What Every PCB Designer Should Know - Crosstalk Explained (with Eric Bogatin)

[@foobarables]

For those who became hungry: Signal and Power Integrity - Simplified by Eric Bogatin. An excellent book.

[@ekarter1981]

Recently Rick Hartley and now Eric Bogatin... Wow! That's the real deal!

[@RobertFeranec]

:)

[@OrsHunorDetre]

@@RobertFeranec Next one Henry W. Ott?

[@Tech2k25]

The last slide has answer to crosstalk. 1- Keep the transmission lines shorter than the rise time of the high speed signal into consideration. 2- Keep a reasonable distance between the two high speed signals e.g 2 time of width of the trace 3- Use stripline configuration i.e. use the high speed transmission lines between the dielectric material i.e. use a separate signal layer for high speed signals that has on top and bottom the dielectric material layer. Then on top of top dielectric the low speed signal layers or any other interconnect and on the bottom of the bottom dielectric use a solid copper ground plane

[@thisoldjeepcj5]

A clear understanding of the basics is your best tool.

[@karthikmaheshrao6021]

As a serdes engineer i come across so many issues of crosstalk on boards and packages. This video was on point and helpful. Thanks.

[@RobertFeranec]

Thank you Karthik

[@mixguru9669]

Eric handles the host very well in addition to the amazing explanation. 😎

[@alexpioner]

25 years ago one electronics teacher has said me: "Alex, the more you study the more you understand you know almost nothing" :) Robert, thanks a lot for proving this to me :) Like your videos very much! I've discovered your channel a couple of months ago, and I'm trying to see as many of your videos as possible, and recommend them to colleagues. I can even say, that after about of 20 years of experience I'm starting to understand something in board design :) Since you've asked for our opinion regarding your sessions with the experts, it can be cute to join some engineers from different industries having some real questions and problems, so the experts can try to help, and then show the results. Best wishes to you and your channel!

[@RobertFeranec]

Thank you very much Alexander for your words. PS: Number of people suggested to make a video with more than 1 guest .... so I keep playing with the idea, I am just not sure how it would turn out.

[@bennybrown4342]

Thank you, Robert, for creating these excellent videos! Books are good, too, but nothing beats a real conversation on a complex topic. And thanks to Dr. Eric for the clear and enlightening explanations! Please please keep making more of these videos, esp. with Dr. Eric :D Some questions/requests I have for you: 1. Can you please discuss termination resistors? What do the signals look like with and without them? Best placement of them? Do most ICs requiring controlled impedances have them built-in, or are they always required externally? 2. What do I do about high-density ICs? When I have something unavoidable like multiple GHz MIPI traces converging into a tight area under a BGA, are there design techniques I can use to mitigate cross-talk? 3. I have seen a number of designs in real products that have high-speed traces on the surface layers that have good SI and EMI performance. But the difference described in this video between microstrip and stripline cross-talk makes me think microstrips should be more problematic. Is it really possible to have good SI/EMI with microstrips? 4. How do these cross-talk concepts apply to differential signals that are tightly coupled with each other?

[@ats89117]

The technical term for that situation is, you're screwed... Eric Bogatin

[@santhoshk8703]

🎯 Key Takeaways for quick navigation: 25:11 *🔋 Energy from aggressor. * 26:35 *💡 Inductive coupling subtleties. * 31:06 *⚡ Capacitive & inductive noise. * 34:24 *📏 Near-end & microstrip crosstalk. * 36:44 *🔄 Crosstalk scaling factors. * 43:41 *🔀 Reflections and crosstalk.* Made with HARPA AI

[@MaxWattage]

Another fantastic video, greatly appreciated. I say this as someone whose restless brain keeps them awake at night with bad memories of a horrendous crosstalk issue I had 26 years ago! It was a system with a 40-bit parallel data bus communicated between two equipment racks over several metres of ribbon cable. It had cross-talk in the ribbon cable, ground-loops, EMI, ground-bounce, simultaneous-switching-noise issues, you name-it.

