Bigger is always better, right?

 A few months after receiving a box full of PCBs with hall sensor arrays on them, something peculiar started to happen - the boards themselves started to "curl" and refused to lie flat on the tabletop.

As the boards were made from fibreglass, they were difficult to reset - and being too forceful with them, trying to get them to bend in the opposite direction, caused more than one to snap cleanly in half! 

We needed something "stable" to mount the boards onto. Like MDF perhaps?

And then we got to thinking - instead of paying for relatively expensive "large format" circuit boards, would it be possible to make something similar, using a "arts and craft skills" (ok, copper tape) and mounting the components directly onto a more stable base (like some 6mm mdf board) ?

It turns out, the answer was "yes".
The cost of materials for producing a large 2ft x 2ft playing surface, with a grid of 16 x 16 sensors plummeted to "a few quid" for the MDF, the cost of half-a-roll of 3mm copper tape, 256 hall effect sensors (AH1815 or A3144 non-latching could be had for about six or seven pounds at the time) and a "controller board" (consisting of a PIC microcontroller and 2 x A2982 source drivers.

Yes, there was some assembly required, but the cost of materials for a DIY kit was substantially less than trying to get something so large manufactured and assembled.

It was around about this time that the idea of giving up on producing a "retail product" started to take hold, and an idea for an open source/DIY project started to formulate.
One major downside of this approach was the modified electronics.
I've long preferred PIC microcontrollers over AVR for anything going into a retail product - PICs are just "indestructible" - and AVRs feel flimsy in comparison.

You can reverse power a PIC, put massive input voltages onto the IO pins and generally abuse them in ways you really shouldn't - and they would still come through unscathed. On the other hand, I've had AVR chips "blow up" simply by having two i/o pins accidentally try to set their state at the same time.

In order to use a PIC microcontroller with such a massive array of sensors, we had to use "source drivers" to power an entire row of 16 sensors at a time, and a shift register on each, to free up the necessary i/o pins on the microcontroller to receive the responses.

Perhaps there was a "middle ground" - where we could provide ready-manufactured "controller boards" and a DIY kit for players to assemble their own playing surface? (one upshot of this idea would be players could make their grids to any size they liked! We preferred 1.5" squares, meaning a 16 x 16 board was 2ft square - but players would be free to make their squares bigger or smaller as they saw fit).

One downside to this approach is that debugging is a real headache!
Even  fully understanding how the whole thing worked (because I'd "invented" it) debugging when one of the sensors failed to "play nicely" wasn't much fun at all. Wiring up 16 rows and 16 columns of sensors and routing all these wires back to a PCB in the middle of the board was a messy and - at times - complicated thing to achieve!