Renesas R7FA8E1AFDCFB - an affordable ARM Cortex M85 microcontroller
The ARM Cortex-M85 core provides a wide range of accelerators, which makes the design ideally suited for machine learning. With the R7FA8E1AFDCFB, Renesas offers a low-priced microcontroller based on this core.
AI acceleration and internal memory
The R7FA8E1AFDCFB is available in an LQFN housing, which - given its generous pitch - can be soldered by hand or in a reflow process. If a smaller housing is preferred, the R7FA8E1AFDCFP has 100 instead of the 144 pins found on the larger version.
Renesas provides ample amounts of memory on the microcontroller. In particular, 1 MB of code flash is provided for remanent memory, while the 544 KB allotment of SRAM holds variables. Should regularly changing constant need to be stored, an additional 12 kB of high endurance data flash are available - the data sheet specifies a write endurance of 125000 cycles.
The above-mentioned AI acceleration is achieved mainly via the ARM Helium instruction set, which is included in the Cortex M85 core. The chip also has a DSP and a vector accelerator to speed up mathematical processes.
Regarding other peripherals, the R7FA8E1AFDCFB is well-equipped to handle external interfaces. For example, a CAN controller can interact with various automotive elements. In addition to that, USB peripherals are also available. Finally, the ADC and DAC peripherals work at 12-bit resolution.
RA8E1 vs RA8E2
During the introduction of these affordable Cortex M85 microcontrollers, Renesas demonstrated two slightly different families. Our part is a member of the RA8E1 family. These microcontrollers are somewhat cheaper than the RA8E2 – Renesas achieved this by omitting the LCD interface required for high-resolution displays. On the other hand, the RA8E1 has an ethernet controller which simplifies networking.
Conclusion
If an application needs an affordable Cortex M85 chip and can make do without high-resolution displays, the R7FA8E1AFDCFB is a prime choice. The combination of affordability, high-performance compute, and AI acceleration leaves them predestined for all kinds of predictive maintenance applications. The presence of an Ethernet controller simplifies data transfer.