I’m just working on my first four layer board. It’s all THT, no SMD and I’m only using the extra layers to avoid snaking a couple of tracks that have to go more or less diagonally across the board.
The project is an analogue input stage for the Android ‘Scoppy’ (two channel oscilloscope) app. Although this could handle up to about 1Mhz I’m unlikely to ever need it do cope with more than a few hundred KHz.
With that in mind, am I better tying the copper fill for the internal layers to GND or leave them floating?
It’s the second. I normally always do a complete fill both sides but should I do the same on the internal layers? I mentioned the project only because someone might have a reason why a fill might not be advisable in this situation.
I understand that you are speaking here about 2 layer board. It is not possible to have complete fill at both sides until you have no other connections at all.
At 2 layer boards I always do complete fill at one side like here:
Complete fill gives to each signal track a best possible return path. These return paths should be as close to signal track as possible so internal layers are much better for it than opposite layer. Your fills should be connected to GND.
More you can read in links I provided at forum long ago:
I assumed people reading what I wrote would understand it to mean a complete fill of the areas not occupied by tracks or pads. It appears I was wrong about that. Never mind.
One of the reasons I use fill is to improve manufacturability. i.e. It greatly reduced the amount of etching required, thereby reducing the likelihood of tracks being etched into, or even etched through, as can happen with fine tracks.
As far as placement goes, sometimes other factors dictate that some tracks will need to travel further than would be ideal.
Anyway, it does appear that perhaps this wasn’t the best place to ask my question, so I’ll try one of the electronics forums.
Yes, that is true, the intention of this forum is to help people with learning how KiCad works. It’s not an electronics design forum, but questions like this are fairly common.
That is a broad assumption. GND planes only work when they are continuous. When they get ripped into little sections they do not work anymore as intended. The youtube video “How to design a proper GND plane” from Rick Hartley is very good. It’s a bit over 2 hours long, but it really is worth watching. That’s how good it is, so grab some popcorn.
Thanks for that info and I’ll endeavour to watch that video (with a strong cup of coffee in hand).
For most of my boards I don’t actually need a ground plane but still use copper fills to aid manufacturability. I got that tip from JLCPCB. It costs me nothing and takes only a couple of seconds to do. However, in this particular project I believe the ground plane is, at the very least, desirable from the electrical side of things. My only concern is what to do about the inner layers.
I have so far always eliminated floating (disconnected) areas of copper by using vias as appropriate and even resort to stitching the planes together where high currents or elevated temperatures are involved. I also ensure the important ground paths are as direct as possible.
It’s very unwise to leave a ground plane ‘floating’ in fact don’t, if it’s just a screen then connect to ground. I also use the planes top and bottom for the same reason ‘Less etching’ and of course they can be used as a form of heatsink. That’s about it, when you start getting into RF then its a whole new all game
FWIW, I generally flood my internal layers (and external). Ask yourself if the potential upside is greater than the potential downside? If you’re not dealing with RF or HV, one downside could be a tad extra weight. If RF, HV creepage/clearance or weight not a concern on your board then flood them. Your paying for that Cu, might as well use it IMO. If the other layers are mostly Cu, then having the ILs be similar can help with warping issues.
I’d second or third the notion against floating them. Then the question is what to tie them to and GND is usually the best obvious answer.
I’ll chime in on one thing I didn’t see above. It’s better to have one/some of the filled planes tied to the power as well as ground. This helps with “return path” and impedance for decoupling caps, power delivery, etc.
Also don’t forget to add vias so that any “islands” created by traces, holes, etc. are tied to the other layers.
I absolutely disagree with the idea of creating power planes just for the sake of doing so. My power traces are all as short as practical and somewhat wider than is necessary for their electrical requirements but not to the extent of creating large EMI antennas, as would be the case with power planes.
Already well covered in my previous posts in this topic.
What is the difference between power plane and GND plane that you assume one being antenna and second not?
Power plane shorted (for high frequency) in many places with GND plane by capacitors behaves like GND plane. For tracks near power plane return currents use it and not GND plane that is further away from it. When signal track jumps from layer along GND to layer along power plane you should place capacitor near that via to allow return currents jump from one plane to the other.
Each on their own is fine. The two in combination are not. Sure, you could tie them together with a whole lot of small capacitors, if that’s what you want to do or have a special requirement for it. Unless there is a very special requirement, he best solution is to eliminate the problem, rather than patching around it.
This getting an awful long way from what has been discussed here. Context is everything. You’re looking at a sentence and ignoring subject.
