Here’s the short answer most pages won’t give you: a standard 93Cxx serial EEPROM has no read-protection fuse and no lock bit. A 93C46, 93C56, 93C66, 93C76 or 93C86 will hand over its contents to anyone who clocks the Microwire bus correctly. So when someone asks for a 93Cxx EEPROM unlock, the real problem is almost always one of four other things: wrong organization, the host MCU fighting you on the bus, data that’s checksummed or scrambled by the firmware that wrote it, or a tired cell array that reads differently every time.
Figuring out which of those four you’re dealing with is the whole job. Below is how we approach it and what changes the answer.
What people actually mean by “93Cxx EEPROM unlock”
Over the years, four distinct requests have landed in our inbox under that phrase:
- “I can’t get a clean dump.” The reads come back inconsistent, shifted, or full of 0xFF. This is a bus, organization, or supply issue, not protection.
- “I have the dump but it makes no sense.” The bytes are there but the values are XORed, byte-swapped, or protected by a checksum the host MCU verifies. The memory is open; the format is closed.
- “The EEPROM is fine but the MCU next to it is locked.” Calibration values live in the 93Cxx, the program lives in a protected microcontroller, and you need both. That’s two separate jobs.
- “The chip is physically dead or degraded.” Decades of retention loss, a cracked bond wire, or a board that got hot. This is data recovery, not unlocking.
The word “unlock” carries over from microcontroller work, where security fuses and read-out protection genuinely exist — AVR lock bits, STM32 RDP levels, the MSP430 JTAG fuse. Serial EEPROM in this family never had that. The only write-inhibit mechanisms on a 93Cxx are the volatile EWEN/EWDS latch, which resets on power-down, and on some vendors’ parts a hardware write-control pin. Neither one stops a read.
The 93Cxx family, and where the read goes wrong
Density and organization are the two facts you need before you touch the part. Getting either wrong produces a dump that looks plausible and is useless.
| Part | Density | Organization | What usually complicates the read |
|---|---|---|---|
| 93C46 | 1 Kbit | 128 × 8 or 64 × 16 | Most common part, most common ORG mistake; often written by an 8-bit host |
| 93C56 | 2 Kbit | 256 × 8 or 128 × 16 | Address field width differs from 93C46 — wrong instruction framing returns shifted data |