I am writing on crafting a microcomputer when I was in the first year of the university. When the Intel 8080 began to be sold at Akihabara, I wanted to build a microcomputer using the 8080. Although I had never developed a digital circuit, I bought a set of the AMD 9080, a second source of the 8080, and eight memory ICs (of 1 kbit), and I dared to wire the memory circuit using a soldering iron for analog circuits and with 1-mm vinyl cables and a universal circuit board. This wiring required much more efforts than I expected. It seemed to me a miracle when the microcomputer worked. But, anyway, I inputted a very short program using 16 address-input and 8 (8-bit) data-input toggle-switches, and ran the program. The result was displayed by the 8 LEDs. It took very much time to input a program by the toggle switches, so the microcomputer soon became unused after I ran only a few programs.
Until then, I had experiences to build hardwares, such as radios and stereo amplifiers, but I had almost never touched softwares. However, in contrast, after the above experience, I seldom touched hardware but often touched software. That was because the wiring of the microcomputer was so painful, and I decided not to build hardware. I can write that this decision led to myself of these years.
P.S. By the way, why did I arrayed toggle switches on the panel? This is because I believed it was the traditional design. At that time, a microcomputer called Altair 8800 was sold at a quite high price. It had a panel with arrayed switches (but they were not toggle switches). The microcomputer hobby club in the University of Tokyo displayed such machines at the Komaba Festival.
By the way, I wrote only on hardware here. But I also wrote on Software and hardware in "Research Topics".