Speed sensor : wrong auto-calculation
Completed
Speed sensor : wrong auto-calculation
When using auto-calculation with a speed sensor, wheel circumference is over estimated on Karoo 2.
I tried the following test: connected the same speed sensor to both my Garmin Fenix and my Karoo 2.
Both are configured with auto-calculation of wheel circumference. My wheel is 700x35c, announced circumference is 2168mm, measured by myself: 2197mm.
After my ride, estimated circumference was:
Garmin Fenix : 2183mm, almost what I measured.
Karoo 2 : 2249mm, way to high.
Here are the stats of my ride, started and stopped at the same time on both devices :
Garmin Fenix: 31,87km, 1:22:20, 23,2km/h
Karoo 2 : 33,27km, 1:22:52, 24,2km/h
How is such a difference possible, how does the Karoo 2 evaluates the wheel circumference ?
Thanks for your help,
Luc
-
Just chiming in here that I have the same problem with Auto calibration. I added the Wahoo Speed sensor last week and already noticed average speeds being unusually high. Today I did a route I usually do which was 67km but it ended up being 71km on the Karoo2 which is a 6% difference. After correcting the distance on strava based on GPS it shows 67km again. Also average speed went down by 3kph (as well as estimated watts and calories) which is a lot. Would highly appreciate a proper fix on this as right now the auto calibration is pretty useless.
-
@Arne I do the same thing and I'm continually disappointed to see no fix.
I was listening to a recent episode of the Marginal Gains podcast. They were talking about their first bike computers and lamenting on how the actually had to manually set their wheel size in the computer. It made me a bit sad that we still have to do this with our $400 Karoo2.
I ended up measuring my wheel circumference. It was very enlightening on why accurate auto calibration is important. I'm running a 700x25 tire (Continental GP5000). The chart says the standard is 2105mm. My measurements at 90psi:
- Without rider weight: 2121mm
- With rider weight (170 lbs): 2091mm
This little exercise showed that that tire brand/model can impact circumference. More importantly, bike/rider weight can have a large impact. While I didn't test, I'm assuming tire pressure would also affect tire circumference. In other words, if I gain/lose weight (or add/subtract to the bike) or adjust PSI my circumference could change enough to through off my accuracy. It's a lot to ask to remeasure circumference on each ride.
While I appreciate the continued development and new features, this needs to be fixed ASAP. Accurate data is the cost of entry in this category. It's the one thing that makes me hesitant in recommending the K2 to other riders.
-
Just a thought... I wonder if the unit drops the connection with the sensor every now and then? It would then have to increase the distance travelled per revolution to make up the lost distance from the sensor.
Does it work better with sensors fitted to the front wheel? They're closer to the unit with less things to interfere with the signal or does the unit itself have difficulty keeping up with the data, possibly from multiple sources/sensors? Or, is it just a calculation error? How about the environment? Could there be interferrence and/or reflections that would 'confuse' the signal (ie urban environments)?
When does the auto calibration take place? At the very begining of a ride, possibly when you're starting off and out of the saddle (weight transfers and bike angle could affect the rolling circumference-the more leant over you are the smaller the circumference).
I suppose it has to happen immediately to get a reading up to the display. Why not make further calibrations as the rider settles?
I've heard that braking can affect the measurements as well (changes rolling resistance), so if it auto calibrates as you're starting off downhill (or in stop/go heavy traffic or even tight, twisty urban areas), measurement could be affected.
How about doing the auto calibration a number of times per ride? It could use the gps to identify a flat section and look for a constant, steady speed before doing the auto calibration. Or, just continually update versus gps, it would only get more accurate like this. Or... just use the speed sensor to smooth out the gps speed reading and not use it for the final distance/speed measurements.
How do the others (Garmin etc) do their auto calibration?
Just thinking alloud really and have not seen any of these suggested. Sorry if they're silly suggestions/thoughts but thought I'd add them, you never know! -
There’s definitely a problem with auto calculation. On a typical 35 mile ride my K2 is a bit more than a mile higher than all the other Garmin and Wahoos in the rest of my cycling group.There’s definitely a problem with auto calculation, And consequently it inaccurately shows a higher average speed. I’ll be trying manual calculation to see if I come in closer to the the truth.
-
Any updates here?
