Add virtual power to your trainer ride on the Tacx Blue Motion T2600. Record your speed during your workout on your Garmin and add virtual power with this Python program. The advantage of vpower2 over vpower is that the convertion step from FIT file to TCX is not needed anymore.
- Clone or download vpower:
git clone https://github.com/cast42/vpower.git
- Create the environment:
conda create -n vpower environment.yml
- Activate the environment:
source activate vpower
- Execute:
python vpower2.py <file>.fit
- A new file is created (or overwritten) called
vpower_<file>.tcx
that contains the calculated power values derived from the speed recorded by your Garmin. - Upload the generated file
vpower_<file>.tcx
to Strava for analysis - By default, it's assumed the lever is at position 5. If your lever was at another position during your workout, specify this as follows (assuming lever position was 6):
python vpower2.py --lever=6 <file>.fit
- Deactivate the environment
source deactivate vpower
Add virtual power to your trainer ride on the Tacx Blue Motion T2600. Record your speed during your workout on your Garmin and add virtual power with this Python program.
- First, obtain a TCX file corresponding with the .fit file from your Garmin.
- Download the file vpower.py into the same directory of the .tcx file.
- Execute:
python vpower.py <file>.tcx
- A new file is created (or overwritten) called
vpower_<file>.tcx
that contains the calculated power values derived from the speed recorded by your Garmin. - Upload the generated file
vpower_<file>.tcx
to Strava for analysis - By default, it's assumed the lever is at position 5. If your lever was at another position during your workout, specify this as follows (assuming lever position was 6):
python vpower.py --lever=6 <file>.tcx
The python fitparse library requires Python 2.7. I created a conda environment to set this up:
conda create -n vpower environment.yml
If you are still running python 2.7, you can just install fitparse and docopt:
pip install docopt
pip install fitparse
- Execute:
python fit2gpx.py <file>.fit
- A new file is created (or overwritten) called
<file>.gpx
that contains the ride in gpx format.
These tools rely on fitparse that is still using Python 2.7. To upgrade, those version should be rewritten with fitdecode that uses a modern Python version (>= 3.6).