I think the third example can omit the magnetic beads. Because the signal transmission uses an isolation transformer, a good isolation effect can be obtained, and the DC supply line uses inductance isolation, which is better than magnetic beads. Since the above two circuits are well isolated, the interference loop of the ground wire is minimized, so the magnetic beads of the ground wire can be omitted.
The third web design could be adopted without hesitation because the signal lines are galvanically isolated placing a bead in the ground path will not impact signal integrity and other considerations may justify this level of isolation between the grounds.
1. Beads are used to create different power domains. It should be used with caution. You should avoid high frequency current going through the beads. 2. The third example is ok. There is a galvanic isolation on high frequency signal path.
I agree, 3rd. The only reason I can see for adding a bead in the ground path would be to reduce power supply HF switching noise in the main circuit or, to reduce noise from the main circuit being introduced in a different circuit that shares the same power source. This would have to be a fairly high frequency switcher in the first case.
In case 2 you can adopt no problem. all circuits close trough a bead anyways. It provides more filtration during low freqencies of switching on different channels and it gets overwhelmed and looses some filtration powers the faster and more channels switch, allowing the receiver to detect highs and lows faster. In case 3 the bead is just another DC filtering component
Is ferrite on ground a good idea? I'd say it can be a well-advised evil, to break ground loops. In example 1, i imagine say an audio output device. Several such devices will be connected to one amplifier or mixer, and some of these devices will be Earthed or share a common power supply, every cable has a ground connection, so it happens, that the enduser gladly builds ground loops, and there's nothing the manufacturer can do to dissuade people from doing so and defeating carefully laid star grounds. Ground current on the analogue output device can be murderous, as to output a particular voltage, it will need to match the existing ground current, and if it's high frequency noise current, it will probably slam right against the slew rate limit. An audio CODEC or such chip can have two sections, the digital section and an analogue section with each their separate grounds. If the recommendation to add a bead comes from the chip designer themselves, they would have foreseen a communication within the IC between these two sections that is resistant to relative voltage noise in ground between the sections, so would be safe to follow. In example 2, this seems to make no sense at all. It's like someone is building a signal integrity murdering contraption. I am going to assume that someone is trying to solve an actual signal integrity issue that they're having there, but the suggested fix looks like it'll only make it worse in so many regards, only the bead on VCC sure won't hurt. For sure ESP8266 being a WiFi SoC, is prone to induce a lot of EMI, but if it's causing issues, either someone has built a ground loop where they shouldn't have, and/or they're talking to ESP8266 via TTL-RS232, which i surmise is the case from TX RX pin designation, and they're having trouble matching the BAUD rates of both devices, they're probably hitting it kinda close but not quite, so it seems to work until it doesn't, they should look at tolerance rates of different BAUD rates for both of the processors, they should really look at the clock circuits they got there, the compensation capacitance of the crystals, and consider using SPI for communication instead, it's not sensitive to clock disparity. Accurate TTL-RS232 is hard and not very tolerant of part quality. In example 3, i don't understand some of it? In the upper half, where they're transmitting a differential signal and coupling it through a transformer, like say Ethernet, this seems good, they can put ground beads between these two devices with no harm. But what is happening with the single-ended transmission lower in the pic where they're sending the signal through the inductor? I don't understand. If that's just power or low-frequency signal, then the bead in the ground sure is fine.
1st example. It depend, without wider scope of problem is hard to say if this bad idea or good idea, 2nd example with UART communication is bad. Ferrite bead should go from power supply to supplied system, not between modules. UART use high frequency, then we have voltage drop on ferrite bead... 3th example look good, for me it look like attenuation for LC filter (high R for high frequency, dump gain in resonant frequency)
I general, as far as I know, placing beads on ground path is intended to attenuate common mode noise [it is separated into two cases: 1) Earthened system PE cable bead placement or 2) not earthened and with capacitive earth coupling - in such a case both gnd and Vin should be filtered since one can not impact PE path capacitive coupling] Looks like 3 is ok since there should be no high dI/dt currents between separated circuits (should be supported by local decoupling capacitors). . #2 is problematic, I would speculate that there will be significant impact on signals risetimes (since those looks like single-ended signals). With low baudrates it will be less of an issue than with higher. Regarding #1: I don't see why one would place bead between analog and regular ground since anyways if analog ground is islanded from the rest of ground - no current will go through to cause any significant ground bounce or similar issues. Waiting forward for the correct answer, prof. Ben-Yaakov :)
Please correct me if I'm wrong. The first example doesn't show the rest of the circuit, and might be acceptable if there are no signals going from the digital domain to the analog domain. The other 2nd example clearly show digital signals interfacing "sensitive" analog circuits (otherwise why filter them ?), but the return current must find it's way back through the bead in the return path - therefore causing ground bounce/ringing. These issues are basically the same as splitting digital/analog ground planes while having digital signals that need to control the analog circuit. In the 3rd case, if it is indeed only a DC power supply - it has only DC current through it and the bead has negligible dc resistance, therefore it doesn't impede the return current flow. If the board in at the other end of the cable is pulling non DC current from that power supply, that current will be supplied by the decoupling capacitors and not the cable.
