Searching for hardware-level sources on Setun, specifically the ternary core memory and logic elements.





Found the primary hardware descriptions. Now digging into the actual logic cell and drum specs.



Pulling the encyclopedia entry on the ferrite-diode cell itself plus the construction memoirs.






One more pass: looking for the primary circuit-level sources (the 1965 book and Brusentsov's threshold-logic paper) rather than summaries.




Found MSU's own "hardware implementation" page — fetching that.



The reason those articles skip it is that Setun's hardware is documented almost entirely in Russian primary sources. The English pieces all descend from the same two or three retellings, and Wikipedia compresses the whole machine into "ferrite core memory".

What is actually documented, briefly, so the links make sense:

- **Logic element.** A ring ferrite core with rectangular hysteresis plus a diode, driven as a current-pulse magnetic amplifier. It is a threshold element with positive and negative input weights: amp-turns sum, the core flips according to the sign of the sum, and that gives three states almost free. Clock 200 kHz.
- **One trit, two cores.** Gutenmakher's binary cell used every second core purely as a "compensation" core against parasitic signals and rejected up to 90 percent of toroids. Brusentsov found a drive scheme where all cores work, just not simultaneously, and turned that pair into a three-state cell. Three of the four binary combinations are used.
- **Physical build.** 3 mm cores, 52 turns wound by hand with a needle (every staff member did five cores each morning as "exercises"), then a power winding and control windings of 5 and 12 turns, diodes soldered on, cells sorted into matched ferrite/diode pairs. About 2000 cells in the machine at 3.50 roubles each. Roughly 3500 magnetic amplifiers, 4500 diodes, 7000 ferrites, ~300 transistors, a few dozen tubes.
- **Memory.** RAM 162 cells x 9 trits, split into 3 pages of 54, page-swapped with the drum, functioning as a cache. Drum 1944 or 3888 words x 9 trits, average page access 7500 µs. That is about 2.3 kbit of RAM and 28 or 56 kbit of drum.

Now the links, best first:

1. **https://ternarycomp.cs.msu.ru/hardware.html** — MSU's own "hardware implementation" page. Maslov's patents as PDFs: ternary D-trigger, threshold element of ternary logic, ternary circuit node and decoders, plus his collected "Троичная схемотехника" (2013) and a paper on one more way to build ternary devices. This is the closest thing to schematics you can get.
2. **https://ternarycomp.cs.msu.ru/Papers/Maslov_Sborn_2013.pdf** — the collected circuitry papers, 190 KB. Start here if you want to actually build something.
3. **https://www.computer-museum.ru/histussr/12-6.htm** — Brusentsov's own data sheet. Element base in one sentence, 200 kHz, 162x9 RAM, 1944/3888x9 drum, 7.5 ms page access, 2000 elements at 3.50 roubles, 2.5 kVA, 25 to 30 m², 27500 roubles.
4. **https://www.computer-museum.ru/histussr/setun_b.htm** — his SORUCOM 2006 paper. Register widths, 9-trit instructions, 18-trit S and R, 5-trit index and counter, page structure.
5. **https://www.computer-museum.ru/books/vt_face/7_brusencov.htm** — Malinovsky's memoir, the only place I found the winding counts, the 3 mm core, the morning winding routine, and his account of why Gutenmakher's cells threw away half the cores.
6. **https://dic.academic.ru/dic.nsf/bse/144408/** — Great Soviet Encyclopedia entry on the ferrite-diode cell, with the labelled schematic: core, diode, write winding, interrogation winding, output winding. Short but it is the canonical definition.
7. **https://www.computer-museum.ru/histussr/12-2.htm** — "Notes on ternary digital technique", part 2. Architecture rather than circuits: full address space, why every third address is a pair, the page exchange encoding, all 24 instructions.
8. **https://patents.google.com/patent/RU2394366C1/ru** — Maslov's 2008 threshold element patent, readable text with modern drawings. Useful because the Soviet originals are cited inside it.
9. **https://www.computer-museum.ru/english/setun.htm** — the English version of the 2006 paper, if you need to quote in English.
10. **https://bigenc.ru/c/setun-evm-a78047** — Great Russian Encyclopedia, the drum figure you can cite.

Two source warnings. The primary document is the 1965 book, Brusentsov, Maslov, Rozin, Tishulina, *Малая цифровая вычислительная машина «Сетунь»*, 145 pp, MSU press. It is not free online; the record is https://istina.msu.ru/publications/book/6637091/ and secondhand copies turn up. Everything above is downstream of it. And the English summaries garble numbers: inventfail says "162 words of 18 trits", which is wrong, the cells are 9 trits and 18 is the accumulator. Element counts around 3500/4500/7000 come from a 2026 Habr retelling, not from Brusentsov, so treat them as approximate until you see the book.