My experience with EMI and RF suppression makes it more of an art form rather than a science, in other words, its pure luck to have success. Its very unpredictable. Very enjoyable lesson, thank you.
Great video as always. I think there is at least one more important property of the common mode choke worth mentioning - the magnetic material utilization. With two separate inductors or beads, the amount of and type of material must be chosen to accommodate the desired (differential mode) signal. Since this current can be quite large you can may need a lot of material or a lower permeability material in order to carry the desired signal without saturating the inductor, which either increases the device size or compromises it's level of noise rejection. The common mode choke elegantly side-steps this by "exposing" the material to only the common mode current, which allows small volumes of high permeability materials to be chosen, giving good common mode rejection even in the presence of high differential mode currents.
The ESP8266 example is referring to the UART which is very low frequency by RF standards. It could go as high as 1Mbps but 115kbps or less is a lot more common. If the capacitor values for the signal lines are chosen carefully, it can greatly reduce EMI issues. The printer example is particularly noteworthy since the Prusa MK3/MK3S/MK3S+ does isolate the USB interface from the rest of the printer. I think it was demanded from the community since some cheaper printers have a bug with the USB power rail tied directly to the printer's main 5V rail, causing problems like the printer not fully powering off when connected to a PC, but that's separate from what such isolation is really supposed to solve.
Thanks for sharing your thoughts and experience. I believe I have said in video that the No2 will be acceptable if the bit rate is very low (1nF caps plus ferrite). Further, the question wa: which example could be accepted without questions or hesitation. Thanks
Hi sir this is a great video and very helpful. Thank you. I just want to understand How will an Inductor or a ferrite bead will differentiate between ac and dc at the same time when a noise imposed on the dc line?
I think the assumption that in #3 that the single ended signal is only "DC" is wrong. I mean who has ever called a fixed DC voltage "single ended". To me it looks like a digital buffer sending potentially high frequency data over. With that, I don't think you can use #3 without hesitation without first clarifying what kind of signal it specifically is.
7:40 hold up, if the power section is takig massive gulps of current like that, why not have that also star connected to the supply negative? Wouldn't a high side ferrite bead also help impede high frequemcy ground currents (KCL)?
Hi Sam! Great video. I was mainly curious about the ferrite bead in the GND path part. I designed a board that sends 5v power at around 1.5-2A through a 10 ft cable and I originally figured I would use a common mode choke but the PCB house I am using didn't have very many so just for quick short term prototyping I decided I would replace the common mode choke between 5V and GND lines with two ferrite beads, both the same ferrite bead product number and component, rated at around 3A with 30mOhm DC impedance. Product name is UPZ1608E101-3R0TF. Would this type of set up present any major issues in my hardware other than additional impedance to the current flowing to the cable? In this case I figured since the current needed to go to the other board is 1.5-2A worst case it would not be a major impedance but am I wrong in making that assumption?
HI Pietro, this should be OK. But if you need to send signals to the further away part, you need to consider the potential difference between the grounds. This is correct even for a common mode choke that will always have a differential component.
Thank you for making this video. My question is what happens when a ferrite reaches the capacitive phase? Does it still absorb the high-frequency noise and convert it into heat or does it store the energy and reflects it back into the circuit, making it inductive again?
@@sambenyaakov So when a manufacturer tells that a ferrite's filtering range is from 1Mhz - 1000Mhz but its impedance peaks at 500Mhz. Will the ferrite stop filtering frequencies from the 500Mhz to 1000Mhz range?
What kind of ripple the "power device" is generating? Frequency? Amps? How much current it draw? I would be very careful with separating grounds, unless you really know what you are doing. I would keep one solid ground plane and place blocks in sensible way on the pcb. Take into account that current of frequency above acoustic will flow in ground plane just below Vcc track. So in my opinion voltage drop on ground plane is unusual situation.
The power supply is generating mostly current ripples, but because the ground has resistance it automatically converts to voltage. The split ground is higly beneficial, because any copper track has a specific resistance and the total resistance between 2 points is specificrez*length between points. Using a common ground would induce higher voltages to the ground of logic components further from the source of ground power for the high current load
@@two_number_nines let's assume schematic from 7:46 shows PCB tracks (all this is one device on one PCB). If you put solid ground layer the ripple current will flow just below Vcc track on top of the PCB. and will not generate additional voltage in u2 . Now if you build this on one layer and Vcc is on top and GND on bottom you will have very large area current loop. All the best in EMC house
@@krysieks You have 10 iq. doesnt matter how solid a ground is it will always be reisitive. any current flowing trough the ground will always cause a voltage drop. the voltage at any point in a trace is current*specific resistace*distance from power source or load but with negative prefix sign. splitting the ground causes near VCC quality ground going to all logic circuits unaffected by the current and its ripples to high power loads. a solid ground would lower the resistance going to the load and raising efficiency, but would increase voltage ripple to logic circuits. To find out more google material specific resistance and ohm's law. And no im not watching a 52 minute enter-cational watchtimebait.
