posted by Bryon Moyer
Accelerometers are used for a wide variety of applications (which we’ll look at more specifically soon). Particularly demanding are automotive applications, not least because of the harsh conditions and huge amount of noise that they must tolerate. That noise can be electrical or simply “ambient vibration” that is not of interest.
Electrically, differential signaling is often used to reject common-mode noise. But one paper at ISSCC took the concept all the way back to the proof mass: a team from Robert Bosch split the proof mass, working then with what are nominally two identical half-masses. This sets up a differential signal flow from the get-go.
Of course, the masses aren’t going to be exactly the same; various differences are averaged out by swapping back and forth (to oversimplify). Chopping is also used to boost near-DC noise up away from the frequencies that matter, although the process does result in some noise being moved into the way instead of out of the way. In order to minimize the effect of this, they used a pseudo-random signal for chopping so that the energy of this noise is, to use their word, “smeared” across the spectrum, rendering it largely impotent.
You can find circuit details and results in the ISSCC proceedings, paper 22.1
posted by Kevin Morris
Altera just announced that they're partnering with Intel to produce FPGAs based on Intel's 14nm Tri-Gate process. This has the potential to give Altera a big lead in the node-after-next war with rival Xilinx. Intel has a well-established leadership position in FinFET technology (which they call Tri-Gate) - a 3D transistor fabrication technique that has much lower power consumption and better performance than traditional planar CMOS transistors. FinFETs give probably an extra process node worth of benefits to FPGAs, so a 14nm FinFET-based FPGA will probably be 2 process nodes better (in terms of performance and power consumption) than the 20nm planar devices that both Xilinx and Altera are already reportedly developing, and 3 process nodes better than the current state-of-the-art 28nm FPGAs both companies are producing with TSMC.
We talked with Altera CEO John Daane about the deal, and Daane says Altera will be exclusive with Intel among "Major FPGA vendors." That means Xilinx will not be working with Intel 14nm Tri-Gate, but likely Achronix and/or Tabula will. This leaves Xilinx without a known partner for Fin-FET FPGAs, although they are rumored to be working with TSMC on the technology. However, Intel is believed to have a significant lead both in 3D transistors and in process geometry at this point, so this deal could be a major coup for Altera.
posted by Jim Turley
Microchip has just rolled out BodyCom, a new way to do wireless nwtworking using your own body as the antenna. This is a body-area network, meaning it connects things you're wearing -- or at least, touching. That makes it useful for sensors and displays, for example, but not for beaming music to a remote speaker. Microschip's got all the documentation, software, and development kits here. http://www.microchip.com/pagehandler/en-us/technology/embeddedsecurity/technology/bodycom.html.