[@RobertFeranec]

Thank you Nicholas. PS: I am very happy, that I am not the only one with really bad experience with crosstalk

[@emredincol]

Thanks for this excellent video. When thinking about electromagnetic effects there is always three types of currents to consider: 1.Conduction current 2.Displacement current 3.Convection current

[@mohammadkarimi8359]

Thanks really for your online interview with Eric and expert engineers like him. Please do like this again. My suggestion is, please at the end of each question , you mention the vital points like that you do for two layers pcb and its crosstalk till we get the exact meaning of their talking.

[@RobertFeranec]

Thank you Mohammad

[@鄭峻杰-i2k]

Thanks. It's much clear on Crosstalk. The cause and responsible solution. Thanks again.

[@maheshu3938]

Finally i got clarity on crosstalk. Thanks Robert for your wonderful video...

[@abhishekupadhyay3419]

Good going, thanks for bringing the Eric Bogatin here... more power to you keep going. :)

[@RobertFeranec]

Thank you Abhishek

[@ElectricNimPedalsPhilippines]

I cant believe im getting all these insights for free :) Thank you guys!

[@sigfreed11]

Yes please continue the crosstalk discussion. Now that we better understand WHAT it is, let’s discuss how to mitigate crosstalk issues as well as understanding NEXT/FEXT plots, and how do we know if we have a crosstalk issue or something else?

[@RobertFeranec]

I asked Eric.

[@amoghjain]

woww!!! thankyouuu soo much!!! another great in-depth and highly conceptual yet simple to understand video!

[@RobertFeranec]

Thank you very much Amogh

[@Retinatronics]

When somebody knows something in depth; he will be able to explain it very clearly. Thank you for the video.

[@jorgemunoz5678]

Great video! Please, create another one about this topic it is both interesting and extremely useful.

[@RobertFeranec]

Thank you very much Jorge

[@magdabalazs5598]

I finally got to understand better the crosstalk! Eric Bogatin explains it so good and you are asking the best questions! Thank you very much for these videos!

[@montvydasklumbys7584]

Really liked the whole video, really useful content! Actually, correct me if you can, but I actually like to think about crosstalk like to what happens in the generator, just on a smaller scale: 1. When we move a magnet next to a coil, we mechanically create a changing magnetic field, which induces current in the coil. We use a coil to induce a lot of current, however we can identically use a single wire, it will only induce less current. 2. Similarly, electricity traveling in the coil acts as an electromagnet. If we replace a coil with a single wire, we will still create an electromagnet, just many times smaller in size. 3. Now we put this all onto a PCB, where a signal travels down one line (electromagnet) and there is an adjacent line (generator coil). We do not continuously move one line against another, but instead we generate changing current (di/dt), which creates a changing magnetic field... 4. The generator coil (victim line) picks up the changing field, which results in induced displacement current!

[@Ghost572]

This video was really good. I Learned a lot in 50 minutes.

[@serhiilozhechka4828]

Well done. Thank you for explanations.

[@gharbisalem1254]

Hey Robert thank you for sharing , we hope you can provide more of these videos

[@RobertFeranec]

Thank you gharbi

[@MatthewHoworko]

Such an informative and useful video! It really demystified crosstalk for me. Thank you Robert and Eric!

[@RobertFeranec]

Thank you Matthew for leaving your feedback

[@stephendeese97]

This is such excellent content. I would love to see a followup where you give practical tips for avoiding cross-talk (by applying what we have learned here).

[@santhoshk8703]

