Question on decoupling capacitors; is there a reason why you chose to place the cap next to the via, instead of placing it on the line between the via and the power pin of the IC?
www.allaboutcircuits.com/technical-articles/clean-power-for-every-ic-part-1-understanding-bypass-capacitors/#:~:text=First%2C%20though%2C%20a%20brief%20note,another%2C%20and%20%E2%80%9Cbypass%E2%80%9D%20refers it will solve your problem after 4 Years if the question still exists.... hahaha
Wonderful video. Thank you! One question is if GND and Power layer are more 0.5mm apart, why we can not add global decoupling capacitor ? if GND and Power layer are less 0.5mm apart, why we can add global decoupling capacitor ?
Great video. Can you or someone please tell me, how to establish a return current path for a low frequency and low amplitude signal which is going to dual supply op-amp? As this op-amp will not have a ground pin. Also please clarify whether keeping the input power connector on the analog side of the ground plane is not going to mess up the along signals due to return path of digital currents.
Yes, that's true. When high-speed circuitry must be located between connectors, steps have to be taken to ensure the connected cables are not driven by the circuitry. This generally involves limiting the bandwidth of the circuitry or providing a low-inductance connection to a metal chassis at each connector location.
Very interesting fact about connectors on opposing sides. I have a related question: Let's assume I have an in-line amplifier in a shielded enclosure. Input and output wiring is shielded with the shield connected via EMC cable glands directly to the case. The unshielded conductors protrude the case through the EMC glands and need to be connected to the PCB inside the case. I would assume that it still holds true that the input and output connectors should be on the same side of the PCB because of the explanation in the video. But what about the EMC cable glands? Do they need to be on the same side of the case as well or can they be on opposing sides of the case, because the circuitry is not "electrically inbetween" the wires when the PCB connectors are on the same side of the PCB? Is it important to minimize the length of unshielded wire inside the enclosure? Is it sufficient to connect the enclosure to system ground at a mounting screw or would dedicated ground connections to the EMC cable glands be necessary? The input in this case will be a millivolt-scale sensor signal, while the output would be either an analog current loop or a digital connection.
Thank you! How to connect the metall audio connector? Should I connect it to common ground, to metal case? The connector will have 2 grounds (loop) - one dedicated gnd from audio-amp, the second from common GND and metall case.
Some issues here. The myth that you should ground shielding to board groundplane is dangerous, Shielding should preferably be grounded to the casing. Reason is that if you are a in a highly noisy environment i.e. an industrial machine hall, that cable shielding will act as an antenna inducing noise into your ground plane.
High-frequency cable shields must always be bonded to the metal chassis. If the cable shield is acting as an intentional current return, it must also be bonded to the circuit board's return plane. In this design, if there were a metal chassis, the circuit board's ground plane would need to be bonded (at high-frequencies) to the chassis near the connectors. In the event that the circuit ground was required to be low-frequency isolated from the chassis, that bonding would be done through capacitors.
@@LearnEMC You are right in the cases you have a connector that doesn't go 360 degree around your cable and have attachment to a casing made of metal. The reason for this exception is that you in this case have a Faraday cage and you avoid getting signals in and out through your shielding. You are right in case of plastic cases and in case of connector not having proper 360 degree shielding attachment (like in a RJ45 with the gaps) then you are forced to attach shielding to signal ground. The reason i mention this, is as i originally commented, that when you have machinery in a very noisy environment you risk having issues with EMI into your circuits. This may sound like an edge case, but having worked with it, there are a lot of machines out there where it is important to distinguish between Shield and GND.
This is essentially correct but it is often better to use separate analogue and digital ground planes rather than running individual analogue ground tracks through a digital ground plane to constrain the path of analogue return currents. The pin layouts on A/D devices are are specifically arranged to allow for this. Manufacturers data will often provide suggested grounding arrangements. Well done, keep up the good work!
