Fixed- and Floating-Point Come Together in New DSP
Low-cost, low-power device offers Ethernet connectivity
Charles J. Murray, Senior Technical Editor -- Design News, April 29, 2009
A new digital signal processor (DSP) from Texas Instruments (TI) combines fixed- and floating-point performance for connected applications that need low cost and low power.
Known as the TMS320C6743, the new product is said to be the lowest cost, lowest power DSP to offer Ethernet connectivity. Texas Instruments says early users of the technology are applying it to such applications as test and measurement systems, oscilloscopes, electrocardiography, conferencing phones, programmable automation, power protection systems and audio foot pedals, among others.
"The good thing about this product is that it offers fixed- and floating-point," says John Dixon, low-power product line manager for TI. "From an applications standpoint, floating point is easier to work with, so you can get to market faster. But fixed point gives you more performance. So you can create a product, get it to market fast and then during the evolution of the product you can upgrade the software and use fixed point where you need the performance."
In the past, engineers who wanted floating-point DSPs typically chose from products oriented toward the audio market. Many of those products offered no USB or Ethernet connectivity, and low-power operation was typically not an option, Dixon says. In contrast, TI's new product operates at 60 mW in standby mode and 490 mW at 300 MHz. In standby, the TMS320C6743 offers days of battery life, Dixon says.
TI says the new product is upward-code-compatible with the company's C6000 fixed- and floating-point cores. The device, which costs $7.85 in quantities of 1,000, also offers 192 KB of on-chip memory.
"It's the lowest cost DSP for both fixed- and floating-point on the market today," Dixon says.
























