EEVblog® Electronics Community Forum
Electronics => Beginners => Topic started by: zimm on September 14, 2026, 07:04:04 pm
-
Hi,
I'm working on a PCB which has the traces on the bottom layer and the top layer is an almost uninterrupted ground plane:
[attach=1]
My question is, is it ok to place the SMD resistors on the top-left side on the top layer surrounded by the ground pour? Would it be better to move them on the bottom layer and have a completely uninterrupted ground pour on the top layer?
The resistors are just for pull-up and voltage divider.
Thanks
-
There are many opportunities for more effective routing by reducing outer pad diameters and routing between in some instances.
You provide no circuit diagram or design intent which makes further comment tricky however with the current information, no bad things will happen as-is with your resistors.
-
Thank you for confirming, the circuit is just an ESP32 dev board connected to 2 TMC2209 stepper drivers + filtering caps and various resistors (for pull-up and a voltage divider). I wasn't sure if it's mandatory, or at least strongly preferred, to have a ground copper pour for all components.
The reason I have those pad sizes and clearances is because I'll make the PCB with the toner transfer method then etch it with ferric chloride and it's my first time, so I prefer to have plenty of space between traces/vias/components. If I were to have it made at a commercial manufacturer, yes I know everything could be a lot smaller and more efficient. But I can't make through-plated vias or solder mask layers at home so I chose distances/clearances that hopefully I can manage with my not-so-great manufacturing and soldering skills.
-
Thank you for confirming, the circuit is just an ESP32 dev board connected to 2 TMC2209 stepper drivers + filtering caps and various resistors (for pull-up and a voltage divider). I wasn't sure if it's mandatory, or at least strongly preferred, to have a ground copper pour for all components.
The reason I have those pad sizes and clearances is because I'll make the PCB with the toner transfer method then etch it with ferric chloride and it's my first time, so I prefer to have plenty of space between traces/vias/components. If I were to have it made at a commercial manufacturer, yes I know everything could be a lot smaller and more efficient. But I can't make through-plated vias or solder mask layers at home so I chose distances/clearances that hopefully I can manage with my not-so-great manufacturing and soldering skills.
Post up pics of your completed project, I gave toner transfer a shot years ago but found the minute carbide drills difficult not to break! Good luck!
-
... and the top layer is an almost uninterrupted ground plane:
Euhmm... no. That does not look like a GND plane at all. For a proper GND plane, the sections have to connect in between all the pads of the THT IC's. Looks like KiCad. (which does this by default for some weird reason). A relatively quick way to fix this in KiCad is to change the pads to "oval" and make them narrower. This still leaves enough room for hand soldering.
For the rest, it does not matter much whether this GND plane is on the top or on the bottom, nor does it matter much where you put your SMD footprints. Either top or bottom is fine (for DIY, for ordering assembled boards putting everything on top is easier). Putting SMT footprints on top also offers more mechanical protection.
-
However, for DIY etching, leaving any large sections of extra copper in place, both sides, is a reasonable practice. First, your etching solution is going to last longer because there's less copper to dissolve. And the process will be a bit faster, because your etching solution remains fresh longer with less copper to dissolve.
I haven't done it in ages. For small boards, it's pretty cheap to send them out these days. And if I need something tomorrow, I'll just do a one-off with a proto board.
-
I don't know whether it's your choice or not, but to label those two row modules the same numbers on both sides is crazy.
Since you're using modules that hover over your PCB you might as well take advantage of that.
Since you'll be hand soldering SMDs, why not give yourself some elbow room?
-
Would it be better to move them on the bottom layer and have a completely uninterrupted ground pour on the top layer?
It would be better to have completely uninterrupted GND but you don't have it if you move these resistors or not.
Much bigger holes in GND are made by rows of pads without GND going between them.
If you look at holes in GND plane it is not the area of hole that counts but the longest dimension. From EMC point of view the 1mm x 5 cm hole is bigger than 3x3cm hole.
-
Thank you for all the comment and suggestions, I'll try oval pads to hopefully get a proper ground plane and moving the SMDs around to make it easier to solder.
Post up pics of your completed project, I gave toner transfer a shot years ago but found the minute carbide drills difficult not to break!
I did a test run with an earlier version of the board to see how it goes, and it turned out ok-ish. Alignment between top and bottom was slightly off, but other than that everything seems fine. For vias I used 1mm drill bits from Aliexpress, I think they're tool steel nor carbide and haven't broken any yet
[attach=1]
[attach=2]
The PCB is the control module for a custom satellite tracker and antenna rotator, for lightweight antennas. Since it's not finished yet all I can show is a KiCAD rendering:
[attach=3]
The whole thing is still WIP but functional, this is it in a action:
[attach=4]
-
If you move everything to the bottom, then you don't need to etch the top at all; once the through holes are drilled, you can use an 1/8" drill to remove the top copper around pins which don't connect to the ground plane. Grounded pins should be soldered top and bottom.
Your test board looks workable, but ground plane between 0.1"-pitch pads will be trickier to etch. If ground plane impedance is a concern then I would add solder to the plane where it passes the rows of pins, or widen the board. Grounding can be important in motor control applications, but this isn't RF or rocket science. I suspect the design would work as-is.