The placement of the vias and decoupling capacitors is wrong. You want connect the power pin as close as possible to the capacitor and then connect the capacitor to the power plane via multiple vias. A via between the IC and capacitor is bad practice
@@LearnEMC For example. The audio ground should refer to the ADC pin, which is the star ground. New ADC have 2 ground, analog ground and digital ground. Supply from via must reach cap first then pin, not between then.
Oh the caps are thru hole, really ancient. It will cost more than smt, not to mention their lead has more inductance than smt counterpart. Digital trace are best route on layer 1, where it's nearest to ground. Layer 3 should use ground if layer 4 will route digital trace. Use power trace instead of power plane at either layer 1 or 4.
@@jasonhanjk A star ground is used for the unamplified acoustic signal, but it is not generally required or desirable for the amplified acoustic signal. Many ADCs have 2 ground pins, but they are connected internally. The decoupling capacitors are connected correctly for this application.
I think this woman doesn't know anything about industrial designer , during my 30 years working as engineer , you can't just move around connector to other place , marketing people would scream on you and tell you that we can not marketing this things. EMC has lowest priority now days and it comes at the end of project when you get stock in EMC lab which is too late to do any change in mechanical design.
A growing number of companies are designing their products to meet EMC requirements the first time they are tested. This avoids iterative design processes that add cost and extend product development schedules. Of course, there will always be some products that require connectors on opposite sides of high speed circuitry. These designs will generally rely on techniques that limit the high frequency currents and/or employ a ground structure or enclosure external to the board.
Using vias for the 64-MHz clock signal is adding unnecessary inductance which can cause reflection and signal attenuation at the load.
Very helpful video, helped me solve some EMI problems in a design. Thanks!
Also, shielding. I ended up shielding my digital signal cables, and analog signal noise on the board was reduced substantially.
Question on decoupling capacitors; is there a reason why you chose to place the cap next to the via, instead of placing it on the line between the via and the power pin of the IC?
www.allaboutcircuits.com/technical-articles/clean-power-for-every-ic-part-1-understanding-bypass-capacitors/#:~:text=First%2C%20though%2C%20a%20brief%20note,another%2C%20and%20%E2%80%9Cbypass%E2%80%9D%20refers
it will solve your problem after 4 Years if the question still exists.... hahaha
Thanks! Very interesting, well explained.
Good stuff. Its hard to find someone who can explain the basics of EMC.
This is such a great video! Thanks!
Wonderful video. Thank you!
One question is if GND and Power layer are more 0.5mm apart, why we can not add global decoupling capacitor ?
if GND and Power layer are less 0.5mm apart, why we can add global decoupling capacitor ?
Great video. Can you or someone please tell me, how to establish a return current path for a low frequency and low amplitude signal which is going to dual supply op-amp? As this op-amp will not have a ground pin. Also please clarify whether keeping the input power connector on the analog side of the ground plane is not going to mess up the along signals due to return path of digital currents.
Product/mechanical design constraints may determine connector placement.
Yes, that's true. When high-speed circuitry must be located between connectors, steps have to be taken to ensure the connected cables are not driven by the circuitry. This generally involves limiting the bandwidth of the circuitry or providing a low-inductance connection to a metal chassis at each connector location.
Very interesting fact about connectors on opposing sides. I have a related question:
Let's assume I have an in-line amplifier in a shielded enclosure. Input and output wiring is shielded with the shield connected via EMC cable glands directly to the case. The unshielded conductors protrude the case through the EMC glands and need to be connected to the PCB inside the case. I would assume that it still holds true that the input and output connectors should be on the same side of the PCB because of the explanation in the video. But what about the EMC cable glands? Do they need to be on the same side of the case as well or can they be on opposing sides of the case, because the circuitry is not "electrically inbetween" the wires when the PCB connectors are on the same side of the PCB? Is it important to minimize the length of unshielded wire inside the enclosure? Is it sufficient to connect the enclosure to system ground at a mounting screw or would dedicated ground connections to the EMC cable glands be necessary? The input in this case will be a millivolt-scale sensor signal, while the output would be either an analog current loop or a digital connection.
