Hi Zach - I'm super impressed with MixedSim and its great to see some real examples of how it can be used. Please consider separate lessons on how to use the tools and how to interpret the results. For this model it would be very helpful if you were to release your schematic and Dashboard settings so that others can reproduce your results. I built my own simulation after your last video and my results were quite different to yours with the addition of bypass capacitance following the ferrite.
I was about to chime in that the ferrite in series is really just a low pass filter for very high frequencies...I wasn't thinking about it trying to act like the world's smallest boost regulator!
Oh man I had not thought about it like that, but you're right! That's almost what it's doing when that current pulse is being drawn into the input port.
I feel like the examples chosen here are not... optimal, at least for highlighting the pros of chip ferrite beads. None expects this specific ferrite bead to do anything at low frequencies, the impedance vs. frequency graph makes it obvious. And for the PLL switching spikes, wouldn't they be tamed with proper/extra decoupling capacitors? As a LC filter feeding a large power plane with low impedance? In RF transistor biasing, no amount of capacitors or their placement will prevent the external signals from entering the PDN, impedance in some form is needed. General guidelines suggest minimum 500 ohms at frequency of interest. But I do wonder if you would see much different results if we add the significant (real world) impedance to the main voltage source (is that how it works)? I am expecting we would be getting the spikes in both cases then, just less amplitude without the bead. Pretty informative video nonetheless, need to get more into modeling circuits. Also would be interested in a video or series on LDOs too, always wondered about their PSRR/low pass characteristics. I typically use LDOs for low frequency noise, coupled with beads for high frequency in RF designs. Newer LDOs have some wild filtering going on, like the TCR5RG series from Toshiba (claims to suppress upto 10 MHz).
Hey smallet, thanks for watching! Yes we want to get more into modeling circuits, this is all part of a lead up to that, and it would be good to be able to compare LDOs as part of a simulation like this. VRM models are more complex than the RL circuit shown in the video, so that would need to be included to get better results. For RF biasing yes you do need some actual impedance, such as if you need to dampen bias oscillations. The point isn't really to show the pros and cons of ferrites generally, just their application in attempting to deliver quiet power to different rails on a PDN. I think of the bead's impedance in this application like a double edged sword. You can block the interference from one rail at high frequencies but you might produce new noise on the rail at those same frequencies, depending on the signal bandwidth that is swithcing on the secondary rail. In the blog I actually compare the case with and without the bead just to illustrate. The conclusion you and I arrived at is, in this case, better cap selection could deal with BOTH frequency components on the secondary rail and would give more desirable results instead of adding the ferrite, which you can see from looking at the impedance curve with the bypass cap. If the 2nd rail ONLY needed DC, then you might target that with a large C to hit the lower-end noise and an inductor (or possibly a ferrite) and you'll hit all of those frequency components.
@@Zachariah-Peterson I meant I need to get more into modelling circuits, not that I would mind more of these simulation tutorials! Have always been too lazy to set up a proper simulation, so many factors (parasitics, etc) to consider.
Just as an aside, a lot of the high frequency rejection of LDOs comes from the passives in the test circuit. This Toshiba part looks fairly decent but they don’t even give you a PSRR graph. Basically the error amplifier bandwidth of all LDOs are somewhat low for stability reasons. You can usually tell from the shape of the PSRR graphs where the active rejection begins to roll off (op-amp loop gain) and the passive is providing the low impedance. The best part right now to my knowledge is LT3042/5. It’s about 20 dB better than the Toshiba at 1 MHz.
@@chrisbradley3224 TCR5RG part does have a PSRR graph, look for a later revision datasheet. ~60 dB rejection up to 3 MHz is good enough for most applications. The LTC part is nice but its price makes it non-contender ($0.3 vs $6). 200mA current is also very limiting. I think TI LDOs have PSRR graphs for varying Iout, they "derate" heavily with increasing current.
Hi Zach, Great walkthrough. Based on one application note(Ferrite Beads Demystified ) published by Analog Device, as the dc bias current increases, the core material begins to saturate, which significantly reduces the inductance of the ferrite bead. It looks like it is not easy to take this factor into simulation since the inductance is always the same for specific condition. How do you that effect?
I'm not aware of anyone who believes ferrites just filter everything and anything out. There is an obvious impedance curve that is frequency dependent. There are thousands of ferrites that all have different response curves for different frequency bands. You pick the ferrite to filter out the unwanted noise you are either trying to reject or prevent from corrupting other things on your power rails. When you have large dips in your rail like you are showing that is because your capacitance is not enough to hold up your rail. You failed to design your capacitance for your load and the response of the ferrite at your switching frequency. There is nothing wrong with ferrites if you understand basic filtering and sizing capacitors to loads.
I think the most common description I see in marketing copy is an ambiguous "they filter out noise in mid range frequencies" and that is it. Sometimes there is no description that the impedance curve switches to resistive in that range. I agree they can be useful but I follow the Barnes rule on this: avoid them unless you can prove you need them. The Sandler rule on this is: the best time to use ferrites is never!