🎯 Key Takeaways for quick navigation: 00:00 *🔊 Crosstalk issues are common in PCB design and can cause various problems such as freezing, crashing, and interrupt requests.* 02:08 *📚 Crosstalk fundamentally involves electric and magnetic fields, where signal and return paths act as aggressors causing noise on victim paths.* 04:02 *🧠 Electric field coupling is approximated using capacitors, while magnetic field coupling is approximated using inductors.* 05:52 *💡 Changing electric fields induce displacement current through capacitors, leading to crosstalk.* 08:13 *⚡ Induced voltage on victim lines occurs when magnetic field lines change due to a changing current.* 09:09 *🛠️ Engineering approaches to reduce crosstalk include increasing distance between traces and bringing the reference plane closer.* 10:36 *🌐 A solid ground plane serves as a reference plane in PCBs, helping to control electric and magnetic fields.* 12:47 *🔄 In two-layer PCBs, the absence of a solid ground plane exacerbates crosstalk issues due to the lack of a reference plane.* 15:33 *📉 Crosstalk occurs when changing voltages induce displacement currents, leading to noise flow between aggressor and victim traces.* 19:20 *🔄 Capacitive coupling between aggressor and victim traces generates noise, with the amount increasing based on the extent of coupling.* 20:56 *🔍 Crosstalk noise propagates differently in the forward and backward directions, affecting the near and far ends of the victim line differently.* 23:57 *📈 Far-end crosstalk can become significant, potentially reaching half the signal magnitude and causing serious issues.* 24:39 *📡 Crosstalk involves both capacitive and inductive coupling between signal lines.* 28:15 *🔄 Induced current in the victim line due to inductive coupling circulates in the opposite direction of the aggressor's current.* 29:00 *🔃 Inductively coupled current propagates backward in the forward direction, causing a negative voltage pulse.* 31:20 *⚡ Capacitive and inductive coupling effects cancel out in the forward direction when of equal magnitude.* 34:24 *🛠️ Near-end crosstalk is mitigated in strip line configurations with homogeneous dielectrics above and below.* 37:02 *📏 Far-end crosstalk in microstrip configurations increases with coupling length and decreases with rise time.* 42:44 *🔄 Reflections from unterminated lines exacerbate crosstalk by causing multiple reflections.* 44:34 *⏱️ Longer rise times decrease crosstalk by smearing the effects of near and far-end interactions.* 46:03 *📊 The magnitude of far-end crosstalk depends on coupling length, rise time, and the coupling coefficient, which is influenced by electric and magnetic field interactions.* 47:57 *📉 Crosstalk saturation depends on factors like length, rise time, and coupling coefficient, with higher values indicating increased noise.* 48:52 *🛠️ Accurate estimation of crosstalk saturation requires field solver tools like Polar Instruments, as approximations may not be reliable.* 49:49 *📝 The video aims to improve understanding of crosstalk, inviting viewers to ask questions and suggest topics for future videos.* Made with HARPA AI

[@marvin19966]

So that's what guard traces are used for, very interesting and well explained video

[@whiterabbitangel]

Guard traces decrease the capacitor coupling. A simple, not entirely accurate, way to think about it is imagine two capacitors in series between the aggressive and the victim.

[@RobertFeranec]

Thank you marvin19966

[@rogerfurer2273]

I know that technical term: "You're screwed!" Thank you both. Question 1: Is ringing result of reflections? Question 2: How do guard traces affect the results?

[@krisjk999]

Qn:1, Indeed the ringing is due to not terminating with characteristic impedance. There will be reflections (Calculated using reflection coefficient). Qn:2, the guard traces can have interesting effects. If it is not grounded, it builds up a charge due to capacitive coupling and there is no shileding effect. If it is grounded, then the field is confined more towards the guard trace. In effect a similar effect of the ground plane. If guard trace is grounded, would expect inductive coupling to treat it as a ground path while those induced currents in the guard trace can still couple to the victim line. I would say that is an "it depends" situation based on my limited understanding

[@RobertFeranec]

Thank you Roger, thank you Jayakrishnan

[@santhoshk8703]

