As always, a nice didactic explanation of a complex problem. Great work, Thank you Sam ! Looking forward for the part 2, a passive damping is so expensive that I look forward to see better solutions :)
somewhat related to the video: Its intuitive that if resistance makes lc tank attenuate over time that negative resistance will do the opposite. The ZVS driver is a prime real world example of negative resistance -you have a main LC tank and a pair of mosfets set up so that the current they supply at 2 points in the circuit is 180 deg away from the voltage on the same points. Seeing inverters trough the lens of negative resistance in resonant tanks is so mindblowing, yet makes it so easy to understand.
Very educational! I am looking forward to finding out in your next video whether the methods that are commonly provided for stabilizing these types of systems are well founded!
Hey professor, I have another little question: The conversion at 10:36, when the approximation is valid? I know this approximation valid some time in fourier plane(near resonant frequency) but for using fourier plane, we have to assume stability, can we use this approximation in laplace plane for stability proof? Thank you.
There are approximation in this transformation. The main one assuming that the resonant network are the generic series or parallel. I think it holds for high quality factors.
Hey professor, thank you for the video, I have a little question, At 14:43, why does Ri=P/VBUS (operation VBUS), at 4:51: Ri= P/Ii^2 ->Ri=P/(P^2/Vi^2)-> Ri= Vi/P, and I assume Vi= VBUS = V operation point, thank you!
Thanks. Indeed Ri=P/VBUS is a typo. But then, Ri= Vi/P is wrong too. 1. missing minus sign. 2. wrong dimension V/P is [1/A] not [Ohm]. So I guess everyone can err😊
Very timely and thank you. I have this exact vexing problem with a power inverter that randomly and under certain conditions will go into oscillation and blow the Mosfets in a twinkling of an eye. Its not possible to replicate this reliably to test , and of course, once it blows, there is nothing to test. I was thinking resonant phase changes and tried lead/lag filtering without success, I think you may just have the answer.
Device is a classic push/push inverter, using 24 Mosfets , 6 center tapped 38kHz transformers. Input , 48vDC 2200mFd x 12. These run HOT despite low ESR . Mosfets are IRFZ44N, not the choice I would make here, . PWM Gate drive low period oscillates around 600 mV above ground despite active driver pulldown, 10 ohm series R. Quiescent current is high. I suspected Dv/Dt issues , lowering freq makes no change. Failure occurrs randomly when un or slightly loaded. Frustratingly, an identical unit functions perfectly.
@@sambenyaakov If I take the input section as a series circuit, L/C and parasitic R, we have the same conditions as you describe. Where the current is injected should matter little, or the direction of current, in my case this can be sourced or sunk via the transformer over parts of , or even over a complete cycle.... therefore the conditions are the same for negative resistance to occurr as you describe. Yes, looking forward to part 2. Unfortunately in the repair industry, time constraints usually prevent any real analysis of these annoying problems.
Hello Professor, I always like your content which is very much informative and easy to understand. However, at 6:13mins the output voltage transfer function i.e. Vo/Vin should not have sL in the numerator, this is slightly confusing. Apart from this, the final denominator expression should be LCRs2 + Ls +R in my opinion changing the quadratic solving. Please enlighten on these fronts. Thanks for your good work.
Hi, thank you very much for taking the time to examine the derivations in the video. You are of course correct on both. On the first point, I already added a comment on that at the TH-cam description page. I was not aware of the second error you pointed out. As you wrote the general final result is the same. But I will also make a correction note or even a video.
Thanks for the very interesting video! However, I don't understand something: If the condition of stability is to have Rp higher than the input resistance of the converter (as shown in 8:22), then from the expression of Rp in 10:30, we can understand that our goal is to have low Rc. But it doesn't make sense because if Rc is zero we know that it is not stable. What am I missing?
@@sambenyaakov Thank you!! Now I see that it is my mistake.. Rp needs to be lower than Ri :) Another question: I assume that Ri depends on the frequency, and the expression for Ri in 4:22 is the DC value. Am I correct?
