Has this ever been seen though? Or put another way how much impedance between rails and ground is needed for instability in a typical circuit - I did some simulation showing perhaps 10µH or so of stray inductance on the ground w.r.t rail is needed to provoke positive feedback via a loose ground like this, and that's a lot of stray inductance. Simulation attached.
I've taken some of the advice found here and decided it's worthwhile to get send a board out to be made by a board house since I want to use the same circuits in multiple applications.
I've decided on a two layer board and have arranged my components in a way that allows a nearly solid ground plane on the bottom layer.
Two bypass caps per op amp going from + to GND, and - to GND. Originally, I only had one going from + to -
My question is, should I also do a ground pour on the top layer too or will that end up just making an antenna?
I've decided on a two layer board and have arranged my components in a way that allows a nearly solid ground plane on the bottom layer.
Two bypass caps per op amp going from + to GND, and - to GND. Originally, I only had one going from + to -
My question is, should I also do a ground pour on the top layer too or will that end up just making an antenna?
I would not bother with a second ground pour on the top as it will be discontinuous and have very poor coverage. You already have the full benefit of the ground pour under the majority of your signal and power lines, which provides a minimal loop area for return currents therein.
Random other critiques that you can take or leave as you see fit:
I think you have plenty of space to fit the filter parts and the decoupling parts and the opamps in a single layer. But if you really want to move something to the back, I recommend leaving all the SMD on one side, and flipping all the through-hole parts to the other side. You can, if feeling rude 🙂 put SMD components between the feet of THT parts with that approach, but it looks like you have plenty of space here to avoid that level of hackery.
Random other critiques that you can take or leave as you see fit:
- Are your corner holes evenly placed? Top two look closer to the top edge. I like to pick a standard board size and hole spacing and stick to it.
- You appear to have SMD components on both sides; this is generally considered a bad idea because it's much harder to load the board, one side must be done manually, per-component, not with a board-level reflow.
- You appear to have IO terminals in the middle of your board; these are often best placed at the outer edge. KiCad has some nice screw-terminal footprints that work well here, in both 0.1" and 0.2" pitch. The extra trace length to bring PHONO_IN t the top edge is irrelevant compared to the wire length, for example.
- Your IO ports seem to be labelled only in the Fab layer, which does not get manufactured. You probably want to hide the Jx designators from the F.SilkS layer and insert human-readable text in F.SilkS for all ports, including their pinout. Some like to do that also for component values - it makes the board much easier to use if you come back to it in 2 years and ask yourself "wtf part goes here?" without needing to print a copy of the Fab layers.
I think you have plenty of space to fit the filter parts and the decoupling parts and the opamps in a single layer. But if you really want to move something to the back, I recommend leaving all the SMD on one side, and flipping all the through-hole parts to the other side. You can, if feeling rude 🙂 put SMD components between the feet of THT parts with that approach, but it looks like you have plenty of space here to avoid that level of hackery.