With Garmin releasing it’s stranglehold on “Garmin Cycling Dynamics” to bring “Cycling Dynamics” to version 5.1 of the ANT+ power protocol means a shift in the usefulness from “What is this pretty picture for… oh… nothing” to “Maybe this could be of some use?”.
Essentially there are 14 pieces of information being transmitted for the new Assioma firmware and Garmin Vectors V1 through V3. Those are power (1), L/R balance (1), left and right pedal smoothness (2), left and right torque effectiveness (2), and four “when / what angle” for each side (8) in cycling dynamics. These encompass at least 4 fields not taking into account averaging, plus a whole other cycling dynamics page. YIKES! Information overload!
The holy grail is high speed, but considering how long FE-C and this took to happen I won’t count on it yet. Pioneer sort of tried with low return on investment, and are now in that phase where they are realizing “oh we need other people’s head units to make this valuable”. They’ve missed the other next step. Analysis.
Every time I edited this video I tried to keep in my rant about standards short, but it was just to divergent so it got cut and I’ll just put it here. ANT+ isn’t dead because BLE is too expensive to participate in making a standard and is mainly a money hole. Giant companies like Apple don’t make sensor standards, they use them when there is a mass market opportunity and will develop their own when needed but they don’t really need. Cycling is not, in a meaningful way, on the radar of any fitness company that is in the real mainstream. What I mean is if their main customer doesn’t know bikes can have third party sensors, then they won’t bother.
I’m not knocking it. I’m just stating that very few mainstream devices like the Apple watch and Fitbit’s have any meaningful cycling connectivity. I tend to put it that smart watches are smart first, an activity monitor second, and a distance 176th they are a sports watch. Whereas Garmin / Polar / Suunto are sports watches first and the other things later — though some of those models have me wondering.
So BLE is the gateway to the mainstream, but you need to be a full member at about 35k USD/year to participate at that level. Now recently they went back to their discount model of $7500 if you make < 100M USD… (when they revised down from 50k USD/year. But to list, not to certify and test, just to list is 8000 USD and if you do something non standard it’s about 30k – 60k to certify a device. If you pay that 7.5k to 35k USD you get a discount to 4k for listing but nothing for those testing. Most people skirt those tests by referencing something else that’s close enough. All this from a group that literally has a document that spells out legal action if you don’t list (which is they kick you out and keep your money if you paid any. It’s like 5 pages long). On top of this, I’ve never professionally had them respond to any questions at any job. Black hole, no response.
So the march goes on. We wait, companies plod through uphill battles be it with ANT or BLE (ha!). The smart ones will realize that first to the standard is better than years of proprietary. Proprietary means you have to support so many levels. It’s a death sentence for startups. Silo your sports sensor company and might as well close it tomorrow if you are consumer facing. We’ve seen so many companies in the running footpod market with new metrics die in their silos. Was it because of the silo we can’t know, but it certainly didn’t help.
So that leads me to doing some, hopefully, clever math. The video explains it better but suffice to say it’s a disassembly / reassembly chain.
- Break power to Power_L and Power_R using balance
- Use cadence in RPM to figure out rad/s
- Use rad/s to figure out average torque
- Use pedal smoothness metrics to figure out peak torque.
- Use cycling dynamics to figure out when 0, 50% rising points are and how much compared to that peak, and the same for 50% and 0% falling.
- Assume peak torque halfway between 50% (peak power phase).
- Assume peak negative torque halfway between last 0% and the next stroke.
- Use torque effectiveness metric to figure out how large that negative peak is by some piece wise integration, and then solving for a triangle in the negative.
PHEW? Got all that?
Python code makes an animation of this in a polar plot so you can see what is going on here.
I do want to make mention that until a brave soul poked a bear the Jan 2019 Power document version 5.1 was yet to be released. But after that poke, 8 months late from it’s last revision, they released it. So as of July 23, regardless of what it says in the document, is when it became an open standard.