Thank you! I wonder if this can be explained with the regular stability analysis methods like phase and gain margin, or bode. and to see how it related to the frequency.
You certainly can. But once you do that you would probably like to plot the phasor equations, to see what is going on, which I am getting without the hard work😊
Thanks professor! Instead of a passive filter, if one more converter is cascaded and operated in current mode while the original converter in voltage mode, would the instability can still occur?
Thanks for sharing. Interesting approach to this phenomenon. Do you think that having a DC source at the input and pulsating power (current source) at the output would give the same result? Where the power changes from stable to unstable level (with 10mW crossover). Voltage perturbation on an inductor should resemble current perturbation on a capacitor in a resonant circuit…
This is a great topic. I just came across it in this video: th-cam.com/video/c1QSn3i_IEo/w-d-xo.html following a product review of a battery powered AC mains generator. Some people in the video's comments claim that this is considered normal behaviour for the product in the particular usage shown. I am of the opposite persuasion but I think that whether this can be considered normal behaviour or not is more of a legal question rather than a scientific one. Nevertheless, there is an interesting question arising from an electronics perspective: can a battery supplied AC mains generator be designed that is stable under heavy inductive loads using a cheap practical circuit? Could this be worthy of a dedicated video?
As always, a nice didactic explanation of a complex problem. Great work, Thank you Sam ! Looking forward for the part 2, a passive damping is so expensive that I look forward to see better solutions :)
🙏😊
somewhat related to the video:
Its intuitive that if resistance makes lc tank attenuate over time that negative resistance will do the opposite. The ZVS driver is a prime real world example of negative resistance -you have a main LC tank and a pair of mosfets set up so that the current they supply at 2 points in the circuit is 180 deg away from the voltage on the same points.
Seeing inverters trough the lens of negative resistance in resonant tanks is so mindblowing, yet makes it so easy to understand.
Interesting. Thanks for sharing.
Very educational! I am looking forward to finding out in your next video whether the methods that are commonly provided for stabilizing these types of systems are well founded!
🙏😊
Very informative. Always wanting more of these videos 👍
🙏🙂
Hey professor, I have another little question:
The conversion at 10:36, when the approximation is valid? I know this approximation valid some time in fourier plane(near resonant frequency)
but for using fourier plane, we have to assume stability, can we use this approximation in laplace plane for stability proof?
Thank you.
There are approximation in this transformation. The main one assuming that the resonant network are the generic series or parallel. I think it holds for high quality factors.
Excellent tutorial!
🙏🙂
Outstanding Job.
Thanks🙂
Hey professor, thank you for the video, I have a little question,
At 14:43, why does Ri=P/VBUS (operation VBUS), at 4:51: Ri= P/Ii^2 ->Ri=P/(P^2/Vi^2)-> Ri= Vi/P,
and I assume Vi= VBUS = V operation point,
thank you!
Thanks. Indeed Ri=P/VBUS is a typo. But then, Ri= Vi/P is wrong too. 1. missing minus sign. 2. wrong dimension V/P is [1/A] not [Ohm]. So I guess everyone can err😊
@@sambenyaakov I meant (Ri= Vi^2 /P), I wrote Ri = P/(P^2/Vi^2). Is Ri=- Vi^2 /P the right expression?
Very timely and thank you. I have this exact vexing problem with a power inverter that randomly and under certain conditions will go into oscillation and blow the Mosfets in a twinkling of an eye. Its not possible to replicate this reliably to test , and of course, once it blows, there is nothing to test. I was thinking resonant phase changes and tried lead/lag filtering without success, I think you may just have the answer.
Hold on to part 2 . Can you share the info about the type a d number of caps you use in the said system?
Device is a classic push/push inverter, using 24 Mosfets , 6 center tapped 38kHz transformers. Input , 48vDC 2200mFd x 12. These run HOT despite low ESR . Mosfets are IRFZ44N, not the choice I would make here, . PWM Gate drive low period oscillates around 600 mV above ground despite active driver pulldown, 10 ohm series R. Quiescent current is high. I suspected Dv/Dt issues , lowering freq makes no change. Failure occurrs randomly when un or slightly loaded. Frustratingly, an identical unit functions perfectly.