Thanks alot for your videos...! :)
Thank you! How to connect the metall audio connector? Should I connect it to common ground, to metal case? The connector will have 2 grounds (loop) - one dedicated gnd from audio-amp, the second from common GND and metall case.
Very useful video for learning about EMC
What if the external connectors can not be moved?
Put the board on a metal chassis and connect the circuit board's ground plane to the chassis near each connector.
Excellent! Very useful!
Great video, thanks!
Very Useful , Thank you
Excellent
Some issues here. The myth that you should ground shielding to board groundplane is dangerous, Shielding should preferably be grounded to the casing. Reason is that if you are a in a highly noisy environment i.e. an industrial machine hall, that cable shielding will act as an antenna inducing noise into your ground plane.
High-frequency cable shields must always be bonded to the metal chassis. If the cable shield is acting as an intentional current return, it must also be bonded to the circuit board's return plane. In this design, if there were a metal chassis, the circuit board's ground plane would need to be bonded (at high-frequencies) to the chassis near the connectors. In the event that the circuit ground was required to be low-frequency isolated from the chassis, that bonding would be done through capacitors.
@@LearnEMC You are right in the cases you have a connector that doesn't go 360 degree around your cable and have attachment to a casing made of metal. The reason for this exception is that you in this case have a Faraday cage and you avoid getting signals in and out through your shielding. You are right in case of plastic cases and in case of connector not having proper 360 degree shielding attachment (like in a RJ45 with the gaps) then you are forced to attach shielding to signal ground. The reason i mention this, is as i originally commented, that when you have machinery in a very noisy environment you risk having issues with EMI into your circuits. This may sound like an edge case, but having worked with it, there are a lot of machines out there where it is important to distinguish between Shield and GND.
Thanks
This is essentially correct but it is often better to use separate analogue and digital ground planes rather than running individual analogue ground tracks through a digital ground plane to constrain the path of analogue return currents. The pin layouts on A/D devices are are specifically arranged to allow for this. Manufacturers data will often provide suggested grounding arrangements. Well done, keep up the good work!
Can you help with an electronic design project?
The placement of the vias and decoupling capacitors is wrong. You want connect the power pin as close as possible to the capacitor and then connect the capacitor to the power plane via multiple vias. A via between the IC and capacitor is bad practice
Some points are still relevant, some are outdated.
Please be specific regarding any information you think is outdated.
@@LearnEMC For example. The audio ground should refer to the ADC pin, which is the star ground. New ADC have 2 ground, analog ground and digital ground. Supply from via must reach cap first then pin, not between then.
Oh the caps are thru hole, really ancient. It will cost more than smt, not to mention their lead has more inductance than smt counterpart. Digital trace are best route on layer 1, where it's nearest to ground. Layer 3 should use ground if layer 4 will route digital trace. Use power trace instead of power plane at either layer 1 or 4.
@@jasonhanjk A star ground is used for the unamplified acoustic signal, but it is not generally required or desirable for the amplified acoustic signal. Many ADCs have 2 ground pins, but they are connected internally. The decoupling capacitors are connected correctly for this application.
@@jasonhanjk None of the components in this example are through-hole. All of the digital traces are routed on layer 1.
does anybody know a good software for EMC/EMI testing ?
I think this woman doesn't know anything about industrial designer , during my 30 years working as engineer , you can't just move around connector to other place , marketing people would scream on you and tell you that we can not marketing this things.
EMC has lowest priority now days and it comes at the end of project when you get stock in EMC lab which is too late to do any change in mechanical design.
A growing number of companies are designing their products to meet EMC requirements the first time they are tested. This avoids iterative design processes that add cost and extend product development schedules. Of course, there will always be some products that require connectors on opposite sides of high speed circuitry. These designs will generally rely on techniques that limit the high frequency currents and/or employ a ground structure or enclosure external to the board.
thanks