If you wish to bring the product to market, manufacturability is as important as design itself. It's no good if you have a perfect product that can't be manufactured, or a crappy one that's easy to manufacture.
Having this in mind, my goal for 1.2 version was to eliminate as much through hole parts as possible. Why? Most probably someone will have to solder all through hole parts by hand.
Unfortunately, since no one thought of making SMD Nixie Tubes, some things will have to remain through-hole, and will be soldered manually.
Touch sensor - TTP223
I spent quite a lot of time searching for a solution for touch detection.
First it was the piezo sensor, which was unreliable as a tap detection mechanism (although it meticulously detected beats from music), and then a friend suggested TTP223.
I bought 5 pcs of the red board with a big, inviting, touch circle, for 0.1$ each.
The chip reliably detects my hand at 8 mm distance, through a 6 mm wood panel. Enough said.
It works! However...
I learned the hard way that not all modules are equal, and that there is a certain flash mode that you must set (dio, qio), in order that your ESP8266 boots after flashing, and that you don't have a ruined day.
Also, when integrating ESP-12 modules, you should follow Espressif guidelines on PCB design, namely, you should not have copper pour below the module antenna. I left the pour by accident, and now the WiFi signal strength is weaker than it should be.
To call it a day, there is only one major change required.
Currently, the device is powered by a 5 V to 170 V DC-DC converter, sourced from eBay. There are no schematics available, so, if anyone wants to reproduce this project, (s)he would have to buy the module directly from eBay.
Furthermore, the dude making the modules can get bored and stop making them. It's not wise to rely on an volatile sources for any kind of volume production.
This calls for a custom designed 5 - 170 V SMPS. Probably flyback topology.