Thanks Pierre.
I'll give it a go
Regarding the WM- I have checked the battery and it's good.I have tried to update the firmware but this didn't happen( says something like 'no dpc file')
I did calibrate the WM compass with the wind and weather app .

Does it matter if Bluetooth is on before opening cdacrr,or letting the app request it?
I did try this without obvious benefit.
It's lashing down with rain,so not testing today
:-(

Thanks for your questions.
There is no need, IMHO, to update the firmware. Good if the compass is calibrated, but CdaCrr only uses airspeed data, and if no barometer is on the phone, pressure from the WM, and if no meteo station is detected, humidity value from the WM. Temperature from WM is never used caused there is a risk to sun exposition that trashes the temperature value and so the air density computed by the app.

It does not matter. CdaCrr detects the Bluetooth is off and ask you to enable it. It can take few seconds before the WM is connected to the app.

---

Blog | Twitter| Bike CdaCrr app

I haven't had any issues with the WM crashing in 4 sessions now. The app *would* crash every time I selected Dark Sky as the weather station after the last update. Using OWM fixed that.

I think it's best to use your runs for calibration. I do one lap at a time with fixed lap length, out-back, and ideally at least 12 laps (6 each for A-B testing). You need to mount the WM where it won't be influenced by the bike/rider, and use the average scaling factor which the app now computes each lap to adjust your CdA numbers after the fact. Average scaling factors have been between 1.072 and 1.094. I realize this isn't going to help you until you can get the device to stay connected!

A correction to what I said earlier about a SD of .0038... that was actually too high. In 4 sessions of 8-10 runs (laps) SD has ranged from .0026 to .0037, with gusting winds >10km/hr each day. Average CdA each day was .205, .205, .205 and .208, over a wide range of wind directions and temperature changes, so good repeatability. Note, this is still processing the data by hand in a spreadsheet. The app's use of weather station data usually results in big errors in environmental variables where I live.

Pierre, any chance we will be able to select the source for each variable? For instance I'd like use the WM for pressure and humidity, and it works pretty well for temperature also so long as the lowest temperature during the lap is used; the sun seems to have little effect if the airspeed is high. I do have a Garmin Tempe sensor on order, but the less hardware needed for good data, the better.

You convince me to add an option (new app version soon) to get temperature from the WM. Good idea too to use the lowest temperature of the lap for air density in order to reduce the sun influence. I need to think a little on the last point to implement it. So for environment data, the new strategy for the app will be:

temperature:
// -WM (if explicitly enabled)
// -Tempe sensor
// -Station
// -Phone (if trainer session)
// -Setting

pressure:
// -Phone
// -WM (less sensitive to speed than the phone but less precise)
// -Setting

humidity:
// -Phone
// -WM
// -Station
// -Setting

---

Blog | Twitter| Bike CdaCrr app

Great! But... shouldn't the Tempe sensor have priority if it is connected?

I was hoping to be able to select where the data comes from for all variables, but no idea how hard that is to implement.

By default temperature from WM would be disabled. So if you have a tempe sensor, it will have priority. But for clarity, you are right:

temperature:
// -Tempe sensor
// -WM (if explicitly enabled)
// -Station
// -Phone (if trainer session)
// -Setting

Sorry, but IMO, too much work for a limited value, and above all, a lot of risk for a newbie to select a bad source for one or several variables.

---

Blog | Twitter| Bike CdaCrr app

That's fine; with temperature measurement I should be good. Oh, except for humidity. The phone is a bad place to get that since it's in front of my face. The WM is better. The station measurements are pretty useless where I live since temperature can vary easily 10C over a few miles, and humidity varies a lot also when there are random thunderclouds around.

I'd asked you earlier about lap lengths being off, but using 1.003s time steps and using the first data point in the next lap appears to have fixed that. For instance I'll tack on the first data point of lap 6 to the end of lap 5. Since "fixed lap length mode" is set up for continuous laps I thought this would be correct. Does that sound right?

Regarding auto zero. With my Powertap the "enable auto zero" is not supported. "Enable app autozero" seems to work except that I see messages on the screen that indicate the request for calibration is sent at odd times, so I don't know if I can trust it. So I've been disabling auto zero and doing a manual calibration. That's fine, but I notice the power field on the screen sometimes says -0 and sometimes 0. Is there some significance to the minus sign?

No problem, I will move WM relative humidity to the first place.

Yes it is right. The app computes the average time between 2 ANT+ messages and it is not 1.000s. I saw, by comparing the files, that bike computers (eg Joule 2.0) repeat (dt>1s) or discard (dt<1s) a recorded value after a long time when trying to record every one second...

In fact, that means disabling autozero seems not supported. Strange, it is possible on a PT. I don't see that on my G3...

