One of the most important comments is "discuss with your board partner". We are busy with an RF design working up to K-band. We had an idea on the materials and stackup we wnated to use, but before we even started the PCB layout, we contacted our board manufacturer and discussed our requirements with them.
TBH for freq3GHz I would also avoid covering your tracks with soldermask. Also what is worth mentioning you should avoid FR4 cores for circuits with amplifiers (eg. RF transmitters) bsc this materials cant stand much heat. For RF boards I can recommend stackup with like: RO4350B core (35um copper and like 0.5mm height) + FR4 prepreg (eg. 0.2 mm) + FR4 (35um copper and 0.6-0.8mm height). I like that you mentioned mitter bends, bcs many designers forget about them. Also folks remember that every bend changes electrical length of your trace. Really nice video as always. :)
It might also help a bit to pick FR4 with a high TG because it has a higher ceramic content and therefore a more stable DC over temperature. I love the mitered bends too, they’re even better than round corners up until certain frequencies.
@@Niels_Dn With the Tg value it depends more where you deploy. I've done test antenna board at 5 GHz wifi with low and high Tg FR4, worked beautifully each time with no problems. The production versions were on high Tg. We talked about that in a different video on stackup materials.
Marcin one thing I forgot to mention about the surface finish, if you're in that low GHz range (2.4 GHz Wifi/Bluetooth) you don't need to even worry about surface finish. John Coonrod has a ton of data that I've cited showing the deviation in insertion loss on microstrips and CPWG with various platings, those curves all overlap up to about 5 GHz. For us, in the high frequency range for the radar and sensor products we do, we prefer silver due to the shorter shelf life with OSP.
@@Zachariah-Peterson I probably saw the data you are writting about and its ofc true. For high frequency and high reliability we used to employ an ISIG (Immersion Silver/Immersion Gold) finish, but novadays its hard to find a seller for this tech. We keep looking for nickless alternative for ENIG (like EPIG or EPAG, but not many providers in Europe have those in offer). Bare silver is not the best option for us bcs it tarnishes and need special hermetic casing - also most of our RF microwave circuits need some sort of tuning and testing before we can put them in enclosures.
Unfortunately Ben went to work at Autodesk for a short while, and now I think he's doing freelance stuff and just having fun building things. Dude's living the dream!
Hey Zach! Amazing videos, seriously! Just a quick question: In a board with an FPGA and DDR3 DRAM I only need 2 signal layers. Should I have 4 layer stackup with 2 signal layers and 2 GND layers, skipping the POWER plane, or should I go for a 6 layer stackup to make sure I have that power plane to ensure good decoupling? But then I will have 1 layer that I really don't need.
S21 is insertion loss for the signal injected at port 1 and traveling to port 2. S12 is in the reverse direction. I think the only time I mentioned S11 is with inputs to discrete components, which can give you very small amount of loss since they are not always perfectly matched to the system impedance, but they can get pretty close.
One of the most important comments is "discuss with your board partner". We are busy with an RF design working up to K-band. We had an idea on the materials and stackup we wnated to use, but before we even started the PCB layout, we contacted our board manufacturer and discussed our requirements with them.
Zack, I really love these videos. Both for brushing up and learning new things. keep them coming and great work
TBH for freq3GHz I would also avoid covering your tracks with soldermask. Also what is worth mentioning you should avoid FR4 cores for circuits with amplifiers (eg. RF transmitters) bsc this materials cant stand much heat. For RF boards I can recommend stackup with like: RO4350B core (35um copper and like 0.5mm height) + FR4 prepreg (eg. 0.2 mm) + FR4 (35um copper and 0.6-0.8mm height). I like that you mentioned mitter bends, bcs many designers forget about them. Also folks remember that every bend changes electrical length of your trace. Really nice video as always. :)
In the types of copper and roughness video that is the conclusion we land on --> don't worry unless you're gettting above that 10 GHz value
It might also help a bit to pick FR4 with a high TG because it has a higher ceramic content and therefore a more stable DC over temperature.
I love the mitered bends too, they’re even better than round corners up until certain frequencies.
@@Niels_Dn With the Tg value it depends more where you deploy. I've done test antenna board at 5 GHz wifi with low and high Tg FR4, worked beautifully each time with no problems. The production versions were on high Tg. We talked about that in a different video on stackup materials.
Marcin one thing I forgot to mention about the surface finish, if you're in that low GHz range (2.4 GHz Wifi/Bluetooth) you don't need to even worry about surface finish. John Coonrod has a ton of data that I've cited showing the deviation in insertion loss on microstrips and CPWG with various platings, those curves all overlap up to about 5 GHz. For us, in the high frequency range for the radar and sensor products we do, we prefer silver due to the shorter shelf life with OSP.
@@Zachariah-Peterson I probably saw the data you are writting about and its ofc true. For high frequency and high reliability we used to employ an ISIG (Immersion Silver/Immersion Gold) finish, but novadays its hard to find a seller for this tech. We keep looking for nickless alternative for ENIG (like EPIG or EPAG, but not many providers in Europe have those in offer). Bare silver is not the best option for us bcs it tarnishes and need special hermetic casing - also most of our RF microwave circuits need some sort of tuning and testing before we can put them in enclosures.
Just found this. What happened to Ben? Awesome set of videos and they just stop. 😢. Seriously, THE best introduction to microwave stuff I've seen.
Unfortunately Ben went to work at Autodesk for a short while, and now I think he's doing freelance stuff and just having fun building things. Dude's living the dream!
This is an excellent intro to RF, but I would like to see some actual calculations.
We have more videos coming on RF with calculations
One of them is quarter wave transformer and it has a homework problem in it
@@Zachariah-Peterson Awesome, I will be watching it for sure.
Wow... Even though I am an RF guy, learnt a lot of new things...!!
Great intro to RF PCB design aka black magic!
Thank you!
You're welcome!
Hey Zach! Amazing videos, seriously!
Just a quick question: In a board with an FPGA and DDR3 DRAM I only need 2 signal layers. Should I have 4 layer stackup with 2 signal layers and 2 GND layers, skipping the POWER plane, or should I go for a 6 layer stackup to make sure I have that power plane to ensure good decoupling? But then I will have 1 layer that I really don't need.
Insertion loss should be S12, S11 is for return loss
S21 is insertion loss for the signal injected at port 1 and traveling to port 2. S12 is in the reverse direction. I think the only time I mentioned S11 is with inputs to discrete components, which can give you very small amount of loss since they are not always perfectly matched to the system impedance, but they can get pretty close.