I like to sometimes make home made PCBs quickly with toner transfer.
Both because this is basically what works for me with Manhattan style and ugly style construction, and also because its literally impossible to create grids of tiny plated through vias to tie top and bottom to one another in the same manner as with commercially fabbed PCBs, with my home made ones, I just dont do that, instead I basically just leave out the top ground plane whenever it seems reasonably justified, since i cant do the close spaced vias thing without it looking really messy.
So...I etch out most - sometimes all of the customary top side GP.. Sometimes I leave copper at the edges- when SMAs will tie the two together well- I leave out the ground plane portions of the top just using the bottom as the GP. This is usually with very simple boards. Where I need a ground, for example, under a SMT component, I drill a hole right there and use a bit of wire that exactly fits the hole and cut it flat with good side cutting pliers. Then I fill the hole around the little short chunk of wire by hand with melted solder. (sometimes I have to put some tape there to keep it in until its done)
So the ground is very short and solid. Where there is an SMA I also leave copper on top to supply it an anchor and far away from the traces I sometimes have a ring of copper around the edges with vias, but I try to avoid surrounding all the traces with a ground plane on top because with my home made boards I see the combination of GP on top as well as bottom with inadequate vias as being a potential resonant circuit. is this a valid approach for a person who just wants the occasional board right away, no delays, no drama?
Note, I still dont have a decent drill setup for drilling my PCBs neatly, I think thats the main reason I am so reluctant to make the committment to enclosing traces in a top side ground plane I cannot supply with dozens of vias automatically.
Doing this seems to work although - most of my boards are passive devices.. filters and so on.
is there any reason its a wrong approach? It seems to work fine for me, although so far- (this likely will change) my stuff is so simple that so far I havent really had to isolate powered traces from one another in quite the same way as I would with more complicated devices.
So far I mostly build little boards with one function per tiny square PCB and use SMAs to allow them to be attached to one another when needed.