Inside a 1970s MMI 5300 PROM Chip with Microscopic Fuses

Looking inside a 1970s PROM chip that stores data in microscopic fuses (2019)

Inside a 1970s MMI 5300 PROM Chip with Microscopic Fuses

I examined a 1970s MMI 5300 PROM chip to reveal how it stores 1024 bits using tiny fuses. Unlike modern memory, this chip uses NPN transistors and Nichrome fuses that melt to program data permanently. By removing the ceramic package, I visualized the silicon die, the address decoding logic, and the microscopic cracks in blown fuses that represent binary zeros.

I expected a blown fuse would vaporize entirely, but instead a blown fuse contains a tiny crack roughly 700 nm wide.
  1. adrian_b

    > I suspect that diode-transistor logic was used instead of TTL for the output multiplexer because the simpler DTL circuit worked better with the high programming voltages. The DTL circuitry uses very large resistors biased to the programming voltage.

    Yes, it was necessary to use DTL instead of TTL, because a TTL gate is equivalent with a DTL gate where the diodes are base-emitter diodes of bipolar junction transistors, but the base-emitter junctions of BJTs have a very low breakdown voltage, higher than the +5 V power supply of TTL ICs, but lower than the programming voltage required by PROMs (typically +12 V).

    Therefore TTL gates would have been destroyed by the programming voltage.

    For some years, after the older DTL logic circuits were mostly replaced by faster TTL logic circuits, which had a +5 V power supply, there remained available some DTL ICs that were intended for industrial applications in noisy environments, where speed was less important than immunity to noise, so those DTL ICs used higher supply voltages, e.g. +15 V, for which TTL circuits could not be made.

    Eventually, those high-voltage DTL circuits were replaced by the RCA CD4000 CMOS logic circuit series, which could use the same high supply voltages while consuming much less power.

    Decades later, the cost of regulated power supplies decreased by orders of magnitude, so it became possible to ensure much less conducted noise for the electronic devices used in industrial environments, which mostly eliminated […]

  2. laughing_man

    Yeah... I remember designing these into circuit boards. They were much cheaper than EPROMs, so we did development with EPROMs and production with PROMs.

    And then we desoldered and replaced all the PROMs when we found a problem.

  3. ggm

    I remember knowing about these at uni and then learning about reprogrammable UV erased PROM and thinking the same magic faries who do the link fusing probably had tiny soldering irons to fix them in the reprogrammable ones (we said if you let the magic fairy smoke escape from your circuit board it wasn't surprising it stopped working...)

More from this day

2026-08-04