Pitfalls and Tips: Designing effects PCBs

z2amiller said:
I think the thing about sig/gnd/sig/sig is that the bottom copper signal is quite far away from solid copper, so you're back to thinking about return currents on those traces and placing a bunch of vias where it jumps a trace on in.2
Click to expand...
The return current for both sides would always be that Layer 2 GND plane. The only reason you'd need the extra GND vias to steer them is if you have two GND planes.

Granted, one signal layer is 0.2mm from the GND plane and the other is 1.5mm. If you're laying out a 2-layer PCB, the ideal would be a solid-uninterrupted GND plane on the bottom and all your traces on the top and 1.5mm away.

What I'm suggesting is a "3-layer" PCB is a foolproof approach that will give better results than a 2-layer design.
 
One hack I found a while back was to always short pins 2 and 3 of pots acting like variable resistors. That simplifies drop in replacements of different tapers should the initial taper not work.

Another recent hack was to create a footprint for the internal and external enclosure dimensions. Then I don't have to draw them every time for referencing internal clearances
 
Last edited:
My hack was making numbered pot functions in lt spice that map to the pins on my diptrace component. Essentially idiot proofing pot orientation. Another was making pot and switch components with just the pins and the component outline on the bottom assembly layer. The origin point for the components is where the hole goes on the drill template, so you can literally copy the values across, essentially idiot proofing placement. If you've noticed a subtext here you would be correct. I am
 
jubal81 said:
Came across another great video with tips for analog/mixed PCB design.

Click to expand...

They say here confidently and specifically not to separate digital and analog grounds without elaborating and then quickly move on as if there is no need to reestablish this

Is there any need to separate grounds? I understand the idea of keeping noisy loops local to their components by placing caps at the pins. Assuming careful placement, is there really no extra benefit to separating ground planes?

Robert here says LFOs should be separated, but that's like the opposite of an MCU or charge pump in frequency terms so perhaps the rules are different. And the oscillation is the whole point of an LFO rather than an artifact of it
 
ftumch said:
They say here confidently and specifically not to separate digital and analog grounds without elaborating and then quickly move on as if there is no need to reestablish this

Is there any need to separate grounds? I understand the idea of keeping noisy loops local to their components by placing caps at the pins. Assuming careful placement, is there really no extra benefit to separating ground planes?

Robert here says LFOs should be separated, but that's like the opposite of an MCU or charge pump in frequency terms so perhaps the rules are different. And the oscillation is the whole point of an LFO rather than an artifact of it
Click to expand...

I stopped having problems when I stopped separating grounds. When you think of a circuit, imagine it as a 3D energy field and "ground" as "return current path."

Long story short, you want to minimize the size of each 3D field by running traces over one big, low-impedance ground plane. Problems come from physically separating a signal from its return current path.

 
@jubal81 thanks mate appreciate that. Do you run traces for your LED returns or no need?

Edit: those vids are great. I've been routing my LED sends and returns around one edge on opposing sides so I think I'll keep doing that

Separating grounds for layout design and then later joining them into a continuous plane is probably good practice anyway because then the noisy stuff is automatically grouped into local islands
 
Last edited:
You must log in or register to reply here.
Back
Top