Thanks professor. Your slides and explanations are truly unique, a true reminder that mastering the fundamentals can make any complex problem easy to understand and tackle.
A fine presentation that showed that the DC power supply design can become challengng when the interaction with a "load" like a HIGH current switching circuit is involved.
Thanks again for another video. I often get confused by the termonology. Stumbled upon a decoupling capacitor while looking at a gate driver. So this is good timeing.
Another excellent lecture - many thanks. I have often observed these kind of behaviours in practice, but never got around to investigating so thoroughly - rather lazily I just tend to experiment with the capacitor values until the problems go away.
Hello professor, Thank you for the excellent lecture. Could you please make videos on active inrush limiters using p type MOSFETs. I believe they have good application in the presence of bulk capacitors in SMPS. Thank you!
The biggest problem with ferrite beads is that they rarely give any information in non-linear inductive/resistive characteristics with change in DC current. Fact is, the rated current for ferrites says nothing of this nonlinear effect-it speaks only to it’s DC current handling current from a thermal perspective. In fact, the impedance collapses at extremely low bias currents for most ferrites. For 0402 ferrites for instance a 100 ohm ferrite might have a 800mA rating, but it’s peak impedance drops by 30% with 20mA, and 99% at the rated current. Their Q also seems to rise with DC bias, so they become even more prone to oscillating and impedance spikes
Earlier in my career I was guilty of the errors you discussed here-you really have to understand what the effects are of the bead on the source impedance seen by the load. Sprinkling these in a design willy-nilly, is the hallmark of the naive. You absolutely want to avoid peeky high-Q parallel resonant spikes in that impedance, or any rapid load changes will result in ringing. I think it would be illustrative to show the source impedance of the load in the different cases, and how that translates to the dampened oscillations. You especially want to avoid those sorts of spikes at frequencies that are likely to be excited by the system. The way to avoid those spikes is to ensure that you have much more capacitance on the load side than you would otherwise. Unfortunately, this limits the applications where you should apply ferrites substantially, since it usually adds unnecessary cost and complexity. It doubly hurts that the benefits are limited by the bead saturation effects, which makes the filter’s impedance highly load dependent, and whose effects are poorly described in datasheets. Beads can be quite effective when designed well, but the level of engineering is much higher to achieve this than it would seem at first glance.
Great presentation. Here is a request from professor Ben-Yaakov: I like to learn about common mode chocks and their applications in Power Circuits. I know their use in Antenna and RF Circuits. However, I also seen them used in Power Management Circuits, and I would like to learn more about them.
Decoupling is very frequency dependent. You mention that Zl is low-that is true at 1MHz for most electronics, but at 1 GHz, it’s actually substantial due to resonances in the power planes/traces. So if you are feeding a 1GHz processor for instance, that close decoupling capacitor is important for the function of the processor (lowering the source impedance at the clock frequency) and EMI (somewhat decoupling the supply). Putting a more caps works to improve overall impedance, but the most improvement comes from using a physically smaller one, one that you can place closer to the load. You want to use the largest capacity you can afford in the package size that you use.
@@sambenyaakov I’m not sure I understand your reasoning. At the beginning of the video, you mention decoupling can’t effectively happen if the source impedance is lower than the capacitive reactance. The only way to increase that source impedance in a lossless way is with inductance. Mostly this inductance comes for free, in the form of a trace or cable. At some frequencies the cable/trace/plane will be self-resonant, and you’ll get all the problems noted with ringing. My point above was just that the source impedance becomes non-negligible at higher frequencies, which is why a decoupling/bypass capacitor is required.
My first inclination would have been to get a bead with a lower permeability resonant frequency ands perhaps move it in-between the two capacitors. However that could just make the resonance lower in frequency. It seems that by changing the smaller capacitor the ferrite bead is now just acting as the series resistor, and not really using the rlc portion of the model. I thought it would be interesting to just replace the bead with a 290 mOhm resistor, so I did it and there was very little difference.
@@d614gakadoug9 It is not well written. The resonant frequency of the permeability is the frequency above which permeability becomes imaginary and the ferrite becomes lossy. Thus a ferrite with a lower permeability resonance would be lossy at a lower frequency.
Sir I am facing an issue in my converter hardware, with the inductor and switch voltages. I would like to get suggestions over that from you if possible. Can you share me with the contact details?
@@biswajit681 I appreciate your interest. Let me tell you the secret behind. Most of my videos are an outcome of my current on going design work. So, as I move forward, going back is not simple. But then, the videos are always "fresh"
Steve Sandler of Picotest once told on a conference at R&S in Munich "Avoid ferrite beads like the plague!" There was no opportunity for me to ask why. Now I understand what he meant.