@@sambenyaakov If I take the input section as a series circuit, L/C and parasitic R, we have the same conditions as you describe. Where the current is injected should matter little, or the direction of current, in my case this can be sourced or sunk via the transformer over parts of , or even over a complete cycle.... therefore the conditions are the same for negative resistance to occurr as you describe. Yes, looking forward to part 2. Unfortunately in the repair industry, time constraints usually prevent any real analysis of these annoying problems.
Hello Professor, I always like your content which is very much informative and easy to understand. However, at 6:13mins the output voltage transfer function i.e. Vo/Vin should not have sL in the numerator, this is slightly confusing. Apart from this, the final denominator expression should be LCRs2 + Ls +R in my opinion changing the quadratic solving. Please enlighten on these fronts. Thanks for your good work.
Hi, thank you very much for taking the time to examine the derivations in the video. You are of course correct on both. On the first point, I already added a comment on that at the TH-cam description page. I was not aware of the second error you pointed out. As you wrote the general final result is the same. But I will also make a correction note or even a video.
Thanks again th-cam.com/video/1p7v8qYRT70/w-d-xo.html
Thanks again Professor 🙏
Thanks for the very interesting video!
However, I don't understand something:
If the condition of stability is to have Rp higher than the input resistance of the converter (as shown in 8:22), then from the expression of Rp in 10:30, we can understand that our goal is to have low Rc. But it doesn't make sense because if Rc is zero we know that it is not stable. What am I missing?
Hi, I think you got it all backward. We want low Rp so the equivalent resistance will be positive (smaller than Ri) , so RC needs to be larger.
@@sambenyaakov Thank you!! Now I see that it is my mistake.. Rp needs to be lower than Ri :)
Another question: I assume that Ri depends on the frequency, and the expression for Ri in 4:22 is the DC value. Am I correct?
Thank you! I wonder if this can be explained with the regular stability analysis methods like phase and gain margin, or bode. and to see how it related to the frequency.
You certainly can. But once you do that you would probably like to plot the phasor equations, to see what is going on, which I am getting without the hard work😊
Thanks professor!
Instead of a passive filter, if one more converter is cascaded and operated in current mode while the original converter in voltage mode, would the instability can still occur?
The instability is at the connection of the converter and an LC filter needed for EMI suppression.
@@sambenyaakov Yes, understood. Making a similar video for instability & oscillations in cascaded converters would be greatly useful. Thanks!
Good suggestion Will see. Thanks
Sir you are a Legend.
😊👍🙏
Thanks for sharing. Interesting approach to this phenomenon.
Do you think that having a DC source at the input and pulsating power (current source) at the output would give the same result? Where the power changes from stable to unstable level (with 10mW crossover).
Voltage perturbation on an inductor should resemble current perturbation on a capacitor in a resonant circuit…
I have no experience in this. Perturbation may perhps be a way to go. Thnks for sharing your thoughts.
At 7:40, the s^2 coefficient should be RLC.
As pointed out in the description section of video, there is a correction th-cam.com/video/1p7v8qYRT70/w-d-xo.html
👍💪🙏
👍🙏🙂
This is a great topic.
I just came across it in this video:
th-cam.com/video/c1QSn3i_IEo/w-d-xo.html
following a product review of a battery powered AC mains generator.
Some people in the video's comments claim that this is considered normal behaviour for the product in the particular usage shown.
I am of the opposite persuasion but I think that whether this can be considered normal behaviour or not is more of a legal question rather than a scientific one.
Nevertheless, there is an interesting question arising from an electronics perspective: can a battery supplied AC mains generator be designed that is stable under heavy inductive loads using a cheap practical circuit? Could this be worthy of a dedicated video?
Thanks for sharing your thoughts. To accommodate heavy reactive loads one needs to increase the current capability for same real power.