I have a 6-layer PCB (F.Cu, GND, +5V. +3.3V, unconnected, and B.Cu) I wish to mount the MCU on F.Cu and its pull-up resistors and decoupling capacitors on B.Cu.
I know I could connect the MCU pad in F.Cu with the capacitor pad through a via. The problem is that the via is not part of the footprint. The via must be laid in the PCB editor. If I do it that way, I loose the freedom of moving and rotating the PCB at my will.
So I would like to use a pass-through pad to connect the pad on F.Cu with the one directly at the other side on B.Cu. If I use a pad, it can be grouped together with the MCU and other local components, so that I can move the whole assembly around the PCB trying to find the best place for it. I would be able to rotate.
So, after my explanation of “what” and “why”. here come my “how” question:
How do I configure a pass-through pad to electrically connect a pad at F.Cu with another one at B.CU without making electrical contact to any iinternal layer (GND, +5V, +3.3V and an unconnected In4.Cu)?
What if I want to ground the MCU pad and I have a GND layer? How do I set up a pass-through pad to electrically connect a pad on F.Cu with layer GND, without any contact with the other internal layers (+5V, +3.3V and Unconnected) and without contact also to B.Cu?
If I create in the Footprint Editor a set composed of a DIP16 MCU with its 16 rectangular pads in F.Cu, one thru-hole pad to connect to the pad of one of the terminals of a pull-resistor in B.Cu, then I can import this directly to the PCB Editor. If I lay it in a bad orientation, I can rotate it, and the connecting thru-hole pad will be shifted or rotated together with the MCU and its resistor.
If I create a Footprint with everything I described BUT the connecting “padvia”, when I import it to the PCB Editor, I still have to lay a “real via” to connect the MCU at F.Cu to the resistor at B.Cu.
The problem is: the via is connected to the PCB, not to the footprint. So, if I moved the set out of its first position in PCB, or if I rotate it, the via will stay put, and no longer connect the MCU pad at F.Cu to the resistor pad at B.Cu.
Do I make myself understand? Does it all make sense? This is my first ever PCB… Sorry if I am saying BS…
KiCad only has to types that go though a PCB. Pads and via’s. To connect to a pad, it pad number must be a part of the netlist, and to do this efficiently, this assignment is done in the schematic, and thus it also needs a schematic symbol, and either a wire connected to it, or a label. You can use an extra THT pad as a via, but the same rules still apply. It has to be present in the netlist.
But there is another option. You can create a group. All objects that are in the same group will always move together. Via’s, tracks, text and such can also be part of a group.
Thank you. So I can place the MCU pad in the F.Cu directly over the respective resistor pad at the B.Cu and then after I import it to the PCB I place a via in the middle of the intersection of the two pads, in F.Cu and in B.Cu?
Yes.
But first - do you have a schematic?
Don’t try to lay out the board without a schematic.
Assuming you do, you use the route tool to start routing from the Front side, and immediately hit the V key to place the via. Then you can finish the trace on the back side (which may be an essentially zero - length trace).
So I start routing on the front layer then immediately hiyt v, then route just a tiny bit on the back. shd end. Got it!
And yes I do have the schematic.I started with it.
I think you’re talking about via-in-pad. Check your PCB manufacturer will manufacture these, and if there’s a cost increase. They have a few gotchas, and are usually reserved for really dense layouts.
I’d just use normal component footprints, and a separate via, as is done in millions of PCBs without hassle. As above, you can group objects so they’ll all move together.
I have never been designing 6 layer PCB, but I would suggest to change unconnected into GND.
You use via. Even you want to connect top layer with nearest internal layer your via goes through all layers, but all other net zones when filled avoid contacting with that via.
There are special vias (micro-vias) destined to connet external layer with nearest internal one. I have never used them so can’t say more.
In my opinion your approach have no sense. When you will have your uC with resistors and capacitors done as a single footprint than you will decide that you want to connect small IC (SO8) to microcontroller and then it would be good to be able to move/rotate everything together. So what? Will you be designing the next bigger footprint having also this IC (and its blocking capacitor) in it. And so on, and so on and you will end with whole PCB being single footprint.
Elements working in groups I connect out of PCB and then rectangle select whole group (including tracks/via) and move/rotate to find the right place for it at PCB.
It seems almost at odds with a 6 layer board. My first (single layer) PCB I have designed around 1970.
Then for many years I was using 2 layer boards. Until I was using ICs in TQFP without thermal pad 2 layer was enough for me. My first 4 layer board I designed few years ago when I had to use uC in QFN with thermal pad.