I have my Karoo 2 connected to a Garmin Speed 2, and on my last ride had a distance that was 3.4% higher than two riding buddies on the identical route. Both of them are using Garmin Edge 530 computers, one with an identical speed sensor to me (Garmin Speed 2) and one using GPS. Their deviation was under 0.4% against each other.
It looks like this has been a very slow bug to fix, and I'm hopeful that it's updated soon. I know that others here are suggesting measuring roll-out manually, which I will do, but I adjust my tire pressure based on conditions (gravel vs road, wet vs dry) and don't want to be doing that all the time when cheaper bike computers from Garmin are flawless in their calibration.
-
Hammerhead is pushing updates regularly that usually include a number of bugfixes, which is great to see. However, I personally don't understand how this issue does not get a bigger priority, as to me it seems more crucial than some of the other bugs that get fixed. Speed sensors are basically unusable right now as long as you don't want to go through the hassle of manual calibration, which actually I won't do as I just fall back to GPS and I think something basic like this should just work. Using GPS is an okay fallback, but as we know, it's speed readings are delayed and it doesn't work great in areas where GPS reception is compromised.
It would be great to get another status update from the team on this, is there a way to escalate an issue?
-
The latest software lists a fix for the wheel circumference calculation algorithm. Can anyone from Karoo confirm if that is supposed to address this issue that has been around for many months now? I guess I'll have to give it a shot on my next ride but it would be nice to know if it is targeting this specific issue or something different.
-
FYI, I just did a ride this morning after installing the latest Karoo2 software update with auto-calculation turned back on for my Wahoo speed sensor. When I compare the distance from the speed sensor to the GPS distance, they are only .02 miles different for a 23 mile ride. This is MUCH closer than I was seeing before with the auto calibrated circumference so "hopefully" this has been fixed once and for all. I'll be interested to hear of others have similar success with the latest software update.
-
No such luck for me. Just did a roughly 33 mile ride (Garmin auto calibrate using Garmin fenix 5+) and the Karoo 2 (auto calibration using same speed sensor) was about a mile more. I guess you could argue I should be comparing with GPS distance, but my Garmin distance as usual was very close to what my other fellow riders got. Back to the drawing board 😤
----
Saad -
I had a century ride yesterday. Just in case used my cheap garmin watch as a backup. Ride was 103.5 miles according to map. Garmin watch showed 103.4 miles, Karoo2 - 107.2 miles. I went to settings and Auto circumference shows 86.2 inches while 700x26 is about 83.3.
Uploaded to strava my garmin ride instead of Karoo, losing watts information, but distance is more important. -
How about a compromise. We care about our immediate speed while riding and if speed is off by less than 2% - so be it. But the most accurate average speed is always ride distance divided by time. Time - you can't screw it. Distance calculated by gps is also acceptable. So have an option what speed sensor will be used for: Speed or Speed and Distance.
If you on 30 miles ride and your hands are shaking or you have to avoid many walkers, runners and other riders, will you care that by calculating distance using speed sensor your total ride distance will be 30.3 miles? -
Sorry but I think there is no need to overcomplicate this. It's an interesting idea but honestly imo Hammerhead should just fix the root cause of this issues as this works totally fine with all of their main competitors on sometimes significantly cheaper devices. There really is no reason we should have to think about this and employ workarounds and/or hacks. They just have to prioritise this issue and finally fix it. As a Software Engineer myself I am pretty certain that there is no rocket science involved here, they just didn't work on it yet which is unfortunate and getting a bit frustrating after all these months.
-
"time - you can't screw it" Do you mean clockwise or counterclockwise? But seriously, there IS an issue with time using bike computers and that has to do with the auto-pause feature. I've given up my Karoo for now and I ride with both a Garmin and a Wahoo. I have noticed that the moment when they auto-start and auto-stop varies between them and I'm sure the same would be true for the Karoo. This may even be part of Karoo's problem with calculating distance, since when you start and stop measuring affects both time and distance.
-
@Arne - you are not the only Software Engineer here. Are you sure other 2 are using circumference to calculate distance? I am not. From riding with and without speed sensor my Edge 530 distance calculation is not changing. One can easily consider that speed sensor is used for immediate and accurate speed displaying while distance and average calculated from gps.
Please sign in to leave a comment.
Comments
60 comments