For example, this project is a 2-layer PCB. Those are bog standard. With this small of a footprint it can be printed on a single surface and cut out. The schematic they posted keeps everything on a single surface for print. This is also an operation any fab can easily perform. If you order from China, even cheaper (even accounting for duty cost + S&H).
I'd be more concerned about the MCU and the components rather than the cost of a custom PCB.
[0]: https://github.com/user-attachments/assets/0f7962d5-38e1-4bd...
The really expensive stuff is through hole components and bulk components (relays, connectors, etc) because they need separate machines (wave soldering, large pick and place).
Placing small components is dirt cheap due to machines like SIPLACE SpeedStar. [0]
This is an stm32l432kc in the form of a yubikey nano.
https://shop.hak5.org/products/omg-cable
https://phreakboutique.com/blogs/tutorials/using-your-new-ev...
thats how the grift starts...
Nothing is safe, unfortunately!
You can absolutely exfiltrate data via the parallel port... that's why you attach printers or zip disks... it's just that it needs host participation.
8-bit AVR is an extremely clean, relatively simple instruction set that can be viably hand-coded. It's fairly straightforward to calculate the exact number of cycles your code will use, which is handy for applications requiring deterministic timing and for knowing worst-case execution time of interrupt handlers.
If the C3 instruction set is anything like the S3, I'm willing to bet it's not as straightforward.
Atmel also tends to do a better job of their documentation. I've lost count of how many confirmed errata I've reported to Espressif, and the time (and steam) I lost troubleshooting them in the first place.
I like the S3, and it's heaps powerful, but for small projects that don't need advanced peripherals like Wifi, DMA, etc. I can envision that AVR as being a fine choice.
Its not, not really anyway. The Xtensa LX7 in the S3 is very weird, though also quite fun to play with. The C3 is RISC-V
Doing asm on the AVR is beautiful, you can count clock cycles easily and then observe them on the scope.
I wrote a bit banging serial interface for an AVR once and had a mystery when I was testing it from a PC just with a basic echo. Every Nth character would be wrong. Was able to figure out a timing problem by counting clock cycles and found the bug in my code.
Was cool to see it align with what I was seeing on my oscilloscope.
With the right software, ESP32 can be incredibly low power. Like "months on 3xAA batteries" for watching a pin with the ultra low power subsystem and then occasionally waking up and making a HTTPS call over WiFi.
Bluetooth is lower energy than WiFi, but in your scenario the energy used for the radio is quite low anyway.