LONDON Touting better energy efficiency and improved performance in industrial controller applications, Texas Instruments will start sampling in September a floating point digital signal controller that provides 300 million floating point instructions per second (MFLOPS) performance at 150 MHz.
The part, dubbed the 320F2833x, combines MCU peripherals with a 32-bit DSP that, the company says, will boost performance by 50 percent when compared with previous DSCs powered by a fixed point DSP at the same 150MHz clock rate.
TI says simplified software development common to floating-point processors and the performance boost allows solar power inverters to more efficiently convert energy from photovoltaic (PV) panels, offers better power efficiency and performance to variable speed alternating current (AC) drives and provides greater performance for automotive radar applications.
Jean-Marc Charpentier, business development manager for DSP products at Texas Instrument EMEA, says the device, the first in a range of DSCs with floating point processors, meets the increasing time to market needs of developers of industrial control gear that has been evident for many years in the consumer sector.
"Developers can save months by using the simpler software associated with floating point optimization processes," said Charpentier.
Software developers typically begin creating algorithms in a floating-point environment for validation, and then convert the code to run on fixed-point devices. By deploying the latest DSCs, they can now eliminate the weeks or months spent contending with scaling, saturation and adjustment of numerical resolution required in fixed-point implementations, he adds.
Some algorithms, such as a Fast Fourier Transform (FFT) used for complex calculations, will see a 200 percent improvement over an equivalent 32-bit fixed-point implementation. Overall system bandwidth is also increased with the addition of a six-channel direct memory access (DMA) controller that offloads the central processing unit (CPU) from servicing the on-chip analog-to-digital converter (ADC) along with user-configurable 16- or 32-bit external memory interface and high-speed serial ports.
The on-chip 12-bit, 16-channel ADC operates at 12.5 mega-samples/s, said to be the fastest on-chip ADC operation of any digital signal controller in the industry today. The F2833x controller series features up to 18 pulse width modulation (PWM) channels, six of which include TI’s high-resolution PWM (HRPWM) technology with 150 pico-seconds (ps) resolution. Communication interfaces include CAN, I2C, UART, SPI and TI’s Multi-channel Buffered Serial Ports (McBSP).
The F2833x devices will also be supported by TI’s TMS320C2000 Digital Motor Control and Digital Power Supply software libraries.
There will be two frequency ranges, 100 and 150MHz and three on-chip memory options ranging from 128kbyte flash and 52 kbyte RAM to 512kbyte flash and 68kbyte RAM. Prices will range from $13.50 to $19.95 in quantities for 1000 units