The app autozero was a try to monitor zero offset as I suspect my PowerTap zero drift was the main issue of the poor CV from run to run during my aero session. So, I implemented an algorithm to send a manual calibration to the PM when conditions (criteria) were satisfied during few seconds: no power on pedals, no cadence. Each time a zero was done, the PM sends back a zero offset value, which appears on the screen and is written in the logs. I am not sure the criteria is the same than the PM auto zero strategy, so I would not recommend to use this feature (I will suppress it soon). But at least, it confirmed me that the PowerTap zero drift was the issue of poor CV from run to run (eg: aero session during early morning when temperature rises quickly), cause even if the zero offset is done between each run, the zero drifts during a run of several minutes when you don't stop pedaling...


A minor bug in the app, -0 is because when smoothing power is computed on several seconds, roundoff error may give a slight negative zero. I will fix.

---

Blog | Twitter| Bike CdaCrr app

Thanks for reporting these good results, even for winds above the recommended range [1-10 km/h]. Repeatability from day to day is a proof of a good protocol: I accessed to that point only last year (after 10 years of virtual elevation practice) when switching gradually from a classic protocol (excel sheets+PowerTap hub) to app+anemometer+tempe sensor+Power2max NGeco.

---

Blog | Twitter| Bike CdaCrr app

The version is out, but still without the constant air density hypothesis (computed with the lowest temperature value) on each lap to reduce sun influence.

I promised to share data, here they are (3 files recorded by the CdaCrr app on a Samsung S4 phone, and sync with the WM mounted on a TT bike), from my last session (laps of ~3500 m).

First, the comparison of bike and air speeds on the two laps (headwind for out and tailwind back), then comparison of virtual elevations (with and without anemometer) which are compared to measured elevation from the barometer (precision +/- 0.6m) of the Samsung S4 phone:



Second, if we compare elevations on the two laps, we can see that the virtual one given by the model with anemometer airspeed is surely the best:



The Fit file was imported into Golden Cheetah then exported, so airspeed data is missing (it is in the CSV files).

---

Blog | Twitter| Bike CdaCrr app

With more readable images:

---

Blog | Twitter| Bike CdaCrr app

Looks great Pierre!

---

What's your CdA?

Got WM working,but have found that if I start thelike WM before connecting to the app,it doesn't work,but starting AFTER works beautifully.
I have been getting CV <2% without WM on low wind days,in a road position (hoods).
Any car throws it off.

I will find out myself,but does the WM increase accuracy when cars are present?

Like rruff cda of WM and mount looks to be an additional 0.06 to 0.09m^2

Just done six runs ,before the heavens opened.Lots of cars.
CV w/o WM is 3.44%, with is 2.5%

I've been getting SDs of ~.003 m^2 (CV~1.5%) on average on high wind days. How is your WM mounted? What is your test course and protocol?

It should improve accuracy with cars, since it will account for the change in airspeed. But I don't know how well it does this. I know the iBike devices had a hard time with drafting, so maybe it's difficult with a simple point measurement. I think avoiding traffic when you are testing is a very good idea.

Airspeed measurement is really helpful on windy days, but the lower the wind the less benefit it adds.

Currently using road bike,WM mounted on end of drop,~8cm outboard.
I have a short (650m) half pipe,starting and stopping run manually.
Could do longer runs,but surface changes.
Run is really handy as right beside my house

Is that "classic lap mode"? I haven't used that one. Do you need to click to start and stop each lap, or do you do a series of laps and just click at the beginning and end?

I've been using "fixed lap length mode", on a 3.1km course that has a steep hill for a turn (quarter pipe?). I have to start it, but it ends automatically.

Mine is also a quarter pipe.
Am using classic lap mode.
I could do longer laps and fixed lap length,but I am impatient!
I would quite like to do fixed lap length so that I could estimate the CDA of the phone and case,(by putting it in my pocket for most of the laps)

If your laps are like mine you can use FLL mode. Saves having to click at the end.

Do you have a TT bike also?

The post below was sent to rruff this am and as per his suggestion I am posting here so that any info gleaned may be of use to others.


Good morning Ron!
So, this fall I had hoped to get in a lot of testing outdoors, but as luck would have it work has slowed to the point that funds are too tight to buy new gear. Nevertheless, I think I need to get my protocol down and practice collecting good data for when I can get some new stuff. Yes, I have never tested outdoors and have relied on my aero-eyeball to get me close and the tunnel to refine/confirm. I have questions about equipment, course, and editing data. All of which I have gathered while I have been lurking on the various Chung threads.


