Microcontroller suppliers are looking to increase the performance of devices through fast flash memory technology and the integration of floating point units.
STMicroelectronics has announced its first microcontrollers based on ARM’s Cortex-M3 processor with 90nm embedded flash memory.
The company is also working on next-generation 55nm embedded flash MCUs for the automotive designs.
Meanwhile Infineon and TSMC are jointly developing 65nm process technologies for embedded flash microcontrollers for the chip card and security markets as well as the automotive industry.
ST’s 90nm embedded flash process is already in production for smartcard and automotive ICs and is now being applied to the STM32 range of microcontrollers.
The company has included a memory accelerator to optimise memory performance for the Cortex-M3 processors.
Called Adaptive Real-Time (ART), the memory accelerator is designed to balance the operation of the
Cortex-M3 so that the processor can operate at up to 120MHz without waits states imposed by the flash memory.
At this frequency the processor runs at 150DMIPS and the accelerator implements an instruction pre-fetch queue and branch cache, enabling program execution from flash at up to 120MHz with zero wait states.
The aim is to increase the performance of the microcontroller and “remove the need to use a more expensive microcontroller or a companion DSP” in applications such as audio codec and video processing, according to ST.
The device draws 22.5mA of power at 120MHz (executing from flash memory, with ART accelerator enabled and all peripherals off).
The chips with 90nm embedded flash are sampling with products expected later this year.
ST has also moved to next generation embedded flash memory fabbed on a 55nm process at its facility in Crolles, France. The first 55nm flash products are still more than a year away.
The first MCUs to implement the 55nm eFlash technology are expected to target automotive applications, including engine management and transmission, car body controllers and safety/ADAS.
“Embedded flash is an absolutely critical technology in a chipmaker’s arsenal for automotive applications,” said Marco Monti, general manager of ST’s automotive electronics division.
According to ST, the power and performance requirements are outpacing that delivered by 90nm-based chips. The areas where this is being seen are functional safety, tougher emission standards or advanced driver assistance systems.
Toshiba Electronics Europe claims to be showing the first ARM Cortex-M3 microcontroller which works from a single 5V supply at Embedded World this week.
The TMPM380FY is the first in a series of 5V Cortex-M3 devices aimed at embedded designs for industrial and white goods applications.
On-chip peripherals will interface directly with motor drives and control IGBTs.
There is 256K of on-chip flash memory and 16K of on-chip RAM.
A three-channel 16-bit multipurpose timer (MPT) is able to combine three-phase PWM control with an ADC trigger and a protection circuit.
The timer can be used to deliver an externally triggered 16-bit PPG output for IGBT control. For motor drive application there is an 18-channel 12-bit ADC, a two-channel encoder input circuit, a watchdog timer and a real-time clock.
The encoder circuit is designed to offer three-phase sensor input and supports rotation direction and absolute position detection.
A two-channel synchronous serial bus interface supports SPI flame, SSI flame and Microwire flame formats.
Toshiba can supply the new microcontroller in a 100-pin QFP or 100-pin LQFP package.
Renesas Technology’s latest SH7216 flash microcontrollers are aimed squarely at industrial applications such as AC servos, building automation and general purpose networking.
The SH7216s are 200MHz 32-bit MCUs. To support these applications the 72 devices in the range include options for FPU (floating point processing unit) and Ethernet functions as well as various memory and package options.
Based on the firm’s SH-2A superscalar core, some devices also include 400MFLOPS single and double precision floating point unit.
“Due to the superscalar core which executes two instructions simultaneously, the device can achieve up to 400DMIPS,” said Renesas.
With the CPU running at 200MHz fast flash access time is a real benefit, and the firm’s MONOS (metal oxide nitride oxide silicon) embedded flash offers deterministic zero-wait operation with 10ns access time.
The core’s CPU registers are arranged in banks, with the 16 registers mirrored 15 times. A subroutine can use a second set of registers, leaving the current ones ready for use upon return from the subroutine.
“This means that the registers do not need to be popped onto and back from the stack, providing a quicker return from the subroutine,” said Renesas. “As a result, the response time to an IRQ is as quick as a simple branch instruction.”
The SH7216 family adopts the SH2A-FPU, which adds an FPU function into the SH-2A CPU core, and can perform double precision floating point operations at speeds up to 10 times faster than previous products.
The microcontrollers also include a CAN controller to enable a wide range of industrial communication protocols. The Ethernet interface includes a media access controller that conforms to the IEEE 802.3 standard, and can connect to a 10 or 100Mbps Ethernet.
Follow Richard Wilson on Twitter for news and opinion on everything from efficient C++ compilers to the lasting legacy of rock music 1971-1985.