This project implements "virtual power" for bicycle turbo trainers where the trainer or the attached bike has an ANT+ speed sensor. The calculated power is broadcasted as such on ANT+ so that any head unit or app will see it as a power meter. Currently supported trainers are the Bike Technologies Advanced Training System (BT-ATS), the Kurt Kinetic range of fluid trainers and the Tacx Blue Motion.

It is easy to add a new trainer - just subclass AbstractPowerCalculator and implement the method power_from_speed(revs_per_sec). If your trainer is not there, please add it and submit a pull request.

Many turbos, such as the Kurt Kinetic or Bike Technologies Advanced Training System (BT-ATS) have well-known and consistent power curves which make it possible to calculate the power from speed. This concept is used in several applications (Zwift, Golden Cheetah, TrainerRoad, Sufferfest) to allow the user to behave as if they had a power meter.

This is great, up to a point. One issue is that the calculated power is not shown on the user's bike computer, but only in the app (except for TrainerRoad). Another issue is that not all apps support all trainers (e.g. Zwift doesn't support the BT-ATS). Even when a turbo is supported, the calculated power may differ from a rider's power meter by as much as 5-10% - this is highly significant when doing intervals in the region of threshold or VO2max.

This project is designed to be run on a Raspberry Pi with an ANT+ stick plugged in, and will broadcast the calculated power over ANT+ so that bike computers and apps will see it as a regular ANT+ power meter. The project is designed to be open-source and highly configurable - for example, the BT-ATS not only allows for a correction factor so that it matches, say, a power2max on another bike, but also it can read from a cheap weather sensor and correct for air density. With these corrections, the calculated power is within 1 watt of a power2max power meter for durations of 10 seconds or more.

If you're not bothered about the air density correction then you don't even need a Raspberry Pi - any computer running Linux or Windows will do - all you need is a dedicated ANT+ USB stick.

The plan is to enhance the project further so that accelerations/decelerations can be factored in, increasing the accuracy in the sub-10 second range and making the project useful for calculating sprint power.

It may also be possible in future to do run-down calibrations to improve the accuracy of wheel-driven trainers such as the Kurt Kinetic, and such calibrations may also be able to calibrate for air density so that even the BME280 sensor is not needed.

This project can be built very cheaply. A sample bill of materials is as follows:

So around $40 for a functioning ANT+ power meter (a little more with the air density sensor), and under $15 for the USB hub/ethernet adaptor to be able to build it :-)

1. Log into the Pi

2. Run sudo raspi-config and tell the Pi to boot to console only (to save memory) and enable I2C in "Interfacing options" if you plan to use a BME280 sensor for air density correction.

3. You can clone the repo on the Pi itself (3a), or simply download the zip file and extract it to the microSD card from a PC (3b).

3a. Clone this repo and install the required Python libraries:

cd /boot
sudo git clone https://github.com/dhague/vpower.git
cd vpower
sudo pip3 install -r requirements.txt

3b. Download the zip file from the "Clone or download" button above and extract it to the microSD card so that the vpower folder contains the files from the repo.

4. Copy vpower.cfg to its parent folder (e.g. /boot) and then edit the values inside - typically you'll set speed_sensor_id to the ANT+ ID of your speed sensor and speed_sensor_type to your type of sensor. Next, uncomment one of the power_calculator lines to tell vpower what type of turbo you are using. If using a wheel-driven trainer you will then want to set the wheel circumference value. If you have a wind trainer (i.e. the resistance comes purely from a fan in the air) then you will want to set the air_density value for better accuracy. Set correction_factor to 1.0 to begin with - you can adjust it later according to your own comparative tests or perceived exertion.

This is just a matter of downloading and setting up the BME280 Python module.

1. From the parent of the vpower folder (e.g. /boot), run:

   wget https://bitbucket.org/MattHawkinsUK/rpispy-misc/raw/master/python/bme280.py

Alternatively (if using option 3b above), download the file from here and put it in the root of the microSD card.

2. On the Pi, run sudo i2cdetect -y 1

   if 77 is shown in the output grid, then edit bme280.py and change 0x76 to 0x77 (at line 27 or thereabouts)

3. Edit bme280.py and surround the print arguments with parentheses (for Python 3 compatibility).

4. Running python3 bme280.py should output the chip's data and the current temperature, pressure & humidity

1. Log into the Pi

2. cd /boot/vpower

3. sudo python3 vpower.py ../vpower.cfg

You should see something like this:

Power meter ANT+ ID: 37127
Starting ANT node
Starting speed sensor
Using KurtKineticPowerCalculator
Starting power meter
Main wait loop
Power: x W

The x corresponds to the power value in the message being sent, so it will only appear once you start pedalling. If you have a BME280 sensor attached then you will also see some data on temperature, pressure, humidity and air density and a o will occasionally appear when the air density is updated.

To stop the power meter running, just do CTRL-C and after a few seconds everything will be shut down and you can unplug the Pi.

You can turn on DEBUG/diagnostic output by setting debug to True in vpower.cfg.

1. From the init.d folder of the repo, copy the vpower file to /etc/init.d, e.g.

   sudo cp /boot/vpower/init.d/vpower /etc/init.d

2. Configure the service to start at boot time:

   sudo update-rc.d vpower defaults

1. Install the libusb-win32 driver for the ANT stick, it can be easily done using Zadig.

2. Clone or download this repo and install the required Python libraries:

   pip install -r requirements.txt

3. [optional] Install pywin32 (to stop the ANT node on terminal window close):

   pip install pywin32