What's the difference between these two, besides the power? Was this just a die revision, or was it complete redesign? Did the update rate issue with the 345 get fixed in the 346?
The issue with the 1600 Hz and 3200 Hz data rates is a function of the design. Depending on the select g-range and mode of operation (10-bit versus full-resolution), the least significant bit of the output data word will always be zero. This occurs in any mode where the resolution is 3.9 mg/LSB (10-bit, ±2 g, or any g-range in full-resolution) when using the 1600 Hz and 3200 Hz data rates. The most recently updated ADXL346 data sheet (PrE) discusses this on page 34 (http://www.analog.com/static/imported-files/data_sheets/ADXL346.pdf). A future revision of the ADXL345 data sheet will also discuss this behavior.
When using the part at 1600 Hz and 3200 Hz in the ±4 g, ±8 g, or ±16 g, 10-bit modes, the least significant bit will change according to the applied acceleration. Additionally, use of any combination of range and mode produces a least significant bit that changes according to the applied acceleration for any data rate of 800 Hz and less.
Retrieving data ...