In the concurrent P2M thread, Hamish Ferguson shared some of the
average results he got from a ride he recently did with his P2M and PT
both mounted. He kindly shared the data with me; I've shown him the
following analysis and he suggested that I re-post it to the group.
It's easy to compare overall average power readings from two devices
but we get into power meters because we want them to be both accurate
and precise. And, features are nice but the most important
characteristic of a power meter is the quality of the data it provides
-- so the real standard isn't "how close are they on average" but
"under which conditions do they differ, by how much, and why?"
Accordingly, I've been using VE as a magnifying glass to peer into the
data and pinpoint where they differ. I can't always tell the "why" but
I can use VE to tell the "when." So that's what I've done here. In
that kind of situation, even if we don't know what the wind, brake
usage, mass, rolling or aero drag were, we know that they were the
same for both the PT and the P2M. So we just choose reasonable values
of Crr and CdA and presume the wind was zero and examine the VE plots.
They'll be off from the actual profile but that doesn't matter for
this type of analysis.
But first, here are some quick observations:
1. The speed recording for the P2M is spikier than for the PT. This
will turn out to be of possible interest because when speed and
cadence are well-measured, we can use them to tell which chainring and
cog were being used (Hamish told me the size of his chainring and the
cogs on his cassette). As we'll see below, the speed spikiness means
that the exact gear combinations aren't as clear as they are for (as
an example) the SRM.
2. There was one data drop-out (lasting 4 secs) in the P2M data file.
I obviously don't know how often that happens but (fortunately) the
drop-out occurred only a few minutes into the file so I snipped it
out, re-synched the PT and P2M files from that point and what you're
seeing starts at that point.
3. Compared to PT, the distance (and thus, speed) of the P2M appeared
about 2% low. I scaled the speed up by 1.02 and then recalculated
distance.
OK. In the attached plot the PT is in black and the P2M is in
red.
Distance in Km is on the x-axis for the entire plot.
The top panel shows PT wattage. I don't show the P2M wattage because
it "overlays" the PT points pretty closely -- this is an example of
what I wrote above, that they can look quite close on average so what
we want is a way to magnify any differences.
The second panel shows the VE profiles for the PT and P2M data. The
P2M VE profile is based on nominal power, i.e., I haven't made any
correction for drive train losses. When I've done this sort of thing
for other crank-based PMs I tend to see the PT's VE profile below the
crank-based profile. Obviously, in this case the lines are the other
way around. Hamish has not yet had a chance to do a static calibration
check on the PT and the P2M so we don't know that the PT is "correct"
but it's interesting to see that relationship.
The third panel shows the difference in VE (i.e., the difference
between the black and red lines in the second panel). Importantly, you
can see that the difference isn't "smooth" -- it's pretty flat (there
was an initial little blip at the very beginning which isn't
important) until you crest the first climb. Then the difference starts
to increase between roughly 5K and 10K into the ride. Then when you do
those two little efforts the difference spikes but settles down again,
and stays constant until the end of the ride. The takeaway message is
that the difference between the two power meters isn't constant, nor
is it a constant percentage -- the difference is more complex.
There's a clue to a possible explanation in the bottom panel, which
shows the gear ratios as calculated from the observed speed and
cadence (in this case, I've colored them red since they're based on
the P2M's speed and cadence rather the PT's. If you "line up" the
parts of the third panel where the VE profiles are growing farther
apart, you'll see that there are a couple of little discontinuities,
and more noticeably, during the section from 5K up to about 10K Hamish
was using one particular gear combination.
These observations are preliminary and, since I've only seen these two
files, perhaps premature but if this P2M and this PT are
representative, we can already see some avenues for further testing
.
, costassero poco sti giocattolini!!!
