Dear Zach, your presentation is excellent. Love to watch your youtubes. You have showed us the technical things that I always wanted to learn from many years ago. Have a good weekend.
Hi Zach, thanks for sharing this important topic! And indeed you can have a 4 - 6dB dips in S21 and this leads to S11 (and S22) degradation as well, causing standing waves that can corrupt your SI and RF performance (or even damage your HPA). In my circuits i use "rule of thumb" lambda/20 pitch for vias that work with most manufacturers up to 6 - 8GHz, above 8GHz you can have vias placed at lambda/10, but it is good to perform CPWG simulation in this case. Regarding the VL - this is mostly determined by the CPWG impedance that you want to achieve in first place and then by manufacturers capabilities. Also for signal content above >10...15GHz it is wort to consider microstrips rather than CPWG as their losses will be lower as well as S21/S11 dips may be avoided.
Thank you for your great videos Zach 👍 Today Raspberry Pis are everywhere... Maybe you can make some videos on high speed routing on a CM4, like PCIE for example?
Sure, I was basing that example on the 14 mil via-to-via pitch. The point was to compare the via pitch to the hole wall separation using the drilled hole radius. I was saying to suppose our drill hole diameter is 6 mils, that means the radius is 3 mils, so subtract 3 mils and another 3 mils from the via pitch distance to get an 8 mil hole wall spacing. That might be a little low for some fabricators, I've been given no bid status by one fabricator in the past because I went below 10 mils. So if your hole-to-hole distance increases from 8 to 10 mils, the half-wavelength you can reliably isolate for the lowest via-to-via structure resonance also increased by 2 mils, or an additional 25%. That means the frequency corresponding to the first resonance in that via-to-via structure went down by (1/1.25), or by 20%.
Hi for pulses with very fast rise time to travel between one board to another board do we need transmission lines like coax or pcb transmission lines????
No it is not required to use a coaxial structure with very fast rise times, although you can use a pcb transmission line that looks coaxial. When rise times become very fast then the transmission line needs to become physically smaller as I demonstrated with the data shown in the video. Making the line smaller pushes the cutoff frequencies for higher order modes to larger values, and this is why you can extend the channel bandwidth with a physically smaller line. You will see stripline coplanar waveguides used in packaging and in PCBs, I have designed these in recent work for clients. You will also see microstrip coplanar waveguide in PCBs, some example boards were used in demos at DesignCon this past year.
@@Zachariah-Peterson thank u Zach now I'm sure to use a transmission line in my pcb I was planning for use an ungrounded coplanar transmission line I tested my designed line with vna and It's good for now I hope it Will works properly in final pcb...
