Should I route reference ground trace next to 400kHz signal?

[liviu.gelea]

Hello. I'm working on a mechanical keyboard module with addressable LEDs chained with a 400kHz signal.
Since I'm using a double layer PCB for cost saving reasons and for other reasons related to standardization I chose to route vertical traces on the front side of the PCB and horizontal traces on the back side.

My LED signal trace needs to wrap around after passing through all the LEDs on a row of keys. Any other solution such as zig-zagging through the rows would provide little benefit and complicate routing.
I'm thinking I might require a return path for the signal especially for the long wrap-around trace. My first idea was to use the front plane as a reference ground plane, but since that ground plane is interrupted by a few vertical traces it would act as a poor reference.

My first question is whether I should route a ground trace next to the long signal considering the signal is pretty slow althought, being digital it does have some fast harmonics.

My second question is whether I could make such a trace disconnectable. Would that provide insight on how EMI would behave without such a trace or should I instead design a separate board without the trace entirely to get accurate results

Am I complicating things for such a slow signal?

Any other tips welcome. Thank you!

[selcuk]

Short rise and fall times of the signal may cause EM radiation. I recommend you to use ground pours on both sides. You may split vertical tracks by routing them from bottom and top to allow copper pours to pass continuously. You can add resistors to LED drive signals to increase the rise and fall times a little.

Check the figures (5.5) and formulas about bandwidth on this book chapter:
https://link.springer.com/chapter/10.1007/978-3-031-14186-7_5

[Smokey]

Like selcuk said, its more about the rise and fall times than the actual clock speed.  There are a million great youtube videos on signal integrity that will do a better job explaining that than I will.

Thoughts:
1) You are probably fine with doing whatever for the board with the addressable LEDs.  They are pretty forgiving.  Not a guarantee, but it will probably work as is assuming your power and ground rails are sized to handle the current (assume about 0.05A per LED).
2) 4 Layer PCBs are cheap enough now that it doesn't make much sense to use 2 layer boards if you are at all concerned about ground/signal/power interactions.  Worry about board costs when you get into the 1000s quantities.
 For a quick personal project just get the 4layer.  www.JLCPCB.com

3) I like the back-mounted LEDs with the board cutouts.  That's a cool idea.  I don't think I've seen that with those addressable LEDs before.  Let us know how well they solder like that.

[liviu.gelea]

Thanks everyone.

The back-mounted led is what the cool kids are using these days. I've previously used sk8612 (way too bright) and sk8605 but currently using Inolux - IN-PI32TAR5R5G5B because they are very available and easily sourced by PCBway. JLCPCB only works with LCSC who only sells the 12mA sk8612.
They are as easy as they get to solder since soldering requires a cutout which holds the LED in place. But the cutout is required. see attached 3d model I made.

I think I will end up using ground pour on the front layer and  some wide ground fills in proximity to the long LED signal traces and nothing else.
I can not route from "bottom and up" since, although featuring just 2 rows of keys, this is a proof of concept and final product will have multiple rows you need to get to. Still, great idea.

Thanks again for all the suggestions.