🎯 Key Takeaways for quick navigation: [00:00] 🚧 Crosstalk can cause issues like interrupt requests and system crashes, especially in two-layer PCBs with closely routed tracks. [02:36] 🧠 Crosstalk is primarily about electric and magnetic fields interacting between aggressor and victim signal paths. [03:19] ⚡ Crosstalk occurs due to fringe electric and fringe magnetic fields between signal and return paths. [04:44] 🔄 Electric field coupling is approximated by capacitive coupling, while magnetic field coupling is approximated by inductive coupling. [07:15] 💡 Changing electric fields induce displacement current, equivalent to conduction current, flowing through capacitors. [08:27] 🔀 Changing magnetic fields induce voltage on victim lines, driving current flow, proportional to the rate of change of magnetic flux. [10:51] 🛠️ A solid ground plane acts as a reference plane, crucial for controlling electric and magnetic fields to reduce crosstalk. [14:37] 📏 Electrically long interconnects, compared to the rise time, are prone to crosstalk due to distributed capacitive coupling. [20:02] 📊 Increasing the coupling region results in more capacitive coupling, leading to increased crosstalk in both forward and backward directions. [23:31] 🔄 Far-end crosstalk can grow significantly and cause substantial problems, even reaching magnitudes comparable to half of the signal, before saturating. 24:39 *🤯 Understanding crosstalk involves considering both capacitive and inductive coupling, with the energy transfer between aggressor and victim lines being crucial.* 26:05 *🔍 Inductive coupling involves changing magnetic fields inducing current, with the direction of induced current determining the nature of crosstalk.* 28:45 *🔄 Inductively coupled current in the victim line propagates backward, resulting in a return signal that's negative in the forward direction.* 30:22 *📡 Crosstalk behavior can differ between capacitive and inductive coupling, impacting signal integrity in transmission lines.* 34:52 *⚠️ Near-end crosstalk can be minimized in strip line configurations with homogeneous dielectric material above and below the signal line.* 37:02 *🔢 Far-end crosstalk in microstrip configurations can increase with coupling length and decrease with longer rise times.* 42:58 *🛠️ Controlling impedance and minimizing reflections through proper termination can reduce crosstalk in transmission lines.* 45:03 *📉 Longer rise times and shorter interconnect lengths can decrease both near and far-end crosstalk, impacting crosstalk behavior significantly.* 46:47 *📊 Crosstalk coefficients, influenced by coupling length and rise time, are essential for quantifying crosstalk effects in transmission lines.* 47:57 *📊 Crosstalk saturation depends on factors like length, rise time, and coupling coefficient, with longer traces and shorter rise times leading to higher levels of crosstalk.* 49:06 *🛠️ For accurate estimates of crosstalk coefficients, using 2D field solvers like the Polar Instruments tool is recommended over approximations provided by other software.* 49:34 *💡 Eric Bogatin's explanations, coupled with animations, provide valuable insights into understanding crosstalk, potentially aiding viewers in grasping the concept better.* 50:04 *❓ Viewers are encouraged to leave comments if they have further questions or if they would like to see more videos on the topic of crosstalk.* Made with HARPA AI

[@superciliousdude]

Another fantastic video! So much easier to understand from the visualisations in your video. Back in the day when I learned this stuff, the best we had was the Howard Johnson & Martin Graham book on high speed digital design. Love it!

[@RobertFeranec]

Thank you PS: I agree, we only had transparent foils at University - not the best tool for animation :D How lucky are today's students :)

[@derstrom8]

I ran into an issue recently with a digital video board I designed. I kept getting very thin (1px thick) vertical pink lines across the entire screen. I am fairly certain it has to do with the extremely tight coupling of tracks I had on an internal layer. This video definitely helped me visualize it much better, and has inspired me to revisit the board design and try to improve it. Thanks!

[@harishrao2952]

Thank you very much Robert for this video. Words are not sufficient to explain but the way eric makes every topic interesting is really good.

[@jeanclgcayer8259]

Thank you Robert (and Eric) I just found many solutions to my problems here !!!

[@RobertFeranec]

Thank you Jeanclg

[@sandeepkumartutiki851]

Wow Robert! really appreciate your work! Bring more topics with Eric and Rick..

[@yel1110]

The learning never ends, really interesting!

[@brokenicry]

PLEASE... You should definitely have more videos about this topic

[@RobertFeranec]

Thank you

[@Elektrotechnik-einfach]

Excellent explanations and great talk (Y)!

[@uccoskun]

Very informative. I like when you stop the video and explain things also.

[@RobertFeranec]

Thank you

[@guillep2k]

Excellent video, Robert, thank you!!

[@RobertFeranec]

Thank you guillep2k

[@burievsardor76]

Great work, Robert. I never thought that someone would make such kind of videos on youtube. This is what i wanted and it's amazing. I also never thought that a two layer pcb with solid ground plane would make crosstalk problems. Soon i will be your student on academy.fedevel. You are doing great work and keep on.

[@DiegoColl44]

Really nice video..!! Thanks Robert and Eric..!!

[@RobertFeranec]

Thank you Diego

[@dimtsk838]

Thanks Robert! Keep up this great contribution!

[@IXSigmaXI]

mindblowing. If i understand what Eric was saying, doesn't that mean that for a microstrip, which had the magnetic coupling outweigh the electric coupling, increasing the impedance would lower the magnetic component in order to restore the cancellation? Please bring Eric back!

