(I'm writing this like half a year later, so forgive me if I'm not explaining everything and probably make some mistakes)

In 2024 the small software company I'm working for had its 25-year anniversary. As a surprise for our two managing directors, we decided to build something special. Inspired by the hackaday-contest (https://hackaday.io/contest/195949-2024-business-card-contest) we decided to build a set of business cards (25 for each boss), which contains a non-contact voltage tester.

The goal was to create a card that is still usable as a business card, which meant it should not be (much) thicker than a normal credit card while providing contact information as a vcard via nfc and offering something useful for the receiver (we develop software for industrial/municipal energy management, so most of our customers know what a circuit is).
To achieve this, I had the idea to use a flex pcb for the circuitry glued to a thin 0.8mm pcb acting as the structural element.

Figuring out how this works

The first step after deciding on the project was to find out how a ncv detector actual works by looking at a cheap one I had lying around (this device also featured a simple stud finder and a metal detector). The ncv-part used a squiggly pcb trace as an antenna and a LM324 op amp (I've colored the hard to see traces).

"Reverse engineering is complete" (like big clive would say ;)). (Note: these are the values I later settled on, and not those on this pcb) 

So it seems to start with a simple low-pass filter. The first op-amp is used as a multiple feedback band-pass filter, which amplifies the 50Hz signal range. The antenna input is centered around half of the supply voltage to give a nice input signal. (I've played around with component values I had at hand until I had something similar to the existing device).

It took me a while to figure out what the second op-amp does. It turns out this is a rather clever hack to turn an ac-amplitude into a pwm signal which can be processed by a microcontroller. The slight offset on the inverting input results in longer on-times the higher the amplitude of the sinusoidal input is. (It can't produce a duty-cicle higher than 50% though).

The stud finder is based on a relaxation oscillator with a single pcb plane as the sensor. It basically measures the change of capacitance.

First prototype

Next was building a first prototype to verify the design and test some variation.  (this picture was taken after I already cannibalized it for parts ;))

This includes two different antenna designs: a squiggly trace, and a simple pad with vias on a corner. Both worked, but I decided to go with the corner design, as this allows pinpointing the live-wire/socket better.
For the microcontroller, I decided on using the CH32V003 in a QFN package, which is really cheap and more than capable for the task (I've never worked with on before so I had to learn something new. I guess I would describe the experience as OK, the documentation had me head scratching a few times, and the provided hal lib... well it's free so...). The leds where controlled via charlieplexing which left enough pins on the microcontroller free to test other things. (Note: this prototype also included a second feature for measuring battery voltages for a possible alternative business card, which was not developed further)
For the nfc chip I tested several variants of the ST25 lineup, but I couldn't notice a significant difference, so I selected the ST25STA02KB for the final product simply because it mentions "business cards" as possible applications in its datasheet ;).
If you want to design your own nfc antenna I recommend checking out this project from Rahim Aziz https://github.com/Raziz1/PCB_Business_Card
The nfc signal seems to be pretty forgiving, as even my small antenna works, even without the right compensation capacitors installed.
The copper tape was a test to shield the "antenna" for the stud detector from erroneous signals like your fingers...