Originally Posted by RChung

I live in Berkeley so I'm pretty aware of the temperature swings we see here, and the climbs. There are very few climbs I can think of that are so continuous that you can't trigger a manual re-zero. The amount of variation you can see in a one-sided PM is greater than that -- plus, from a data quality point of view it's actually pretty handy to know that a re-zeroing occurred at a specific point. Relying on automatic temperature compensation "smooths" the zeroing so you can't tell exactly when it's happening.

I'm pretty familiar with the temperature issues. I did the original analysis of the Stages and PT and Quarq for Ray Maker, and identified both the temperature drift problems with Ray's Quarq and was the first to identify the cadence/power problems with the Stages. Ray's situation in terms of temperature change was extreme -- he kept his bike in his heated apartment, and the tests were done at close to zero celsius. He was supposed to re-zero each of the PMs a few minutes after leaving the apartment but he forgot, and didn't do that until he got out to the test venue. If you go back to the original data sets, you'll see that after that initial zeroing, the Quarq became much more stable, but the Stages continued to drift a tiny amount. That's what I mean when I say that it's sometimes handy to know for sure when a zeroing occurs rather than to allow for automatic compensation that sometimes lags and takes a couple of minutes.