This is a long story. You first need to estimate the parasitic capacitances. I am considering a video presentation on this subject. Have you seen: th-cam.com/video/5NRiLOvmXdA/w-d-xo.html
@@sambenyaakov I have seen the video but am still don't understand I want to really understand how to calculate the value of the common mode choke, differential choke values also the Y and X capacitor value plaçed used to attenuate the noise and at what frequency does noise occur and what is the required cut off frequency for ac line filter
I don't get the phrase "it's ok to include a bead if the grounds are isolated". If the ground are isolated, this means that there is no physical conductor connecting them (galvanic connection). Obviously, we can have a parasitic capacitance between them, which is capable to transfer some noise. But in this case, how we can place a bead?
Consider a PCB island that you need to power from main part of board. In such a case beads in plus an minus lines will attenuate backward noise from island to main. Capacitance in between the two will just lower the effectiveness of noise rejection
@@sambenyaakov Thanks for the reply. In this case we agree! But to fully isolate two part of the board (eg the power and the control part) is achieved if we provide each part with a separate PS (and separate grounds). I may got confused by the circuit @12:28 which uses an opto but its two parts seems to have the same PS. So, in this case we use the opto to isolate the "u1" block from the "power" and suppress any noise coming out of the first. But since we supply both circuits via the same PS, beads are put on the power line, to suppress any noise transferred through them. Did I get it right? Another solution would be to have two separated PS, one for each side.
@@thece5511 Yes, the idea is to isolate the signal, and for the PS to provide clean voltage by capacitor and beads. Thanks for conversation. This is more economical than using an isolated PS.
Just ran into a situation where ground isolation and ferrite in the ground is beneficial. Everything spot on here.
👍😊🙏
My experience with EMI and RF suppression makes it more of an art form rather than a science, in other words, its pure luck to have success. Its very unpredictable. Very enjoyable lesson, thank you.
Good point. But even an art painters needs to know basic rules of canvases, brushes, point's and color.
@@sambenyaakov great answer to a good point
Great video as always. I think there is at least one more important property of the common mode choke worth mentioning - the magnetic material utilization. With two separate inductors or beads, the amount of and type of material must be chosen to accommodate the desired (differential mode) signal. Since this current can be quite large you can may need a lot of material or a lower permeability material in order to carry the desired signal without saturating the inductor, which either increases the device size or compromises it's level of noise rejection. The common mode choke elegantly side-steps this by "exposing" the material to only the common mode current, which allows small volumes of high permeability materials to be chosen, giving good common mode rejection even in the presence of high differential mode currents.
Thanks Larry for excellent input. The common mode choke certainly deserve a video on its own while here it was just an addon.
@@sambenyaakov A common mode choke video from you would be very interesting, I can't wait to see it!
The ESP8266 example is referring to the UART which is very low frequency by RF standards. It could go as high as 1Mbps but 115kbps or less is a lot more common. If the capacitor values for the signal lines are chosen carefully, it can greatly reduce EMI issues.
The printer example is particularly noteworthy since the Prusa MK3/MK3S/MK3S+ does isolate the USB interface from the rest of the printer. I think it was demanded from the community since some cheaper printers have a bug with the USB power rail tied directly to the printer's main 5V rail, causing problems like the printer not fully powering off when connected to a PC, but that's separate from what such isolation is really supposed to solve.
Thanks for sharing your thoughts and experience. I believe I have said in video that the No2 will be acceptable if the bit rate is very low (1nF caps plus ferrite). Further, the question wa: which example could be accepted without questions or hesitation. Thanks
Great video as always. Please keep doing intuitive videos.
👍😊🙏
Thank you for sharing your knowledge.
👍
Hi sir this is a great video and very helpful. Thank you.
I just want to understand How will an Inductor or a ferrite bead will differentiate between ac and dc at the same time when a noise imposed on the dc line?
dI/dt=V/L
@@sambenyaakov got it sir .. thank you
Thank you sir for the valuable lecture.
Thanks
Thank you very much for your work!
Thanks
How to choose bead specifications, just want to isolate digital and analog power supplies.
Look up other relevant videos in my channel. In TH-cam search box look for "sam ben yaakov bead"
Thank you for sharing knowledge.
Thanks for kind note.
What if signals are D+/D- differential. Would a ferrite in PWR and one in GND be a problem?
Thank you Sir.
You are most welcome
I think the assumption that in #3 that the single ended signal is only "DC" is wrong. I mean who has ever called a fixed DC voltage "single ended". To me it looks like a digital buffer sending potentially high frequency data over. With that, I don't think you can use #3 without hesitation without first clarifying what kind of signal it specifically is.
Clearly, the inductor at the high side implies a very low frequency signal, probably analog. This is why I simplified it to a power supply.
Can you use a ferrite bead in place of a mosfet gate resistor?
It has been suggested to use ferrite beads in gate drive to arrest oscillation . This might be useful for very fact transistors like GaN MOSFETs.
7:40 hold up, if the power section is takig massive gulps of current like that, why not have that also star connected to the supply negative? Wouldn't a high side ferrite bead also help impede high frequemcy ground currents (KCL)?