Depends on where is a current leaving its source and how the current coming back. And where and what is a common reference (some people use the word "ground" for common reference). So, if the circuit dealing with the same reference source on both ends, the bead on the gnd (common reference) is a bad idea. Also, you (usually) dont want currents from different sources to superimpose (by keeping the current loops separately, the gnd beads will kill this noble approach. .But, I probably can make up an artificial scenario when beads on the source line and the return line may help. Theoretically, isolated power supplies can be visualized as 2 sources with "indefinitely large beads between them, but then look at the differences on common nodes between these 2 sources. it can be very high if switching is involved.
I would ask, what is the reason Ferrite beads are being inserted into the system in the first place? If it is to block radiated and conducted emissions, then the un-intended RF transmitters need to be fixed. Most likely bad PC board layout and decoupling of RF generating components. i.e. digital stuff. Seeing the use of Ferrite beads on computer cables, that is there to bandaid the system to pass FCC emissions requirements. If it is to block HIRF susceptibility energy from entering the system, then I see no issue with them in any signal path. However if susceptibility is that big of a design requirement, to pass during a qualification test, then there should be no unit in the system with an isolated ground that has an external pin as the ground return path, the chassis is the path. The unit may have special internal isolated grounds for precision analog and such, but external power input, assuming chassis is the return path, are there mainly for unit test outside of the integrated system interconnect. My experience is avionics using +28VDC power. There the ground is the aircraft airframe or some other metallic structure in the case of carbon fiber composites. Note in those aircraft there must be enough metal to be able to survive a 200Kamp lighting strike. The avionics are bonded to this super structure with low inductance connections.
I don't think I'd adopt any of these without hesitation. I'd considering the return paths first . A high speed SE signal I definately wouldn't. High speed differential would be ok with bead in the gnd as parasitic capacitance would likely provide enough path for any even mode currents from pair skew. Slow edge rate signals would be fine with the beads in the ground but I'd still only put them in if it was likely a source would conduct through that ground connection instead of where it was meant to be going and even then I'd be looking at seeing if I could bypass the source return path closer to it's generation. Cabling is where it makes most sense to stop any common mode sources from using your device as a return path but then you have to have beads on all your signals in that cable bundle which may mean you need to use common mode chokes instead to retain your signals return path. In summary #1 no (unless I was trying to do a separate DC return path with AC someplace else) #2 ok since I believe those ESP signals are very slow although you could replace the signal path ferrite beads with 1k resistors. I'd also be concerned with making the signal driver drive 1nF caps. #3 ok. Might need a ferrite on the signal line too if the inductor has large parasitic capacitance. I'd use a common mode ferrite choke instead if it were a fast signal.
Hello Professor and thank you for doing these videos I have gained more knowledge here than before - although I regret my maths is dreadful! I have been working on power supplies for my own company for a number of years with some success and a lot of failures of course ( good learning) I have often thought about the circular currents that DC - DC converters have, and thought we should keep the tracks short ( on switching nodes of course) and maybe isolate them with Ferrite beads - I have tried - and it seems to work, to put a bead in to the input ground before any sort of input reservoir cap and then one after the first output reservoir ( in it ground to the rest of the circuit) to keep the circular currents from going on the common ground? I am using a 70Khz converter right now but I have used ones up to the 1Mhz range too. - I have not been able to prove 100% the right or wrong for this shorting out the ferrite has made it worse or better in some cases so this has to be board layout. - I am super critical to keep this at the far end and away from sensitive circuits - but the best i have is - it depends! What are you thoughts - If I may ask?
