You've filled in so many of the knowledge gaps I had from undergraduate studies. One of, if not the best teaching channels I've ever had the pleasure to come across. Thank you for consistently delivering such good content.
That's very kind of you to say - thank you! Glad to hear it's filling in some of the teaching gaps found at most uni courses (I had the same experience). And thanks for watching :)
I usually don't comment on youtube, but I wanted to let you know how much I appreciate your videos. They are flawless not only from a technical perspective but also regarding sound quality, speaking rhythm and animations. Everything is extremely clear and what is most, you answer the comments regularly. Thank you for that! Greetings from Argentina.
As someone dabbling in PCB design after many years working with networking cabling, I finally understand the difference in signaling and design for PCB lines with this video. And I seen to learn something new and useful in every one of your videos so I am now a patron, thanks for sharing your knowledge!
Awesome content, as usual. Detail and care is shown on every video. I have some years of experience in doing Highspeed PCB design and I can say your explanations are right to the point. I would add "broad coupled" diff pairs to your presentation and also the requirement of a common ground return path for those currents coming from the imbalanced or losely tight diff pairs. Thanks for these videos!!
Really great video! Encapsulates a lot of good details regarding electromagnetic principles of and routing differential. I would love an equivalent style of video for DDR memory lines
Brilliant video as always, many thanks for the information you’re putting out there! Working on a project incorporating USB and putting out this video came at the right time!
A video on termination techniques would be awesome! 🙂 I am currently struggling designing a PCB with 8 daisy chained ADCs using SPI. The speed is around 10 MHz and trace lengths up to 80 mm. I am currently using a length matched star topology for SCLK, CS and some control signals with a single termination resistor at the transmitter. Would be awesome to hear what your approach or advice would be for this configuration. As always thanks for the interesting topics in your videos and keep up the good work :)
Thanks! Yes, definitely on my list of videos to make, which'll cover series and parallel (AC/DC) termination schemes. I'll go into detail when it's necessary as well.
@@alexanderquilty5705 I'm kind of late to the party, but I've got the same question you did. So how is Vi + Vg correct? If I understand correctly what he says about ground potential difference, shouldn't Vg always be greater than or equal to 0? Therefore Vo should be always equal or less than Vi, so Vo = Vi - Vg should be correct. What am I missing?
Thanks for this, incredibly clear and helpful for me. One question though, does it really say 50 ohm single-ended impedance in the USB spec? Most sources I've looked at (admittedly just random internet forums) seem to claim that USB specifies a differential impedance of 90 ohms and has no comment on single-ended impedance? Also, at 14:00, you suggest to set the single-ended impedance as recommended by the spec, and then adjust the spacing to achieve the diff impedance in the spec. But that example has the differential impedance as exactly double the single ended impedance, so applying that strategy would lead to an infinite spacing?? I do realize the overarching point you're making is that the spacing doesn't really matter than much, but still.
Hi, as always very well explained. Thank you! As I have a very specific task at the moment, hence I have a special question: I have to replace a 3m long Ethernet cable CAT-5 (100MBit) with a thin Flex-PCB cable for an application. As I have to roll up the cable and space is very limited, I am aiming for a Flex-PCB with a maximum thickness of 200um. According to PCB-Way, this can even be achieved with 4-layer FlexPCBs. Now the question: Ethernet cables would like to have 100 Ohm differential impedance. But there are no more reference lines in the cable except, of course, the other conductor of each pair. How do I then lay out the conductor spacing and widths? Or do I simple use the GND line of one of the circuits (before the magnetics) at one end of the cable? I would be thankful for any hints.
What about the rule of the return path? When the return path on the transmission line is not solid there is no impedance matching. How to setup in Altium for all differential pairs or x-traces?
why can't we route twisted pairs on PCBs? Can't a square wave approximation of twisted pairs be done by alternating layers with vias? Or a "stitch pattern" approximation with 45 deg races, also with vias? Is it any better than the pure planar version?
I'm very confused, wouldn't we want, LOWER capacitance between the two pairs and it's ground so the transitions are faster and sharper? and so we don't end up making a low-pass filter that filters the signal either. also why is the impedance so important and has to be exact? Sorry i'm a newb here!
Outstanding content as usual, do you have anything in the works for tips on laying out an ARM processor (one that can run Linux, A series for ex) and DDR2 / 3 memory? Like one of your step by step guides, that would be fantastic to see. Top quality work as always Phil. Thank you!
Thanks, Aaron! I'm working on a advanced hardware/PCB design course, which'll feauture some (embedded) Linux-capable device with DDR3 memory. I'm afraid that is quite a large amount of work and I won't be putting much of that on RU-vid (other than smaller how-to's).
Heey, great vid. By the way, what about the part 4 on sensor fusion, with extended kalman filter and the implementation of the part 3? Also, i was looking through your git repo, couldn't find the code from part 2 and 3 os sensor fusion, iam looking on the wrong place? Anyway, thanks for all your shared knowledge, Iam currently on a engineering degree, it's been really helpful!
... what about differential signals routing after a transformer, e.g. ethernet signals up to the connectors. I think this should also be an interesting additional topic
great video, one question what if we route/overlap the diff pair in adjacent layers? in this case do we still need a reference plane for the return path?