A resolution of 0.1 knot is impressive Marc. Can you describe the specialized code you use to pull the signal from the noise?
Jeff, did you notice this thread is nearly 6 years old?
By "specialized" I meant specific to the implementation of a variometer. In our case, using the microcontroller onboard ADCs and digital filters, combined with simple linear boxcar filtering in the code was good enough to achieve that level of resolution on a desktop. In a glider we use the variometer in conjunction with a total energy compensator, which mixes altitude with airspeed to calculate total energy altitude, either before the variometer pressure sensor sees it (using a pneumatic compensating probe), or in the variometer code (using airspeed sensor data). When in turbulent air, a glider encounters horizontal gusts, causing transient fluctuations in airspeed which, in turn, show up as (sometimes large) fluctuations in variometer reading. This was addressed by implementing a "gust filter" as the final step in the variometer code. We tried a variety of filter implementations, but simply rate limiting the change in variometer indication was sufficient to damp down the transients, while not interfering with the all important task of finding rising air.