Monday, September 23, 2013

DIY Power without the Meter


That really didn’t take very long for a proof of concept!

Okay, so you can see the code in the background there and ANT+ is getting super easy for me especially since I re-use a lot of code. The real goal would be setting this up as a small updatable device so that other trainers can be configured. The best way to do this is to aggregate data from multiple peoples rides on different trainers with powermeters to develop quality regressions that can then be uploaded onto the device for people who don’t have powermeters but still want to train with power.


  1. Hi Keith; very interesting stuff! Is the idea that you would create a device that would allow you to select your trainer make/model, then it would transmit power values via ANT+ so folks could see and record the power on their heat unit? I'm not sure there's a huge market for that specifically, as I suspect most people would have a PC close and just that with with Golden Cheetah/TrainerRoad/PeriPedal, etc?

    1. Ya. Eventually I can use the custom calibration commands to upload the specific trainer curve to the device so it wouldn't be locked into a single device.

      There are potentially a couple of advantages. If I were to sense at the higher speed of the flywheel then I can more accurately gauge accelerations which means I can calculate power much more accurately.... impossible to do with the 4Hz update over ANT+. Means more accurate power compared to any other regression algorithm based power. Also record the data using your existing head unit for near automated uploads. Etc etc.

      I get your point though about precieved market based on existing solutoins. I am however seeing a lot of interest for something so simple especially considering Kurt Kinetic decided there was a market for one based on their equation. I'm not in marketing, and I made a piece of software LONG before TrainerRoad / Peripedal existed or the anyone was doing regression curves with Cheetah. I'm build things that are useful to me or others. If I get enough response maybe I'll look at building prototypes / beta's for people like my powermeter.

      As for market: I subscribe to the concept of lean startup. I don't believe in speculation. I believe in build, test, gauge response, pivot as needed.

  2. Hi Keith,

    Will be interesting to see how you fare with the accuracy of these regressed trainer curves.

    There are plenty of people using virtual power with TrainerRoad/GC etc, so there's certainly an argument that it's good enough.

    However, my experience with building a homebrew ergotrainer has been frustrating day to day variation between the estimated power from the curves, and the actual power readings from a PowerTap. This is still something I'm striving to improve (will be looking at adding roll down calibration this winter), but in the meantime I've had to work around it with closed loop control based on the actual power meter readings.

    I know you're focussed on accuracy for your crank based power meter, do you think there is any risk of 'damaging' your Accuity brand by releasing a virtual power product if it's not up to scratch accuracy wise? Perhaps power meter users are all smart enough to make the distinction between the two approaches, but it still struck me as a risk when I read these posts?

    One other thought sprung to mind when I read your previous post. Reading the speed data from an existing ANT+ sensor, and then calculating/broadcasting the power from your device. This would avoid any fiddly hardware mounting on the trainer, but you'd add a bit of lag, and lose the potential benefits of higher sample rate from the flywheel.

    Anyway, keep up the good work. Your blog is always an interesting read.


    1. Hi Jon,

      Seen you're blog before. Glad you're looking at getting back to it.

      That's a good question about brand. It illicited the response: "I have a brand identity?"

      I think there are a few markets, and based on the interest I've seen in this regression setup I think it's something people would like... given that someone else did good quality accurate regressions for the trainer they are using. While true power is king, having something that might not capture the peaks but comes within a few percent for short term (2 - 5 second) averages is quite valuable.

      So in terms of brand, I don't see a problem. I would encourage you to take a look at Eric Reis' book Lean Startup. It's explains how testing the market becomes much more important than protecting a brand image. If things go poorly with a product and you are forced to pivot, finding a new application, then that's an opportunity to re-invent a new brand or consider a different physical market test area. True tested market knowledge is much more important in a startup scene than building a brand -- however this holds much more true for "blue ocean" or completely new, no competitor technologies. Powermeters have their "gold standard" detractors already, and with more people entering the market with very very good products that's falling away into pure competition.