I hope in your " other life " you are an educator , you have an excellent way of presenting material - a rare gift 💪, as at university one is often on the calculus treadmill and dealing with abstract concepts, so it often takes companies at least 2 years to get interns to be productive.
Exactly my thinking when watching his videos. 👍 If he is a teacher in an university, his students can be proud to have him.
It seems to be your 200th video 🎉
I didn't even notice, but I guess you are right :D
High quality content. Really informative. Thank you very much.
I always like your videos. Technical concepts explained in clear manner.
Thank you very much for your excellent presentation. Much appreciated!
Very well explained, thanks.
I encountered this precise situation on a buck SMPS. The product was in its final version and the EMC testing revealed a terrible mountain of noise in the 100 MHz range. We found that a combination of schottky diode parasitic few nH inductance coupled with a capacitor was the source of ringing. Changing the diode reference only changed the ringing frequency. Finally, a RC snubber was the solution, without degrading the power supply performances. The cost was a PCB re-design 😭, lost time and the commercial team very angry 😡
Thank you, very informative. Appreciated.
Nice video as usual. One note (if it hasnt been mentioned in the comments yet): your comment regarding the contribution of the low side free wheeling diode was not correct. It is actually important to consider and can be a major contributor to ringing after the low to high transition. It is the reverse recovery charge that causes a very high current spike, depending on the implementation. The already conducting high side switch needs to discharge it, before the output even can start rising. The intrinsic MOSFET diode usually has a high Qrr. Higher voltage FETs tend to get even worse. It helps to add a Schottky, but only if it manages to take over most of the current (which is peak inductor current, to make things worse). I use to choose the low side MOSFET to have an as low Qrr as possible. Your ringing waveform looks like you do not have an extra Schootky, as ringing is quite asymmetrical.
Very good lighting
Super, multumesc pt video !
Pe mine ma intereseaza orice subiect din categoria SMPS, asa ca la mai mare 😊
Master class!
Happy 200th!🥳🎉 Excuse this late comment, but i am always backlogged in viewing my subscribed channels. At any rate, yet another excellent video with a lot of new fascinating details i need to learn.
However, i have a hard time understanding why the forward voltage of the diode would cause a negative voltage (5:20). Do you have any other video that dives deeper into this weird phenomenon?
Well, the diode is placed with anode to GND, and cathode to Switching node; when it is conducting, the cathode is at a voltage potential ~0.5V lower than the anode (exact voltage will be diode type and current dependent). As long as the GND (anode) is 0V, then the cathode (switching node) will be at a negative voltage - the exact value being the forward voltage drop.
@@FesZElectronics I feel like my electronic major has missed something fundamental here... Please let me dumb it down further...
If we froze time at a moment the diode is conducting, and i use a multimeter, negative lead connected to GND and positive lead to the other side of the diode, would the multimeter show a negative voltage?
If your answer is yes, then i REALLY need to sue my teacher.....
Lets consider a different example, you have a 10V supply, that drives a resistor trough a diode (diode has the anode towards the positive terminal of the supply, and cathode towards the resistor A -|>|- K); this way the diode is forward biased, and you get a current trough the resistor; to keep numbers simple, lets say the forward voltage is 1V, so the resistor has 9V on it; coming back to the diode, the anode is at 10V, the cathode at 9v, the the cathode is at a lower potential than the anode by 1v; Coming back to the smps circuit, if the diode is forward biased, and the anode is at 0V, the cathode needs to be at a negative voltage; This negative point can also be seen in the simulation at 6:23 ; for short periods when no switch is conducting, the switching node goes negative - that is when the diode is conducting.
I think im starting to get it, but now my head hurts 😅
😃😄😁😆😅 5 Smile's ever had that in a older relay makes a humming noise some flybacks will help in high frequency and current not in mine unfortunately mainly for that diodes reason😁. Helps fantastic if you want 4 voltage regulator working as the material adds a block to that 5v reg and needs pulling all working to a amp at 32v great for a dynamo ....
12:05 - Here looks like we are referring to parasitics... worth mentioning just to make traces shorter from Buck IC to MOSFET's drain... which will decrease parasitics inductance..
Right, at 12:05 he's discussing the parasitics that he explicitly drew on the diagram at 9:30.
@@Graham_Widemanyep, but would be helpful to mention to make high-current traces shorter
Indeed I am talking about the parasitic inductance, but formed between the input capacitor and the 2 switches (mosfet+mosfet or mosfet+diode); this was not linked to the driving signal - if that is what you mean by traces from the buck IC...
@@SignalProduct-be2js I'm sure that will be one of the topics of the second part of this series of videos.
