Overview
- Dr. Semiconductor produced and tested a 5×4 DIY DRAM array that stored data, with each cell measuring about 12 pF and holding charge for roughly 4 milliseconds.
- The cells were too small to wire directly, so he used micromanipulator probe tips to place voltage and read current on individual structures.
- He followed standard chipmaking steps in a class‑100 shed cleanroom, including oxide growth, photolithography, etching, spin‑on‑glass doping, and aluminum metallization.
- Results showed leakier capacitors than commercial DRAM, so the array needs refresh much faster than the roughly 64 milliseconds typical of off‑the‑shelf memory.
- He plans to stitch larger arrays and try a PC hookup next, with outlets calling the project a notable proof‑of‑concept during a period of high RAM prices rather than a practical replacement.