In industrial embedded systems, memory may need to preserve configuration or operating data without a backup battery and tolerate repeated updates. I2C FRAM offers a way to address those requirements using a familiar two-wire interface. The Fujitsu MB85RC16V is a 16-Kbit ferroelectric RAM device organized as 2,048 words by 8 bits. Its datasheet specifies high read/write endurance and writes that do not require the programming delay associated with Flash or EEPROM.
Those characteristics can simplify memory handling in a controller or other embedded design, but they do not remove the need to check capacity, bus timing, power conditions and data-retention requirements. Understanding the device’s behavior at the interface level is essential before connecting it to a system.
I2C FRAM fundamentals: capacity and write behavior
The MB85RC16V stores data in a nonvolatile FRAM array. Unlike SRAM, it retains data without a data backup battery. Its 16-Kbit capacity is arranged as 2,048 bytes, addressed through an 11-bit memory address.
The specified read/write endurance is at least 1012 operations per byte. The datasheet notes that the total number of reads and writes defines the minimum endurance because FRAM uses a destructive-readout mechanism. This figure should therefore be considered in relation to how the application accesses memory, rather than treated as a write-only limit.
Writes can be performed in byte units, and the device does not require a programming wait time after a write stop condition. It also does not require the acknowledge-polling sequence used to check whether an EEPROM write has completed. The master can proceed without waiting for a separate internal programming cycle, subject to normal I2C communication and system-level timing.
How the two-wire interface addresses the memory
The device operates as an I2C slave. A master initiates communication and supplies the clock on SCL; SDA carries bidirectional address and data information. Because SDA is an open-drain output, the bus requires external pull-up resistors. Multiple devices can share the bus, although the MB85RC16V does not use external pins to select its slave address.
After a start condition, the master sends an 8-bit device address word. Its upper four bits are the fixed device type code, 1010. The next three bits represent the upper portion of the memory address—not externally configurable slave-address bits—and the final bit selects read or write. The lower eight memory-address bits are transferred separately for byte write, page write and random-read operations.
Data moves in 8-bit units. The receiving device acknowledges each successfully transferred byte by pulling SDA low on the ninth clock. For a read, the master supplies the final NACK before ending the transaction. Start and stop conditions are exceptions to the normal SDA timing rule: SDA changes while SCL is high to signal the beginning or end of a transaction.
Write and read modes to consider
Byte and page writes
A byte write sends the device address in write mode, the lower memory address and one byte of data, then ends with a stop condition. A page write sends additional data bytes within the same transaction. Addressing rolls over to 000H at the end of the 2-Kbyte memory space; sending more than the capacity can overwrite earlier data from the same operation. Firmware should therefore control transfer length and account for the array boundary.
Current, random and sequential reads
A current-address read uses the address retained in the device’s memory address buffer and advances to the next address. The buffer’s current address is undefined immediately after power-on, so this mode should not be used until the address state has been established. A random read first sets the target address, then uses a repeated start and a read-mode device address. The upper memory-address bits in the address-setting and read phases must match.
Sequential reads continue across bytes and roll over to 000H at the end of the memory array. This can be useful when software needs a contiguous block, but the rollover behavior should be included in buffer and boundary handling.
Electrical limits and integration checks
The specified supply range is 3.0 V to 5.5 V, and the operating ambient temperature range is −40 °C to +85 °C. Maximum I2C clock frequency depends on the mode and supply conditions: standard mode is up to 100 kHz, fast mode up to 400 kHz, and fast mode plus up to 1 MHz, with fast mode plus specified for 4.5 V to 5.5 V. Confirm that the controller, bus loading and pull-up network meet the relevant timing requirements.
Data-retention specifications vary with temperature: at least 10 years at +85 °C, 95 years at +55 °C, and more than 200 years at +35 °C. These figures are not interchangeable; select the applicable condition when assessing the required retention period.
The WP pin protects the entire memory array when held high; reading remains enabled. A low level permits writing, and the pin is internally pulled down, so an open WP pin is recognized as write-enabled. Do not change WP during an active communication period. Designs should also follow the specified power-on and power-off sequence, since the datasheet warns that memory data cannot be guaranteed if read, write or power sequencing conditions are not met.
Practical implications for automation designs
The MB85RC16V combines modest storage capacity with a two-wire interface, byte-level writes and a high specified endurance. These features may suit embedded designs that need frequent nonvolatile updates, provided the required data volume and retention conditions fit the device’s specifications.
Before selecting it, engineers should verify address handling, rollover behavior, bus timing, pull-ups, write-protect wiring and power sequencing against the complete datasheet and the application’s requirements. Treating those details as part of the memory design—not just firmware implementation—helps ensure predictable integration into an industrial control system.