Could you please make video on current mode buck converter and it's advantages/Disadvantages and how it's better than voltage mode buck converter
I will try to cover the control strategies - current mode and voltage mode at some point; not sure when though
@@FesZElectronics We are tremendously fortunate to have you covering the range of topics that you do. If you need any encouragement, I too would eagerly watch current mode vs voltage mode.
At 4:48 you introduced a diode across SW2 to ensure current is always flowing through the inductor. How does the diode achieve this? Its anode is at ground so the only way for it to conduct is if Vs2 < -0.6V rigbt? Why is that node swinging below ground when it wasn't before (the only change is that you introduced the delay).
Is it that with the delay, the left hand side of the inductor would be floating for a short period of time, changing its voltage to try and keep current flowing, and the diode now allows current to flow from ground, and into the inductor?
The inductor current in a buck converter will always flow from the switching node towards the output; when the high side switch is on, the inductor behaves like a load, getting charged with current, so the left side is at a higher voltage potential than the right side; when the high side switch is off, current still flows either trough the diode or the low side switch, an this time the inductor behaves like a battery - it is producing a voltage which is inverted to when it was behaving like a load - so the left side is at a lower voltage potential than the right; if the low side is an ideal switch, then that side of the inductor is at 0V; if its a real switch (with non-zero voltage drop) or a diode, the switching node goes negative - this is how the diode is being polarized.
In the video you mentioned it is not easy to measure the series resistance to determine the Q, that is not accurate. A simple 2-port impedance measurement can be done to determine the Q very easily. We do this measurement all quite often. We have published multiple papers and app notes on this topic.
Maybe you can share the names of some of the App notes or papers (but not links - youtube usually hides comments with links)
Anyway, my point was that the resistance is not just trace resistance, its also in the switches and the decoupling capacitor; you will need to keep one switch on, the other off, and then the other way around to get a proper measurement, as I guess you already detailed in your work; it can be done, but its not as easy as measuring the stray capacitance and inductance :D
The method for measuring L/C that I will try out in part 2 is to measure the ringing frequency with and without some known added capacitance.
@@FesZElectronics Sure, you can find multiple app notes about 2-port impedance measurements on the publication page of our website. Also I would steer you to the Signal Integrity Journal where lots of our peers have publications on measuring the Q using the 2-port shunt through impedance measurement.
I was literally just searching for this. How did you know!!??
I have a 6th sense :D I hope you found it helpful!
@@FesZElectronics You are making a joke, but it's uncanny how you do this! Hahaha!
TH-cam has millions of videos, but i don't see a single video for a complete practical (line powered) flyback SMPS design with explanation.
Why no one has the courage to make one? How difficult it is? If anyone has such video link please share here, but from my knowledge no one makes a video about proper flyback converter😢
Three thumbs up for this topic. Particularly the magnetics.
I'm concerned that much of the ringing (at 1:40) may be caused by your probing technique. The scope ground is not referenced to the measurement point (on the board), so there is a lot of inductance in series with the scope probing point. A local (on board) ground should show much less ringing, and even better results can be obtained with alternate scope probe connections (without the standard 6-inch scope ground lead).
Maybe it was not pointed out, but I am using the short ground connection - the probe tip spring ground thingy; its visible at 1:37 . I am not using an alligator clip or anything longer...
@@FesZElectronics - Oh, I see it now if I look carefully. Have you ever made a video about the importance of this detail of measurements? Since we both know of the importance, and since it is difficult to see (especially in a small-window video), you may want to consider mentioning it at times when it is important. Otherwise, people may continue to use the scope probe tip measurements and relying on scope-grounds back through power supply wiring. Many hours have been wasted troubleshooting scop-ground induced ringing that was not really there.
Keep up the good work!
Thanks,
Carl
I will try to point it out more clearly in the future, thanks for the suggestion!
Your probe is not properly grounded during the measurement.
How would you propose doing it differently? As another viewer pointed out, it was not clearly highlighted in the video, but I did use a ground tip spring on the probe; do you suggest a better ground is needed?
@@FesZElectronics could be the circuit board design issue, low cost PSU usually uses 2 layer PCB
The board its self is something I made a while back, and it has some intentional issues; I did things this way to be able to better highlight possible problems and solutions. It was never supposed to be perfect :D
“Simulation”……things tend to change in real-life circuits ….,lol
If real life is different from the simulation, then the simulation doesn't reflect the real life circuit. FesZ takes steps to remedy that by adding in the parasitics explicitly to the simulation. The next video will show the proof in the pudding!
I'm less than two minutes in and I see someone probing a node with lots of high frequency content with no ground lead on the scope probe .... annnnnd I'm out. That probing technique is so astoundingly bad that I doubt the ability of this person to do anything correctly.
:) look a bit closer, you might see the probe tip ground spring