...more... power!...

garmin vector pedal based power meter - (http://sites.garmin.com/vector/#power )

One of the latest new power meters seems to be suffering from product development woes. Some heard about it back in 2009, and it seemed promising enough. We have power measuring hubs, cranks, and chains (sort of), so why not pedals? Fast forward three years, and it's still not available.

The way I see it, there are three critical components to these things...




Firstly, there's the part whose physical response - to pedaling input - can be measured. In this case, it was shown as a strain gauge covered beam inside the spindle. It seems that there are reasonable odds that this was a marketing display that bears only a passing resemblance to the actual construction. Considering that the thing is sealed up, anyway, it's not quite clear why one needs an extra component and its associated mechanical coupling. Anyway, we might assume that the gauges are affixed to the inside of the spindle itself. Further, given the typical bushing/bearing layout in road pedals, they might be closer to the inboard (crank side) section of the spindle, where the amplitude of the moment may be less susceptible to changes in rider stance, rather than directly underfoot. These are all just examples of minor details (there are many more) - the bottom line is that I see little reason why the mechanical design would stall the development of this product. Compared to what's being done in robotics, there seems to be a fair amount of unused real estate in a road pedal, which would be suitable for instrumentation (without compromising the structural integrity of the system).

The second major component is the wireless communications system. There are ways to get this wrong, but at human pedaling speeds, it's not that hard to make a wireless communications kit that works (and stays powered up). Heck, if you go at it all ninja engineering like, you should be able to make things like this self powering, with truly negligible relative power cost to the rider. The actual bandwidth necessary - for sixty-times-per-minute updates to a power reading - is quite low. To stretch it further, consider that the pedal passes within a few centimeters of the front derailleur, at this rate. If the front derailleur is already electronic, adding short range communication with the pedals should be an efficient endeavor. But, I digress. This, also, isn't the hard part, and it's easy to see how practicality would just dictate the use of an existing protocol (such as ant plus).

In my opinion, the fussy part is data processing and control algorithms. The output of these strain gauges might require deconvolution (or separation) from differences in system (crank+pedal) stiffness that depend on crank position, for instance. Add to this the necessary ability to properly integrate various pedaling stances/modes that riders tend to switch between (spinning, sprinting, etc.), and you'll have a reasonably sophisticated pattern recognition system. While difficult, I have to say that this does seem perfectly doable, still (easy to say, from the comfort of my futon).

My old powertap (with the egg shaped brain) will, for obvious reasons, calibrate incorrectly if the wheel was being used during the calibration, and even (infrequently) drift a bit at the beginning of a ride if the outside temperature was very different from the bike storage temperature. My friends' newer ones seem to auto-calibrate and have the ability to gradually make corrections to account for the latter phenomenon. That's a nice example of some smart data processing built in.

Possible proof of this is an experience that one friend had, when the auto-calibration apparently did a brain fart. With an expectation of doing intervals at ~250W, the rider saw double digits (tens of watts), gradually increasing over time (to <200W), while doing intervals at consistent perceived effort. Upon restarting (and re-calibrating) the computer, everything was fine. The interesting thing to see was how the auto-correction function sensibly has a limit on its slope (and perhaps overall delta)... that is, if this has anything to do with what actually occurred, and how these things actually work! If so, then someone over at cycleops found a well generalized pattern with which to base their auto-calibration routine, because for the vast majority of the time, it just works.

Maybe, it's the task of finding such a pattern that is holding up these pedal power meters. But, then again, maybe it's just patent issues... Look and Polar seem to be listing an incredibly similar looking product, which doesn't seem to be getting as much press as the Garmin system. Anyone out there know anything about it?

Anyway, when are we going to have in-frame power measuring systems that use the conductivity of structural carbon fiber composites as the strain gauge (and structural health monitoring system)?