1. First case it depends on return path- layout? If there is no return path for high frequency component then it is not good. The return path should be as short as possible. Basically it could be done if the communication of other part does not have high speed communication with isolated(via ferrite) device and we want to prevent high frequency currents . I would not reccomend to do so. 2. Case I think is not recommended since it will limit the communication speed roughly to 100kHz. If communication speed is low then it is reccommended. 3. It is recommended since signal is differential and it limits common mode voltage high frequency part only.
If you want to absorb HF energy on power rail, I think it's better to do on V+ only. Doing it on GND is equivalent for the rail in isolation (current flows in loops) however now all your signal nets get their transitions rounded, and worse they now crosstalk/ground bounce. In the third case the bead is applied to a single net so the above caveats don't apply, so I think that's fine. Prof, would you consider doing a video on selecting core materials for various applications? How do you estimatr how much impedance and at what frequency would work for a design - or is it usually trial and error like I've seen at the EMC test house?
I think the 3rd is a differential circuit above and single ended below with bead in return path. I don’t think it is a good idea to put ferrite bead in the ground path because doing so can result in signal integrity problem. Considering this, in my opinion the only case where it can be used is the first one since the bead will keep the HF curren.t confined in the digital plane, preventing like this the digital noise from entering the analog area.
When we say ground it should always be a low impedance return path. Adding ferrite bead on the gnd increases resistance in the path in mohm range during low frequency and in ohm range during high frequency of return current. It is certainly not a good thing.
#1 OK if only local ground is used in the circuit and there is no further reference to the main Earth. #2 OK too, however unnecessary. The ESP8266 is a pretty insensitive device. Will work just fine with or without that filtering. Since it draws around 70mA it is probably not going to induce a noticeable voltage drop as long as its AD converter is not used referenced to to the system ground. Acceptable solution if EMC and minuscule RF leakage is a major concern, however I'd suggest smaller value caps of 10pF instead of 1n ones close to the ESP8266 side. #3 We don't know what device is connected on the other end of the cable. If the isolated signal at the source is led back to the grounding in the connected device, it is reasonable. If the connected device is referenced to the Earth, then it is not that great idea.
1. Not a good idea, the bead in the ground path could corrupt the signal since the ground impedance is not zero anymore. 2. Third one, since the bead is in the power supply return path, and the signal path is isolated by the transformer.
I am not sure of any of the answers, however I'm going to take a chance on Q2. Q2) In my opinion it is example #2 because by having capacitors on both sides of the ferrite it presents to both domains a preferred path for the high frequency noise current to close through. I would like to add that my laptop, as many other probably, has a ferrite enclosing both ground and supply lines. What is the difference from this case to what is being presented? Looking forward to you reply.
@@sambenyaakov Hi. I am talking about this: www.radiationhealthrisks.com/wp-content/uploads/2019/02/Ferrite-Bead-Laptop-Charger.jpg As far as I know they are ferrite beads used to filter out high frequency noise from the laptop chargers.
I must ask, as one who is in the field of engineering in general, is it a good idea to bead between two grounding rods in an apartment system, such that the interference and RF of other systems in my apartment complex do not disturb my sensitive electronics in my home? It would make sense if the main apartment grounding rod is of horrible quality, since at any point a higher or lower point in potential, even if it's the ground of my computer case, that point can act like an antenna to distribute the charge that it absorbed somewhere else -- I presume. So while I have improved the grounding of my entire apartment by feeding everyone's grounding through my own ground (I'll need a GFCI breaker at the power point where the second rod is attached), I'm not certain that it's going to improve the quality of the ground plane without an isolation bead against the main ground (which is rusted pitted and very very old, made of galvanised steel.)
I have been questioning the same thing. I even noticed a circuit with an inductor in the ground path. In general, I don't think it's a good idea to put a bead or inductor in the ground path as this can cause a large voltage difference for some frequency between the analog and digital ground. This might be harmful to some components, for exmaple an ADC.
Can we calculate the turns rato of an smps ferrite transformer , if we measure the inductance of all the winding. The transformer primary has 1 milli Henry. 1st secondary of 5volt has .01 mH Auxiliary has also .01 mH and the 2nd secondary has .02 mH