[@ColinMacKenzieRobots]

Very informative. Agreed, great animation! Thanks Robert & Eric

[@sergeyivanov3453]

Please continue this series

[@LucasFernandesAndrade]

Awesome video, thank you very much Robert and Eric!

[@amigo2x]

Thanks guys! Very interesting and useful!!. And if possible yes would be nice to watch second part

[@placiduzzu]

Congrats for the nice discussion and the material presented. Engineering feast.

[@anuradhapatel6021]

BEST Video I have ever seen.

[@evolutionaryelectronics6787]

Amazing vedio on cross talk .....

[@BeMuslimOnly]

Thank you very much both of you. 💕💕

[@duttabikash0013]

These Video sessions with Eric Bogatin are exceptionally informational. Thanks Robert. GROUND BOUNCE/ RETURN BOUNCE would be another topic which Eric explains extremely well. Please do a video on this too.

[@RobertFeranec]

Thank you Bikash

[@ThomasBomber]

I am just starting to get in the field of PCB design and your videos are full of very useful information that helps me growing my knowledge. I hope can use that in my career in the future. Thank you!

[@Anaxroche]

thanks for your videos Robert

[@RobertFeranec]

Thank you özhan

[@venkannachimata6946]

Thank you very much for the great video

[@barismetin1014]

Very good video. Thank you very much!

[@edouardmalot51]

Very good teacher

[@sailtogether3236]

Really interesting details. Thank you!

[@Gengh13]

Awesome video, I'm starting to get an intuitive understanding of crosstalk. Keep up the great work.

[@RobertFeranec]

Thank you very much

[@mdchethan]

Hi Robert, Excellent information. Waiting for more of these. Also its been a while we have seen your videos related to PCB design, missing those.

[@RobertFeranec]

Thank you Chethan PS: Yes, during Christmas I used different topics

[@arsalansyed4709]

Thank you so much for making this video! This was super informative!

[@yunusYT7]

Awesome explanation!

[@RobertFeranec]

Thank you Yunus

[@66impala34]

Very well presented. Great work

[@gsuberland]

Fantastic video, and yet another great guest!

[@RobertFeranec]

Thank you Graham

[@hugoboyce9648]

Thanks for the video! It was very informative!

[@michaelk.1108]

Great video, Robert! Thank you! The last slide was very important for me, because it really showed some values. This gives me a better feeling of how much distance of traces is required. I would really like to see another video with 2 or 3 examples showing something like a good situation and one where it is really screwed up be crosstalk. And maybe a difference between stripline and microstrip situation. This is a really interesting topic. And also thanks to Eric for his great explanations!

[@RobertFeranec]

Thank you Michael. PS: I have already contacted Eric if he would be interested to continue talking about this topic.

[@CasualInventor]

Fantastic video. Helps me understand a problem I'm dealing with at work. I'm looking at 5 ns rise-time at 200 VDC! Over 100 signals that need to be routed with minimal crosstalk AND all length matched! I like a challenge though :)

[@hichemsetif1741]

You are welcome. It was a pleasure.

[@haribabuk850]

Thanks for this video, This type of videos are really interesting, this video really helps lot of engineers to get clear understanding on cross talk, really love the animation 😀, we were also happy when guessing that capacitive and inductive coupling will cancel out, there our Eric explained a complications about microstrip 😀 again it's worst for people who design 2 layer and 4 layer Pcbs, you are doing such a wonderful job by sharing this type of content with free of cost, this channel deserves alot 👌 Kindly make one video on how the tabbed routing helps to control impedance and cross talk( if you found it is interesting)

[@RobertFeranec]

Thank you very much Haribabu

[@22icyo]

please more! very interesting and well explained. Thanks!

[@RobertFeranec]

Thank you very much 22icyo

[@thenoisyelectron]

these videos are priceless thank you for your time and effort. I am going through a re-spin of multiple boards and I'm trying to soak up as much information as I can to make this the final revision

[@XxfieryfirexX]

A Video of pro tips on how to avoid crosstalk would be really great please!!! :D

[@michaellove7912]

this is good stuff, thank you both

[@giannisasp1208]

Fantastic explanation from mr Bogatin! Thank you Robert! It would be really interesting to speak a little bit more about how can we estimate or calculate a maximum coupled length for our signals (based on RT, track separations, e.t.c.) with some examples like we saw in the last slide. Also how would we estimate or calculate at what separation distance are our tracks “safe” and are not considered coupled any more, something like a “threshold”.