Equipment:

• I plan on using CdaCrr App w/anemometer
• is it possible to upload the files to snip turnarounds?
• if so, I am assuming GC's AeroLab would be the preferred platform

Mounting location

• Phone is okay in a jersey pocket?
• Anemometer: the Bayo3Devox has two mounting locations for extensions, I am using the narrower set and plan on using the outer set for the anemometer. I think this should be a pretty clean spot, but would like your opinion.

Course description (please critique)

• I have a ~1200m course with roundabouts at both ends.
• I will have to sit up on the downhill end (eastern roundabout) and brake heading into it also, as speed will be high regardless of power.
• Not sure about sitting up at the northwest roundabout but if I am clipping it I'm not sure if there is good cause for not
• there is a painted line I can follow, so, accurate repeatability should be attainable with a concerted effort
• there is elevation change but nothing great....maybe 3% and it is steady(ish)

anything else I am not thinking about? I know there is. ;)

---

My YouTubes

I have the anemometer mounted to the base bar handle and sticking up and out about 18cm from there. I'm not sure that is far enough away though. Mounting it farther forward would be cool, but I wouldn't place it lower or closer to "stuff". You want the height near the center of drag too, because wind speed increases with increasing elevation above the ground.

Your "platform" is the CdACrr app itself. It will give you a CdA along with other stats every lap. And you can download it all for analysis when you are done.

GC won't snip. Neither will CdACrr. It is however a feature I'd like, and if enough people ask Pierre nicely he might incorporate it.

At least for the mode I use, I need to access the phone to start a lap. I have mine mounted on the extensions (the case is taped in place and I pop the phone in and out of that). Put the computer in a pocket or bag, or leave it at home.

I use the "fixed lap length mode", so I only need one turn on a hill. I have to hit a button at the start and it automatically stops (and starts a new lap at the end). I then stop (brake) and turn around and that lap will automatically end. The app will ask if I want to comment on that lap and tell me what the computed wind scaling factor was and if I want to change it. This is never a good idea on a single lap; just leave it be and make adjustments when you are done, based on the scaling factor for all laps. Then the app is ready to start the next lap....

If it's really a 3% grade, I think that is too steep. Mine is ~1.2% and I wish I could find a flatter one. Particularly if you have a tailwind on the uphill, then you will need to go really hard to outrun the wind, plus all of your aero-drag-energy for the lap will occur on the downhill part, which isn't ideal. If you could find a flattish road with a hill at one end for turning, then that would work well.

Depends the exact profile but, yeah, 3% average over the entire length of the test section is pretty steep. These methods work best if you can get a wide range of speeds/powers not only within a lap but also across laps, and if it's too steep and you need to brake at the ends then you'll be less inclined to hit high power on the downhill part. (If you always coast at zero power on the downhill part, then all of the downhills are going to be identical, so you don't get any extra information from additional runs -- that's not bad per se, but it's inefficient and a waste of time).

You can use my app (https://MyCdA.app). You can segment your activity however you want, analyze each segment separately and it will give you an average CdA and other stats for the segments you created. The app supports .fit files with airspeed data produced by CdaCrr and you can do whatever you do in GC plus and in a more useful and user friendly way :)
Let me know if you run into any issues.

---

What's your CdA?

Thoughts since posting my questions:

• Since I am measuring windspeed do I need to go out and back? My thinking is I could do just the “uphill” leg, then roll through both roundabouts rinse & repeat.
• Also, the pavement is perfect, smooth, and new enough to be fast but not so new that it’s tacky/greasy/oily. In addition, there are “brick” crosswalks at the start and end of the uphill lap (same for downhill). I bet this would show a noticeable CRR difference compared to the glass pavement making lap markers almost not needed.

rruff wrote:

I have the anemometer mounted to the base bar handle and sticking up and out about 18cm from there. I'm not sure that is far enough away though. Mounting it farther forward would be cool, but I wouldn't place it lower or closer to "stuff". You want the height near the center of drag too, because wind speed increases with increasing elevation above the ground.

yeah, that makes sense. I was looking at it from a clean air perspective, but also aesthetics (oh my vanity). So, above the bar and as far out in front that I can keep it rigid enough to prevent bounce? Again, i’ll try to offset it away from me without placing it in the path of me reaching for the bullhorns.

rruff wrote:

GC won't snip. Neither will CdACrr. It is however a feature I'd like, and if enough people ask Pierre nicely he might incorporate it.

Looks like he kinda has that with custom segments and I was also thinking of going into the fit file and snipping the data manually.

rruff wrote:

If it's really a 3% grade, I think that is too steep. Mine is ~1.2% and I wish I could find a flatter one.

Well, it looks to peak at 3%, but it really never feels like that and it does “roll” a bit, well as much as a very false flat can.

---

My YouTubes