and to complete the issue:Deanj wrote:I was playing around with settings on the powertap computer, I'd noticed in one of Ant's posts he was talking about data smoothing so I sent him a pm a few days ago asking which he had been using.toolonglegs wrote:I have no idea what you are talking aboutAnt. wrote: 5s display smoothing for me.
I done an awesome MIET through some very undulating terrain without looking at my meter at all, and it is -very- isopower'ish. I impressed myself
1. on the road, power output is highly variable, and is the first thing people notice when they see it on a bike computer for the first time (or when they inspect a power meter file)
2. there is a mathematical artefact of the way the Powertap calculates data, which exacerbates the variability in the second to second data. It isn't an accuracy problem (on average). The issue is caused by the fixed duration sampling of the torque data, which results in an effect known as "aliasing". It is more noticeable at certain cadences over others. I can explain more if anyone is interested.
3. to help reduce the "signal noise", it is useful to set the display to show a rolling average power, usually something like a 5-second rolling average tends to reduce the variability in the second to second data somewhat. over time, if you are using the meter to guide say an interval effort on the trainer, then you get pretty good at doing a running integral in your head and not concern yourself with the ups and down of the more instantaneous data but rather focus on the overall effort level.