Agree with your reasoning about error. Thinking now that it might be useful to measure both tangential and radial in all four positions to check for symmetry as well and to also have numbers that are comparable between tangential and radial forces.

LikeLike

]]>FWIW, I simply held the opposite crank steady and horizontal (by eye) to make my measurements. A few degrees of error will have essentially no effect on the tangential reading but could yield a few % error in the radial reading (since cos(small angle) = 1, sin(small angle) = small angle). Along the same line, the radial component I observed on the R would require to about 6 deg of misalignment in the R strain gauge (or my estimate of horizontal (unlikely, I think)).

Thanks!

LikeLike

]]>1) My assumption for the Pioneer was that pedal and crank weight (in the case of the Vectors just the pedal weight – especially as the pedal body can now be swapped to Shimano Ultegra) would be “taken care” of by the zero offset calibration and the force displayed is any additional force working on the pedal. Will think about this…

2) Thanks, good to have some comparison data.

3) Yes, the radial force is usually close to zero in horizontal crank position. The radial values I gave were measured independently from the tangential ones and in (nearly) vertical crank position. I had nothing to measure and lock the position of the cranks, so I found it more convenient to read the maximum value while slowly moving the cranks fore and aft, and applying this method, I reasoned I should measure tangential in horizontal and radial in vertical crank position. Not sure what to make of your left-right difference…

LikeLike

]]>1) It is worth noting that the weight of the pedal/crank is included in the hanging weight. To account for this, record the response for several weights and use a linear fit to extract the response (slope) and pedal/crank contribution (intercept).

2) On my Pioneer L/R system, I found nearly identical results to yours (L is about 2% low and R is about 2% high) for the tangential response.

3) Pioneer radial force should have been near 0N for each case (L and R). It is odd that yours were -102N. In my own case, I found the L radial response was consistent with 0 N (< +/- 3N) but the R radial response appeared to register about 10% of the (purely) tangential load. I'm not sure what to make of this but it could be a limitation of the more complicated mechanics of the R crank/spider yielding an overlap in the radial and tangential strain responses at the strain gauges.

LikeLike

]]>LikeLike

]]>Wheel rotational accelerations are very small, especially at steady state, although the oscillations about the mean are a bit larger in the initial seconds of a hard acceleration (but that’s the case no matter the chainring shape). I went through this in a couple of my own blog items:

http://alex-cycle.blogspot.com.au/2015/01/the-sin-of-crank-velocity.html

http://alex-cycle.blogspot.com.au/2015/01/accelerating-sins-crank-velocity.html

SRM Powercontrols from PCV to PC7 enabled 2Hz recording rates. PCIV before that enabled 10Hz recording rate, although (as with 2Hz) it would only update power values once a crank revolution was completed. IOW the power file would step up/down more frequently than 1Hz, but each step would last X tenths of a second, with X being the whole number of tenths of a second it took to complete a pedal stroke.

LikeLike

]]>Concerning hub-based meters … it’s often mentioned that they work with oval rings which implies higher sampling of wheel rotation than the sampling of crank rotations of a crank-based meter, but your information reminded me that I haven’t really thought that through yet. A simple calculation would give 250 wheel rpm at 30km/h independent of crank rotations. Assuming a magnet switch triggers at least once per wheel rotation for wheel rpm calculation (even when their output is at 1Hz), that’s just 2 to 3 triggers per crank rotation at cadence 125 or 83 respectively … which seems too small to deal with an oval ring or with the brief bursts of just 1 or 2 crank rotations as in this case. Would you know if they have more “switches” or additional accelerometers inside?

Thanks for the suggestion of using standing starts for measurements. Will attempt sometime … but that also reminds me that at some point I will need a pedaling robot for evaluation … which would in turn create the problem of calibrating that robot. ðŸ˜‰

LikeLike

]]>LikeLike

]]>Providing data at 1Hz in itself creates an issue for most meters, as power is really calculated on an event basis, i.e. crank rotation duration which has variable timing, typically less than 1 second since cadences are often > 60rpm.

It would be better IMO for crank and pedal meters to provide a time stamp for each cadence tick, and the average torque for that revolution. Then power can be displayed based on an average over X number of crank revolutions (SRM does this when calculating power but then converts to data stream at user’s chosen reporting frequency which can be 2Hz or less), rather than over Y number of seconds (be it 1 or more) .

Of course meters that rely on accelerometers for cadence have to infer when the crank passes a given point. That in itself is a source of variable aliasing error, and is quite possibly a reason for lags in power data as well.

One test of a power meter’s abilities over short durations is plot the average effective pedal force v circumference pedal velocity for each data point from a maximal effort track type standing start. The initial 5-6 seconds of data should plot a straight line as pedal force and pedal speed are inversely related during maximal effort before neuromuscular fatigue sets in.

Not all meters have the temporal resolution to do this, or they are unable to reliably show such a plot.

LikeLike

]]>This paragraph may need editing?

“So, my recommendation would be, to either try a hub-based power meter (although I admit Iâ€™ve never used one before and have no idea how theyâ€™d perform under such conditions) or go with the P1 (which seems to provide honest data, an impression that also somewhat aligns with their claim of using a large number of sensors (8) and enough computation power).”

Isn’t the P1 a hub based PM?

Did you mean ‘or go with the Pioneer’ instead of ‘or go with the P1’?

LikeLike

]]>