I am not sure if we are talking about the same thing. In a DAC perspective with EMC, I am focused on practices avoiding the pcb to be the victim, not foccused on being the agressor. Of course the analog signal is going outside... and some DIYs do have separated front end pcb so both can matter.
Btw, wanted to ask : in the pcb edges surrounding square guard trace from the above illustration: do you need to remove the front and back masks ? I Mean thiese are the stubs than are guarding the interplanes radiations towards outside ! Is still working ok if some layers are not merging with the stubs or should we, as a good practice make a square guard at the edges on all layers tied to each others with the th Vias and let the main pcb layers recessed from the edges and spreading towards that 3D guard ?
What are the speeds/HF it becomes mandatory and then the average advised vias spacing, please ? (I mean about the pcb edges guarding); It could sounds as a sweeping generalization question, but what is the good limit with the DAC rise/fall time and new HF op amps ? 1 Ghz EMC design ?
Btw, wanted to ask : in the pcb edges surrounding square guard trace from the above illustration: do you need to remove the front and back masks ? I Mean thiese are the stubs than are guarding the interplanes radiations towards outside ! Is still working ok if some layers are not merging with the stubs or should we, as a good practice make a square guard at the edges on all layers tied to each others with the th Vias and let the main pcb layers recessed from the edges and spreading towards that 3D guard ?
What are the speeds/HF it becomes mandatory and then the average advised vias spacing, please ? (I mean about the pcb edges guarding); It could sounds as a sweeping generalization question, but what is the good limit with the DAC rise/fall time and new HF op amps ? 1 Ghz EMC design ?
Think of an antenna. It really doesn't care whether you think it is a transmitter or a receiver - it is going to ignore you and do both. Same with your board/design.
Are you referring to the drawing in #42. If so, you can leave the solder mask on both top and bottom. That drawing does show that the edge is plated - something I do routinely on high speed boards but may not be available for low volume orders. If you edge plate then you don't need the via fence.
To repeat myself... unless you transport you and yours back to the 1970s, you are living in an EM rich environment almost everywhere you can be. Yes, there are exceptions but I am guessing that you do not live in a bunker! So I suggest that high speed design is mandatory.
Go back to #42, the answer is there. Decide what your fMAX is i.e. the highest frequency on your design or in your environment. Then calculate the wavelength, divide that 50 to give the ideal via spacing.
Use something like Saturn PCB Toolkit to calculate the wavelength.
Are you referring to the drawing in #42. If so, you can leave the solder mask on both top and bottom. That drawing does show that the edge is plated - something I do routinely on high speed boards but may not be available for low volume orders. If you edge plate then you don't need the via fence.
To repeat myself... unless you transport you and yours back to the 1970s, you are living in an EM rich environment almost everywhere you can be. Yes, there are exceptions but I am guessing that you do not live in a bunker! So I suggest that high speed design is mandatory.
Go back to #42, the answer is there. Decide what your fMAX is i.e. the highest frequency on your design or in your environment. Then calculate the wavelength, divide that 50 to give the ideal via spacing.
Use something like Saturn PCB Toolkit to calculate the wavelength.
