PCB: Ground Copper Fill...or not?

[tenser75]

I just noticed (after a year I'm using the software duh!)
that I can fill the PCB with copper...

I never had problem before without... but I started noticing some pedal manufacturers, they do use it. Sometimes they use it only on one side though...

Is it useful? do i have to use it on both sides or not? why?

Thanks for any advice!

[cloudscapes]

Couple reasons

1. For home-etched boards, some people use copper fills so that the etchant doesn't have to work as hard. It's quicker too!

2. Good for shielding sensitive/high gain parts of a boards from unwanted noise generated by another part. If a board shared both analog and digital circuitry, it's good to give both sections their own copper fill linked to  ground, and linked close to a single common ground point.

3. Higher speed digital chips need ground planes for proper inductance.

[miech]

4. Providing a heat sink to components dissipating a lot of power.

5. Providing power to a lot of components at once by making a power plane (4-layer PCB and up). Proper local decoupling is required, and must be implemented correctly in order to have any effect (that is: inbetween the supply and component being supplied to).

[amptramp]

We used to buy in boards with ground planes but the planes were not solid - they looked like a grid structure.  This was done to avoid warpage which can happen if a board is flow-soldered with copper on one side of a layer and not the other.  The grid can be filled in where a heatsink is needed on the board itself.  Multilayer boards are a different story - the ground plane is part of the decoupling capacitance.

Where components are soldered to the ground plane, some boards had circular spaces around the component hole to limit the flow of heat out of the joint when soldering by hand.

[PRR]

Ground fill is over-kill for most small audio like pedals.

Ground fill does save etchant, a real win in small etch-your-own work.

[LiLFX]

It absolutely makes a difference in a lot of circuits, but in the analog guitar pedal world it's kind of a coin toss. I've seen a lot of layouts in the guitar pedal community (even commercial products) that have issues way beyond what a ground pour will fix.

The idea for the ground pour is to capactively couple the circuitry to ground instead of neighboring traces which reduces the potential for noise and crosstalk. In the RF world you will see traces that have the ground pour running the entire length with dozens of ground vias stitched on either side of the trace.  On the other hand, when you have differential pairs you want them routed as tightly to each other as you can get them.

Long story short, the ground pour is an intentional functioning piece of the circuit whether it is for signal integrity or heat dissipation.

[R.G.]

Ground pours are a tool. Like all tools, there is a time when they are the right thing for the job, a time when they are nice to have lying around, but serve no particular function, and a time when they are in the way or just the wrong thing.

Ground >planes< are essential tools when suppressing power supply impedance is important. A ground pour is not a ground plane, it's just filling in around other stuff and usually has huge areas that are not continuous. A ground plane really needs to be more or less a continuous plane to properly do what it was intended for.

Nice to have lying around is where most ground pours in pedals come in. They look "official" to many people, which is a high-profit thing where it only takes one or two clicks to make the pours. This feeling of official-ness lasts until you have to hack and correct a mistake or two. It grows decidedly thin about the third fix you have to do. And on #4, you start vowing to be more careful about when you click those "pour me a plane" buttons.

Just the wrong thing instead of being in the way happens when the pretty plane actually capcitively connects things that ought not to be connected. Usually this is when a section of the plane-oid is not well grounded, or otherwise has a high impedance to real ground, and a low impedance noise maker gets coupled to a high impedance input - like a FET of some kind. Or a tube grid, if you've managed to shoehorn in a tube.

In the small, low frequency (that is, audio) world of pedals, mostly you get away with whatever planing you do, whether it's functional or just decorative. This is a lot like cheerfully ignoring grounding and just connecting things up in pedals - usually the application is undemanding enough that you get away with it. But it's going to cost you a lot of debugging time when one of those stars-align-wrong situations come up if you don't really understand what is going on with currents on the ground net.

[tenser75]

thank RG... always thoughtful deep answers... I'm learning a lot here...

[LaloFP]

When using ground pouring we can cause ground loops right? We delegate the ground path design to randomness.

If thats true, and we want that "oficialness-look", can just use copper fills that only connects to ground in 1 point? (each copper "island" connected to a single ground pad)
No copper "island" unconnected, and not all the grounds randomly connected.

Is there something Im missing?

Can those islands act like antennas?

[merlinb]

When using ground pouring we can cause ground loops right? We delegate the ground path design to randomness.
If thats true, and we want that "oficialness-look", can just use copper fills that only connects to ground in 1 point? (each copper "island" connected to a single ground pad)
No copper "island" unconnected, and not all the grounds randomly connected.

Yes you can do that, but in a pedal (and in fact in most audio circuits) there is no need. The ground loops are so small that they cannot suffer any significant self induction, especially once its inside a metal box. (I once had someone email me to tell me how I could significantly improve one of my pedal PCBs by breaking a loop that existed in the ground pour. The PCB was less than 2 inches square    )


The main exception to this in a pedal would be an LFO where you have heavy gulps of current throbbing through ground. Then you want to keep the LFO ground separate from the audio ground, right back to the power jack, to avoid ticking. But many LFOs are low current (esp. when we're designing for battery power) in which case you normally get away without this precaution.

Can those islands act like antennas?

Any patch of copper is an antenna, usually in the GHz range, but since they're inside a grounded metal box, which also contains no internal GHz transmitters, who cares?

[LaloFP]

Thanks merlin!!

I was searching about this yesterday and found that my suggested solution wasnt ideal neither. That the ground pours are less effective if they are connected to ground in only one point. That makes different ground islands that are not that "equal" and looks more like antennas. But... everything I find in the internet is usually for high frequency signals, so it gets a little difficult to put things in the right context.

Thanks!!