Wednesday, January 2, 2019

2.4.x Release notes

2.4.5
-(Model) Calibration algorithm improved
-(Output) Write vertical vibration into SMO2% FIT field
2.4.4
-(Settings) Add default relative humidity into the settings
-(Output) Estimate and monitor gearing (big chainring only)
-(Output) Change format of the Lap.csv file for training mode
-(Format) Fit SDK updated to 20.88.00
2.4.3
-(Output) Airspeed added into the SMO2 field of FIT file
-(Feature) Fix wind forecast screen (3 hours period instead of 1 hour, though)
2.4.2
-(Output) GPS Lat/Lon added to Raw.csv file
-(Feature) Outliers detection (Grubb's test) for CdA of a run (purple flag is shown, suggesting to delete the lap)
-(Fix) Typo in the line Mean in the Lap.csv file
-(Fix) Comment with micro was not working on Android>=6.0 (needed authorization)
2.4.0
-(Mode) New training mode (add training metrics on lap/run)
-(Mode) Automatic segment mode (live CdA during a ride) is not a premium feature anymore
-(Mode) Home trainer mode is suppressed, but try to guess if home trainer is used
-(Feature) Add wind direction/current location to Wind forecast menu + fix a bug
-(Feature) Ask for powermeter manual calibration each time the zeroing strategy is changed
-(Model) Try to calibrate the different elevations by the more accurate one
-(Model) Several modes (TimeTrial, Segment, Criterium, Training) can now be used with GPS altitude data
-(Model) Estimate total mass (rider+bike) during long and steep enough climbs (training mode only)
-(Model) Try to estimate anemometer calibration factor after each lap/run
-(Model) For crank based PM, lap is now cancelled only if pedalling less than 90% of the time
-(Format) Fit file format replaces iBike one
-(Fix) Useless laps added during velodrome sessions when bike is stopped
-(Fix) Distance overestimated if datadrops happened
-(Fix) Try to improve ANT+ speed sensor reconnection
-(Fix) GSM location fixed
-(Fix) Turn off Bluetooth when the Weathermeter is disconnected in order to improve reconnection
-(Fix) Improve speed spikes fix and add a sonore warning
-(Fix) Averages for power or wind could be wrong
-(Fix) Graphics crashed with old Android versions
-(Fix) Bike stops may not be detected correctly with ANT+ speed sensor
-(Fix) Error in watts gain/loss when the Weathermeter was activated
-(UI) Option to record data from two powermeters
-(UI) Setting to disable beep warning for ANT+ data drops
-(UI) Search of ANT+ devices is not automatic anymore
-(UI) Add three more configurable fields (swipe down current fields to make them appear)
-(UI) Add new fields plus lap, run and ride averages for some of them (speed, power, ...)

Saturday, December 22, 2018

Test of the dynamic calibration of the airspeed device

Bad and good news from the velodrome test today. I implemented recently in the app, a dynamic procedure for calculating the correction factor to be applied to the raw air velocity measured by the anemometer. This factor is calculated at each lap. The anemometer (blue) was placed on the left drop of the bike during the test:
Several turns are made where I changed the position of my hands (hoods then drops, again hoods). Thus, we can see that when the left hand on the hood is close to the anemometer (certainly creating a local air acceleration), the calibration factor decreases by about 2% (1.095 to 1.075), and so a similar erreor increase of ~4% on the calculated CdA:

The good new is that if we keep a same position, the factor is well calculated with acceptable variability +/- 0.5%. We can also assume that by placing the bike's anemometer in another location, the issue may be reduced.

Sunday, August 12, 2018

Power meter data quality

Yesterday, for the first direct comparison during a turbo session, I got results, which confirmed my suspicions, between my two current powermeters, a Power2max NG-Eco (2 months old) and a PowerTap G3 (recently changed). Unfortunatly, I deleted some of the data. Today, during a ride with intervals of several minutes, I recorded the following averages:


One (or the two?) gave an unexplicated difference during the first lap of the two runs (the relative difference does not depend of the gear ratio and is constant at +/-0.5% during the whole lap). Auto-zero feature was activated on the two power-meters and I stopped pedalling during the laps. I have some guesses about the suspect. More to come.  

Tuesday, January 2, 2018

Cheapest Android phone to use CdaCrr bike computer

After some research (cause people asking me when the app will be available on iOS, my answer is: sorry, never), here is the cheapest way to buy an Android phone to use the app (prices are given for December 31th, 2017) :


So for 83$*, you should have an ANT+ enabled Android phone (follow this) to run the free version of the app (classic lap mode) and aero field test on velodrome or other closed loop. There is also GPS and Bluetooth on this phone, so more CdaCrr modes could be used. Not tested yet, but I plan to buy and give a try soon.

*: You would need also an ANT+ powermeter (~450$), an ANT+ speed sensor (~35$), a phone bike mount (~10$). And a bike of course :-)

Edit 05/01/2018
I bought the Blackview A7 and it seems great for the price, but unhappily, it does not have USB host feature so we can't use an ANT+ stick :-(

Edit 14/01/2018: 
-List of ~130 Android phones with ANT+ built-in on www.thisisant.com
-List of ~460 Android phones with USB Host on www.gsmarena.com

Sunday, December 3, 2017

Virtual elevations

Last week results. The anemometer improves CdA accuracy when drafted a lot by cars. Supposed Crr at 0.004 may be too high.

Monday, August 7, 2017

Track model

According to Google Maps, the 250 meters track has straight of length 74.8 meters and turn radius of 16.0 meters.

The measured incline along the black line in the turns is ~30°, for 6° in the straights.

Friday, March 17, 2017

Velodrome session with high wind

Back to the outdoor velodrome. I wanted to see if the anemometer could improve precision with high wind conditions. The results, after three runs of 10 laps (trying to keep the same relaxed position on the hoods of my road bike) are:


We can notice that the wind, measured near the ground (the anemometer was mounted on the drops), is in the 5-15 km/h range (weather station indicated 22 km/h this day). 

Quick updated conclusions for the moment:
  • with low and constant wind conditions (0-5 km/h), data anemometer is useless as the effect is nearly cancelled (CV~1-2%)
  • with higher wind conditions (>5 km/h), anemometer is mandatory to maintain acceptable precision (CV~1.5-3%) whereas for apparent CdA (wind data is not included into the model) variability is way of.
  • variability from run to run is still an issue in my protocol (0.343, 0.353, 0.342) even if it should be lower with the use of a time trial bike.