Dissecting TI’s Active EMI Filter (AEF)

Relevant videos:
An intuitive explanation of various active filters for ripple suppression.
• Active Ripple Filters ...

[@n6mz]

24:50 someone pronounces "giga" correctly!!!!!!! Your outstanding analysis makes me smile also. THANK YOU, Professor.

[@sambenyaakov]

😊Yes, this is how it is pronounced in venues that count such as in the film "Back to the Future".

[@k7iq]

😂😁😍😎

[@akosbuzogany2752]

There might be just a handful of people on this planet to challenge a TI design! Chapeau!

[@sambenyaakov]

😊Thanks. Believe it or not, TI offices and labs are populated by humans which are bound to err from time to time. The sad point is not that people do not challenge TI but that many do not take the time to check for errors.

[@robson6285]

Unbelievable interesting! this circuit would never been understandable enough without this great and superinteresting explaining. I find this such an inspiring circuit now i see the way this circuit can work. Great teaching and around this great new circuitidea!(at least for me this was such an interesting and new circuitway, this whole videolesson i find really inspiring! Thank you so

[@sambenyaakov]

Thanks Rob for comprehensive over view.

[@arthurswanson3285]

As always, excellent tutorial, Professor. I feel like I owe Ben-Gurion University tuition payments from all the knowledge I get here! :)

[@sambenyaakov]

Wow, thanks!

[@1102gourav]

Thank you Professor for the lecture on Active EMI filter from TI. Your lectures always means a lot and is building my power electronics knowledge. Professor, one point to add on the value of Cinj. The Cinj is connected between phase and ground. Hence, at low frequency (50/60Hz) its value should be of sufficiently low, so that the leakage current value is small to meet the regulatory standards (high leakage current causes nuisance RCD tripping and may become lethal to livestock). If it make sense, please let me know. Thank you once again sir.

[@sambenyaakov]

Thanks for comment. I am not sure if follow you. We are talking about DC circuits. What do you mean by "phase to ground"? the power line?

[@th600mike3]

This is awesome! Thank you for making this video!

[@sambenyaakov]

Thanks

[@user-hi2to8ln5w]

Good presentation! Thank you!

[@sambenyaakov]

🙏👍😊

[@trevthea5781]

Always brilliant

[@sambenyaakov]

😊🙏

[@k7iq]

Awesome as usual ! :) I remember once that LTspice would not converge... I added a resistor with ONE leg connected to a portion of circuit and it ran ! I think they fixed that particular type of convergence error not long after that.

[@sambenyaakov]

Thanks

[@tseckwr3783]

Professor Sam ... Thanks for the latest video.

[@sambenyaakov]

😊🙏

[@popuassmf]

My avation! Nice explanation step-by-step.

[@sambenyaakov]

🙏😊

[@mohammadhassanzade6893]

Thank you Mr professor 🙏

[@sambenyaakov]

You are very welcome

[@Metalloys]

Thank you professor.

[@sambenyaakov]

Thanks

[@falconhawker]

I remember 60 hz power supply design was necessary and B O R I N G . Well times have changed ! Switching power supplies delivering 10Amp DC , with 20mvpp ripple , operating at 100 of khz, requiring sophisticated and difficult ( for me ) analysis.

[@sambenyaakov]

Indeed. Thanks for comment.

[@mauriziogarino8353]

Why not simply using a ceramic capacitor as a first stage of the pi filter? What about the opamp output stage current amplitude, Is It a push pull needed? What's the main advantage vs full passive filter?

[@sambenyaakov]

"Why not simply using a ceramic capacitor as a first stage of the pi filter? " all that is discussed is AFTER the cap. Or, do you think that just a cap is a good solution to the EMI problem? As for: " What about the opamp output stage current amplitude, Is It a push pull needed? What's the main advantage vs full passive filter" you correctly defined the home work.

[@peddadasaiganesh4651]

It worked on me like a magic professor. I am currently working on the stability of the filter. Do you suggest any reading materials to analyze the stability of such filters and also the selection of component values?

[@sambenyaakov]

look up IEEExplore, there are some papers on active filters.

[@van_augur725]

Hi, I have one question. Doesn't the feedback loop have to be broken up when simulating the loop gain and doing a stability analysis of an op amp circuit?

[@sambenyaakov]

No. Very difficult to keep the operating point when opening the. The method shown is the recommended.

[@harrydellamano1025]

Hi Sam, great job but one question: Your BODE box can only supply 50 MHz, how do you explain 1 Ghz display?

[@sambenyaakov]

Hi Harry, Thanks for the comment. You are correct but the Bode 100 is illusrative regarding the method of measurement while the plots, as described, are simulation results. As I understand, Omicron is working on a new analyzer with a wider BW, so this apparent incompetability will be resolved.

[@alirezatanehkar2009]

Very sweety analysis!! What is the max power that can be used for active filter?

[@sambenyaakov]

Probably few tens of Watts.

[@klauskragelund8883]

Great presentation. One small error if I am not mistaken: when talking about the bode measurement, the resistor is added to the compensator/opamp output. You would like to have low impedance in that point (10mOhm), but adding a resistor/inductance of the injection transformer will impede that function. You will probably measure something, but not the correct function. You could add a 10mohm resistor, but that is impractical and the bode 100 wont be able to supply the needed source voltage. It's better to find another injection point.

[@sambenyaakov]

Hi Klaus, in principle you are correct. But.. since the feedback is taken after the resistor , the reflected resistance is much much smaller.

[@klauskragelund8883]

@@sambenyaakov Yes, to a degree. The compensator path needs however to have not just low reflected resistance, but also low inductance. The AEF will not function well if you have 50nH+ inductance in the compensator output, since the phase will roll of as compared to the 3ohm resistor. Could be interesting to do a simulation with trace and component parasitics

[@sambenyaakov]

@@klauskragelund8883 Where di you get 50H+?

[@klauskragelund8883]

@@sambenyaakov 50nH, from 50mm total traces from the device to the injection point, via traces and components

[@klauskragelund8883]

Just a comment about your videos, they are simply great and deep content of high quality. If I only had had a teacher like that in school.....

[@ayushsinghal6557]

Hello prof, I have a circuit. A supply voltage connected to the NMOS and then to the capacitor (1000uf).I want to charge the capacitor with 1A current in around 200ms. SO for this, I have to use NMOS in a linear region and it will be closed switch. But my question is how to control this NMOS so it can work in linear region and provide 1A current for particular time....could you give some suggestions?

[@sambenyaakov]

Perhaps this will be of help?

[@ayushsinghal6557]

@@sambenyaakov sorry i didn't get?

[@tamaseduard5145]

🙏👍🙏

[@sambenyaakov]

😊🙏

[@GadgetChannel-ErCanEverything]

👍👍👍🤝🤝🤝

[@sambenyaakov]

🙏😊

[@ats89117]

Great video relevant to a very large number of products, but the circuit isn't very useful if you can't get an amplifier with a high enough gain-bandwidth product to make it work. Or as the saying goes: Further research is necessary...

[@sambenyaakov]

Thanks for comment.

[@veeraa8695]

It will help for RE or RI Or both