For SD Card project shown: Note that the ATmega328P-PU is being powered at 3.3V by the SD Card regulator; therefore signal conditioning (resistor network, FET or transistor voltage translation, or specialized inter grated circuits) is required if the logging signal is higher than 3.3V. FLAT-duino can work from 5V to 3.3V @16MHz in testing here at SofKinetics. Just note that you were reminded that 3.3V is outside the published uC specification for 16MHz but within the 12MHz range.
Want to shrink your homebrew Arduino project without using surface mount components? It does not get much more raw than the flatduino... a naked ATmel328P-PU chip with its little legs all spread out, soldered to a piece of copper project board, with the headers facing horizontal rather than vertical. The final module is approximately 1.35" wide x 1.90" long x 0.35" deep.
Last month, I posted my "Under $8 Serial Data Logger" project: https://www.instructables.com/id/Under-8-Arduino-Serial-Data-Logger-Record-to-SD/ which was built on perfboard. While thinking about how to shrink the project, I decided to piggy-back the Arduino on the SD Card breakout board. My first physical layout simply had the chip soldered to the board with the headers in a more traditional vertical configuration. Then I thought about flattening the concept and I arrived at the flatduino format.
I positioned the crystal sideways on top of the lower-numbered pin header. A strip of adhesive backed copper tape was placed on the top-center of the uC running up the length of the chip. This was grounded by soldering a wire to Gnd at one end. The top of the copper was used as the ground connection for the 16MHz crystal bypass capacitors and also for connecting a wire to chip pin numbers 8 and 22.
The SD Card breakout board contains the 5V to 3.3V LDO regulator, bypass and bulk capacitors, so these are not replicated on the flatduino. In other configuration, anyone duplicating this project should use traditional bypass and bulk capacitors.
I used my UNO with an ADAfruit Arduino ISP ZIF shield to program the ATmega328P.
Testing proved the implementation and unusual physical design worked perfectly. It was then coated with a thin coat of 2-part epoxy to provide a rigid and permanent module. I used an A + B product that takes about an hour to fully cure and I have used this successfully on many projects; however, I would suggest care be used in the epoxy selection process as some fast-cure products become very hot when mixed. As I have no personal experience with these "minute" products, I cannot attest to concerns regarding electronic components.