Modular PCB Logic Tiles: A Physical MUX and Wire Alphabet
January 31, 2026
After twenty-five years of design in thought, paper, and CAD, Brian Greenforest placed a physical computing alphabet on the desk. Purple MUX, wire, constant, and intersection tiles now snap edge-to-edge into circuits that read like a language.
Hermaphrodite connectors and an unambiguous one-sided orientation turn separate boards into a fabric that builders can assemble, lift, rotate, and extend.
A Set of Symbols Becomes a Machine
Each triangular tile carries one computational role. MUX tiles choose, wire tiles carry relationships, constants establish values, and intersection tiles let paths cross inside a regular spatial grammar.
The breakthrough lies in the set: a circuit no longer begins as a fixed board. It begins as letters whose physical arrangement expresses the logic.
Build the First Stateful Word
The first connected patch crossed the line from concept to manipulable material. The next construction uses three MUX tiles to form a D flip-flop whose state change occurs inside the assembled fabric.
Builders, educators, circuit designers, and programmable-matter researchers can extend the alphabet. Connect the tiles, create a circuit, and help turn Boolean relationships into a shared physical medium.
Build the MUX Alphabet From Tile to Algebra
The photographs and hands-on learning path show the physical symbols; MUX Algebra and the Logisim-to-LUT chapter connect those tiles to complete logic.
LinkedIn status when archived: Visible to anyone on or off LinkedIn.
Soldered the first batch of hermaphrodite connectors onto my purple MUX tiles.
After ~25 years of “designing in my head / on paper / in CAD,” I finally have a physical tile alphabet on the desk (MUX / wire / constants / intersections).
I snapped a little patch together and it instantly reads like a language, not a random pile of PCBs.
The big boss this whole time wasn’t electronics — it was getting unstuck long enough to hold the thing in my hands.
For context: these are small triangular tiles that are meant to connect edge-to-edge into a fabric, where each tile is a symbol (a MUX tile, a wire tile, a constant tile, an intersection tile…). The point isn’t “a PCB exists” — it’s that a set exists, like letters you can actually arrange into circuits.
This weekend’s move is simple: stop theorizing and make the first real, connectable patch. That means soldering the connectors onto the tiles (with the correct orientation), chaining a bunch together so it behaves like one long object, and then using that patch to build the first tiny piece of compute.
And yeah — the fear was real: I’ve shipped ideas before, but this one could’ve stayed forever in the “beautiful concept” stage. Another year of upgrades and re-designs and cost optimizations and “next revision will be the one,” while nothing ever clicks together.
But now it’s past that line. The alphabet is physical, the connectors make the tiles obviously one-sided (so orientation is unambiguous), and the patch is finally something you can assemble, pick up, rotate, and extend. Next step is the fun part: take just three MUX tiles and prove a D-flip-flop behaves like a D-flip-flop — not in a schematic, but in a little chunk of this fabric.
These 4 comments were also preserved verbatim from Brian Greenforest’s LinkedIn data export or the public post page. The image attached to one of these comments is localized below.
With rigid tiles and tall M/F “hermaphrodite” headers, removal is an over-constrained move:
• The center tile is mechanically locked by multiple neighbors.
• When you pull it, you apply torque and side-load.
• Halfway out, you’ve created partial disengagement on several sides at once → neighbors loosen, tilt, and you lose alignment.
• The taller the connector stack, the bigger the lever arm → more wobble, more accidental unplugs.
So: insertion can feel great, and removal can still be awful.
🍀 Brian Greenforest 🍀
Quick summary of the final design
• Goal: Use only 4 physical tokens (each can be flipped) to represent 7 tile types, and have the baseboard sense which type is placed—cheaply and reliably.
• Tokens: Passive pieces (can be PCBs): just color + markings for humans and copper patterns for sensing. No LEDs, no electronics, no connectors.
• Baseboard: A PCB with capacitive-sense electrodes at each triangle cell (or “cell region”).
• Sensing method: Each token face has a unique capacitive signature (simple copper “islands/pads”). When the token sits in a cell, the base electrodes measure the capacitance change and decode a 3-bit ID (8 possible faces = 4 tokens × 2 sides). Use 7 IDs, reserve 1 as invalid/empty.
• Mechanical alignment (your correction): Instead of a rod through the tile, the base uses plastic side separators/fins that:
• run between tiles (not through tiles),
• locate tile edges precisely,
• optionally pass through holes in the base PCB as structural posts,
• enforce repeatable seating height/position so cap-sense decoding is stable.
Comments added by Brian Greenforest on LinkedIn
These 4 comments were also preserved verbatim from Brian Greenforest’s LinkedIn data export or the public post page. The image attached to one of these comments is localized below.
Comment 1 · (2026-01-31 19:25:22 UTC)
View the LinkedIn post
Comment 2 · (2026-01-31 19:48:04 UTC)
View the LinkedIn post
Comment 3 · (2026-01-31 20:13:55 UTC)
View the LinkedIn post
Comment 4 · (2026-01-31 21:29:26 UTC)
View the LinkedIn post