Below is what he wrote, but I do not attached photos right now as it takes so much time to browse in my phone (my photo gallery contains to much photos, and I am lazy now"
Enjoy his article below:
"At the risk of increasing my nerdiness image even further to scary proportions:
Although nowadays I spend most of my time coding, when I was younger it was dividied around 50/50 into coding and electronics, electronics initially from grade 3 and coding later on at grade 6, going in phases between the two. Electronics were more stimulating as I lacked internet access and hands-on experimentation brought faster results than banging my head against compiler errors for days.
My hobby culminated in building an 8-bit processor, entirely from 74-series ICs. This was the most interesting project I've ever worked on in my life, and it lasted years, many failed designs.
Technical details in thread to not spam. Pictures follow, sadly I have none of the fully assembled machine. It was 4 vertical PCBs on 2 backplanes and 1 control card extending out via a ribbon cable. 2 more cables connecting the backplanes. 100% manual point to point soldering.
The itch remains, so I wanted to enter the competition as well. You're posts have inspired me to pick it back up again, one needs some diversity in life. Wonderful posts in this channel! It's great to see the human side of bfx.
It was microcoded, 40-bits wide, and the instruction set (16 bits, 2 bit src register [4 registers], 2 bit destination, 4 bit ALU opcode/instruction options, and an 8-bit immediate value which could be used for jumps. No stack pointer, and as development ran late I had to disable the instruction set itself and code directly in microcode.
40 1s and 0s side by side on grid paper, then entering them by hand in 5 8-bit EEPROM chips, with tiny DIP switches and 8 LEDs for validating output, one byte at a time. It took ages, and was amazing. I plan on making another one, after so many years, it's a truly fascinating and enjoyable project to take on. These days I only code, lacking time and my old stable workspace.
Cards were 1) 8bit ALU from 2 4bit halves, w/ 2x 4bit digital comparison TTL ICs (less than, greater than, eq, etc), bufffered inputs via ALU A and B registers, and a tri-state output, all 3 linking to the same 8-bit data bus. So it took 3 clock cycles to do an operation (load reg ALU A, reg ALU B, then read buffer output into an arbitrary register). Then 2) a register card, 3) exec control card, 4) testing card w/ switches + LEDs tied to the data bus and all registers for debugging, 5) control card w/ 5x microcode 8bit EEPROM + 2x program code 8bit EEPROM. Some flip flops thrown in for control, I don't remember the details there. No stack, but it had a BRA(nch) command which could jump anywhere in RAM by clobbering the instruction pointer directly. Sadly only 8-bit addresses were supported, I never got around to adding paging, so 256 bytes out of ~4096 available or so, maybe 11k but I'd have to look up the ICs, they had a strange address width, SRAM.
Less than 5khz clock speed :) Never measured it, but could not have been more. Point-to-point wiring was too noisy for higher speeds, also with older-model 74-series ICs (LS, some HCT iirc, CMOS would have been better). Clock was 2 inverters in series with an RC (resistor + capacitor) circuit, very noisy and not square. It drew 1A at 5V, so 5W (a ton really), switching 7805s power supply."
As a bonus, let me attach his photo here. FYI, I do not know how old was he in the piture, but he looks smart, cute, and handsome guy <3