[@RobertFeranec]

Thank you. It's on my list ...

[@simonndungu1196]

Thanks Robert!!

[@RobertFeranec]

Thank you Simon

[@painchaud49]

This was amazing, thank you !

[@MyMcMichael]

Thank you a lot!

[@RobertFeranec]

Thank you Michael for leaving feedback

[@RadimPolasek]

thank you for this video i've learnt so much. now many things make sense :-)

[@pendalink]

Loved it. Learn your physics!

[@Philip8888888]

Great video. Almost makes me want to have this problem so I can understand it in a real-life situation! :D

[@KamleshGadhvana]

so how do we reduce cross talk in double layers pcb ?,thanks for the great video

[@RobertFeranec]

Thank you Kamlesh. PS: How to reduce crosstalk in 2 layer PCB? Honestly, after my very bad experience with crosstalk on multiple 2 layer PCBs, my best recommendation would be do not use 2 layer PCB. The 4 layer PCBs are not expensive anymore and that is what I would use instead.

[@ahmedalphy6104]

It is really great explanation!!

[@SerafinoConvertini]

Great as usual

[@chromatec-video]

Very interesting and great explanation of crosstalk between traces. Can I now suggest that Eric returns in future to analyse the effect caused between differential pairs when the edges are not phase matched - presumably this will result in crosstalk and is even more dangerous since the traces will be closely coupled over most of their length. Robert can you also please consider this problem with regard to package flight-times which mean that the signals are not in phase at the source before any tracking has even begun.

[@sajinm7625]

Please add another video for this topic.. i really appreciate your help

[@RobertFeranec]

Thank you Sajin

[@samihawasli7408]

Eric's answer to "why is the maximum coupling only 50%?" was pretty wishy washy. The fields just "equilibrate over time"? In my opinion, a more mathematical answer is to think of the "aggressor" as a signal source with its own impedance, if the impedance of the source and load are matched, the maximum power transfer is 50%. Love the videos you all produce, great stuff!

[@timpeng1269]

Great interview. Could you please have Eric again for some power integrity topic and simulation in hyperlynx?

[@dogukantombul1045]

49:12 what is Eric saying? which tool?

[@TD-er]

This really is a great and informative video. One thing I would like to know is to have some kind of rules of thumb to know when we enter the very scientifically defined "you're screwed". For example, what kind of ratio of length vs frequency can be used for 2 parallel traces over their full length at 1*w separation on a 2-layer board ? (somewhat of a worst case scenario) I do realize the rise time plays a very big factor in the "it depends", but let's assume it is 1/4th of the period.

[@RobertFeranec]

Thank you Gijs. PS: I may do some experiments with this and then a video, but you can find some answers if you play with a crostalk calculator. E.g. this may help: kzbin.info/www/bejne/hWW3aqyVlq58qqM

[@Xentura]

Fantastic! mind=blown

[@RobertFeranec]

Thank you Xentura

[@tablatronix]

What happens when you put ground between the 2 strips on 2 layer?

[@RobertFeranec]

We may be doing a video in future about that.

[@CarstenGroen]

Great and VERY informative!

[@RobertFeranec]

Thank you CarstenGroen

[@ThiagoSouza-ji9xy]

Very good.

[@RobertFeranec]

Thank you Thiago

[@Believer34545]

Great 😊👌

[@randomviewer896]

Does the polarity of the inductive crosstalk change if the victim trace is to the right of the aggressor trace instead of to the left?

[@manideepdunna827]

Thanks Robert for the amazing video with lot of good information. Can you please make a video with Eric Bogatin to understand the effects of Microstrip vs co-planar waveguide traces?

[@AbdullahKahramanPhD]

When I use SaturnPCB calculator for crosstalk, I don't get the same value in 49:00 for a 2 line width separated 50 ohm track pair that has coupled length of 2 inches and a rise time of 0.5ns.. The height to the reference plane is 0.1mm and er is 4.. I get 0.135V that is 10%. Anybody help me here? What am I doing wrong?