Should i ground pour an empty layer on my multi layer board

[drakejest]

Im about to tace my IC on my 4 layer board with a stack up of:

Top Signal
GND
Vcc(5v)
Bottom Signal

I tend to prefer using 4 layer boards in my projects as it makes routing the gnd and Vcc so much simpler only using a via. Even on simple projects i tend to use a 4 layer as it only a couple of dollars more which i don't mind paying in exchange for the sanity of having clean traces.

As i finished laying it out I can already see that i will be able to trace all the signals on the top layer making the bottom layer very empty. I have read from a quick google search that pouring gound on the signal layer is not a good practice. But i have this bottom layer totally empty exempt from a few pads for through hole components (which are routed on the top layer). 

Would placing a ground pour on the bottom layer still a bad idea?

[Siwastaja]

a quick google search that pouring gound on the signal layer is not a good practice.

This is a ridiculous claim. (Beware of "not good practice" / "good practice" tips. Require an explanation if you can't figure it out yourself.)

Of course, pouring ground on your routing layer in 4L design does not add much grounding benefit since the ground/power plane is already very close below, and pouring ground on signal layer as well would then require large number of vias so that you don't end up with isolated copper sections. So much work for little gain. Maybe that's what they were thinking by not recommending it.

For a totally empty layer, I would likely pour it, and stitch it to ground with vias. As an empty layer, you won't struggle finding places for those vias as the last step, so it's not a lot of extra effort.

Benefits:
* Plane capacitance with your Vcc layer. If the layer separation is 200um, this isn't much, but it's something.
* Sideways thermal conductivity, improving heatsinking greatly. Just bring heat to this bottom layer with vias. Even heat brought to the Vcc plane (for example: if you use a Vcc regulator which connects the output to the package pad) would couple to the nearby ground plane through the thin FR4, now the two planes act in parallel transferring heat sideways.
* Very theoretical shielding of the power plane. If the power plane is properly bypassed, it's neither noisy nor susceptible to noise even without such shielding, but hey, you get the extra shielding for free, so why not.

[drakejest]

a quick google search that pouring gound on the signal layer is not a good practice.

For a totally empty layer, I would likely pour it, and stitch it to ground with vias. As an empty layer, you won't struggle finding places for those vias as the last step, so it's not a lot of extra effort.

Thank you very much for your advice, it felt very odd to me to not place anything in there. So for my top signal layer which is populated with the signal traces (less than 50% of the space is used) i do not pour a ground on that layer?

I do have an LDO that will benifit from this extra "heatsink".

[Siwastaja]

You can pour ground on the top signal layer, but stitch it with vias not to leave totally isolated areas, or long antennas only grounded from one end. I used to do that, it's quite a lot of extra work with little benefit.

Nowadays I still like to pour the top, but I use larger clearance (something like 15 mils!) and neck removal parameters so that the ground does not pour in every small gap, to reduce the workload of stitching everything. Copper anywhere helps distribute heat, even if not electrically connected to the same net where the heat is coming from, and having all layers relatively full of copper would help reducing the risk of board warp (usually a small issue, overdiscussed in old literature just like acid traps and right angles). For EMI, there is not much difference because you have the ground just some 100-200 um away anyway.

For a 2-layer design, now filling both layers with copper with fairly small clearances and stitching carefully everywhere becomes important.

[drakejest]

Thank you for that sir, my software does not actually create pours on isolated islands which is great for spotting them also my traces are packed which helps reduce the