Lol I paused the video right then and continued watching, yup I thought so 👍👍
👍
Hi Sam! Great video. I was mainly curious about the ferrite bead in the GND path part. I designed a board that sends 5v power at around 1.5-2A through a 10 ft cable and I originally figured I would use a common mode choke but the PCB house I am using didn't have very many so just for quick short term prototyping I decided I would replace the common mode choke between 5V and GND lines with two ferrite beads, both the same ferrite bead product number and component, rated at around 3A with 30mOhm DC impedance. Product name is UPZ1608E101-3R0TF. Would this type of set up present any major issues in my hardware other than additional impedance to the current flowing to the cable? In this case I figured since the current needed to go to the other board is 1.5-2A worst case it would not be a major impedance but am I wrong in making that assumption?
HI Pietro, this should be OK. But if you need to send signals to the further away part, you need to consider the potential difference between the grounds. This is correct even for a common mode choke that will always have a differential component.
Hello Sir, thanks a lot.
Thanks
Thank you for making this video. My question is what happens when a ferrite reaches the capacitive phase? Does it still absorb the high-frequency noise and convert it into heat or does it store the energy and reflects it back into the circuit, making it inductive again?
@sambenyaakov
At very high frequency the bead behaves as a cap in all respects which is reversible if course.
@@sambenyaakov So when a manufacturer tells that a ferrite's filtering range is from 1Mhz - 1000Mhz but its impedance peaks at 500Mhz. Will the ferrite stop filtering frequencies from the 500Mhz to 1000Mhz range?
@@sambenyaakov I'm if my question sounds basic. I am not an engineer.
What kind of ripple the "power device" is generating? Frequency? Amps? How much current it draw? I would be very careful with separating grounds, unless you really know what you are doing. I would keep one solid ground plane and place blocks in sensible way on the pcb. Take into account that current of frequency above acoustic will flow in ground plane just below Vcc track. So in my opinion voltage drop on ground plane is unusual situation.
The power supply is generating mostly current ripples, but because the ground has resistance it automatically converts to voltage. The split ground is higly beneficial, because any copper track has a specific resistance and the total resistance between 2 points is specificrez*length between points. Using a common ground would induce higher voltages to the ground of logic components further from the source of ground power for the high current load
@@two_number_nines let's assume schematic from 7:46 shows PCB tracks (all this is one device on one PCB). If you put solid ground layer the ripple current will flow just below Vcc track on top of the PCB. and will not generate additional voltage in u2 .
Now if you build this on one layer and Vcc is on top and GND on bottom you will have very large area current loop. All the best in EMC house
@@two_number_nines btw see this video. th-cam.com/video/kcU-nT9jcSs/w-d-xo.html
@@krysieks You have 10 iq. doesnt matter how solid a ground is it will always be reisitive. any current flowing trough the ground will always cause a voltage drop. the voltage at any point in a trace is current*specific resistace*distance from power source or load but with negative prefix sign.
splitting the ground causes near VCC quality ground going to all logic circuits unaffected by the current and its ripples to high power loads. a solid ground would lower the resistance going to the load and raising efficiency, but would increase voltage ripple to logic circuits. To find out more google material specific resistance and ohm's law. And no im not watching a 52 minute enter-cational watchtimebait.
Thanks for participating and sharing the knowhow.
Sir how to calculate the value of the common mode choke for practical power supply
This is a long story. You first need to estimate the parasitic capacitances. I am considering a video presentation on this subject. Have you seen: th-cam.com/video/5NRiLOvmXdA/w-d-xo.html
@@sambenyaakov I have seen the video but am still don't understand I want to really understand how to calculate the value of the common mode choke, differential choke values also the Y and X capacitor value plaçed used to attenuate the noise and at what frequency does noise occur and what is the required cut off frequency for ac line filter
Thanks prof 🔥
Thanks
I don't get the phrase "it's ok to include a bead if the grounds are isolated". If the ground are isolated, this means that there is no physical conductor connecting them (galvanic connection). Obviously, we can have a parasitic capacitance between them, which is capable to transfer some noise. But in this case, how we can place a bead?
Consider a PCB island that you need to power from main part of board. In such a case beads in plus an minus lines will attenuate backward noise from island to main. Capacitance in between the two will just lower the effectiveness of noise rejection
@@sambenyaakov Thanks for the reply. In this case we agree! But to fully isolate two part of the board (eg the power and the control part) is achieved if we provide each part with a separate PS (and separate grounds). I may got confused by the circuit @12:28 which uses an opto but its two parts seems to have the same PS. So, in this case we use the opto to isolate the "u1" block from the "power" and suppress any noise coming out of the first. But since we supply both circuits via the same PS, beads are put on the power line, to suppress any noise transferred through them. Did I get it right? Another solution would be to have two separated PS, one for each side.
@@thece5511 Yes, the idea is to isolate the signal, and for the PS to provide clean voltage by capacitor and beads. Thanks for conversation. This is more economical than using an isolated PS.
At 12:12 you write path of interference however there is bead in path that should have high impendence for interference so path through U1 is wrong?
Thanks Marcin for note. Well, I may have been carried away, but strictly speaking, there would be some injection even if you have a filter😊