What's new

Question about speed sensor distance accuracy

Yogi

Speeding Up
May 17, 2023
120
63
Cheers all, new guy here.

As a new cyclist, I was curious about tracking cadence on my fitness rides with my Polar training watch so recently purchased the low cost IGPSports CAD70 and SPD70 sensor combo. The sensors connect to the watch via Bluetooth.

I've been riding this same route thru Toyosu-Odaiba-Ariake 3-4 times a week for several months now and my phone GPS has always recorded it as being ~20km measured in 1sec resolution. So I was quite surprised the speed sensor first reported it only as 16km, 20% less than the iPhone GPS!

I had first set up the sensor for this first run using the recommended default 700x32c wheel circumference of 2155mm. This turned out to be a little short when I measured the wheel to actually be 2180mm. After reconfiguring both sensors (as an aside, the watch also wants to know wheel size for the cadence sensor during pairing. Why?) with the new value, the distance on the second run was 18km and still 10% less than GPS showed.

As a check of the phone GPS accuracy I then rode around the 4.8km Toyosu promenade section that has distance markers stopping at each marker to compare the distances recorded by the watch/sensor against iPhone GPS. By the end the watch and sensor showed 3.55km while the iPhone was spot on 4.8km +-20m.

So what to do and does accurate distance even matter? Should I keep increasing wheel circumference beyond what I actually measured until the watch and sensor agree with the GPS or just keep the cadence sensor and return the speed sensor and rely on phone GPS for speed and distance? TIA.
 
Does the speed tally with what your phone recorded? It sounds like the speed sensor is inaccurate, even though these things should be a solved problem by now its still a quality issue i guess.

If you had a cycling gps computer it usually adjusts circumference automatically to match with what it records.
 
Nope. Avg and max speeds vary too with the watch being slightly faster. Maybe I should also mention the watch also has GPS capability and had always been close enough for me to what the phone recorded previously so maybe that's more confirmation for GPS accuracy. But when paired with sensors the watch GPS data is not used. Anyway, I suspect the sensor only sends wheel rpm to the watch and the watch calculates the speed and distance using the wheel circumference. I guess in the end it's just fun with numbers!
 
Thats exactly what all speed sensors do in a nutshell. Which is why you need the circumference in the recording device.

If the device is not logging the rev correctly thats gonna skew things. Is it a magnet based or orientation sensor (usually zip tie to hub). Orientation would probably be more prone to cheap quality issues.
 
Nowadays, speed sensors are used in conjunction with a GPS head unit. These struggle in e. g. forest, mountainous or urban terrain where GPS reception might be crappy. A speed sensor will smooth that out.

Using a speed sensor by itself is quite outdated. I'd definitely trust a GPS over a speed sensor for most metrics, provided you are using a suitable app to record your GPS track.
 
Increase the tire size until you are happy. Then just use it for cadence. Was a time we'd count revs or heart beats for 10 seconds and multiply by 6. And then I bought a polar watch.
 
I've been using the Polar Grit X Pro smart watch in "headset mode" strapped to the handlebar and their OH1 optical HR sensor on my forearm. Gives me all the cycling data I could process from a typical cycle computer except mapping. Increasing the wheel circumference factor again will be next test. Basic geometry would lead one to calculate at least a 2199mm circumference for a 700c tire while I measured 2180mm, but clearly there are other factors in play according to this wheel standard chart. I'm already factoring 700x38c size though my tire is marked as 32c. Some real black box voodoo going inside these devices!
IMG_0211.png
 
You arent running so high pressure that your tyres are perfect circles. There is tyre deflection from sag, and manufacturers are funny with sizing. e.g. Panaracer gravel king 700x38c is 40-622 ETRTO.

That 1% difference between 32and 38c *should* be within margin of error just to use the chart but 🤷‍♂️
 
UPDATE: Success! Discovered the Polar watch settings had a new-to-me BIKE CRANK LENGTH setting buried in the menu. This setting was only supposed to be available for Power Meters according to the manual! Simply updating it to 175mm from default 172.5mm and setting wheel size back to the measured 2180mm now results in nearly total agreement with what iPhone reports for distance, avg speed, and max speed. Why this worked is a mystery as the Cadence sensor sits at about 150mm on the crank. But I'll take the win!
 
Maybe phones have gotten better, and I do know they have barometric altimeters now, but I guess I've never seen something being compared to phone data as a way to judge accuracy.
 
They have improved but so has GPS surveyed maps. Ever since Japan launched its own QZSS satellite GPS a few years back, accuracy can be within a few meters. Hopefully we will have centimeter accuracy once all 7 satellites are in place by FY2025 but the new H3 rocket has issues. In addition iPhones at least also use cell tower triangulation and known WiFi networks to assist. Also, the Toyosu-Odaiba-Ariake wanterfront where I am testing is almost flat with unobstructed sky views and the Toyosu promenade is a marked 4.8km course.
 
  • Informative
Reactions: jdd
Ever since Japan launched its own QZSS satellite GPS a few years back, accuracy can be within a few meters.
What devices use Japan's GPS system, though? As far as I remember chipsets only advertise GPS and Glonass compatibility. (I cannot recall seeing Galileo compatibility being advertised either.) But perhaps all modern GPS chipsets just are compatible these days?
 
The official QZSS website will probably answer most of your questions about GPS compatibility, https://qzss.go.jp/en/

Phones and other devices that use dual frequency reception like the top line iPhone 14's and some Androids should be especially accurate even areas that have poor cellular and WiFi augmentation coverage.
 
Nice, thanks, I didn't know that. I'm glad that I get the benefits of the additional accuracy (compared with GPS).
 
If you want your wheel circumference to be super accurate, you have to do a rollout test while actually riding the bike.

The measured circumference of an unladen wheel is actually going to be a bit longer than one with half a rider's weight on it.

It will also depend on your tire pressure.

But if you get it right, it will be more accurate than GPS.

Of course, your ride distances might differ just because you don't ride perfectly straight...
 
I've now have a distance difference of <1% between what speed sensor/Polar watch calculate and what the iPhone 13 GPS records with the sensor/watch always slightly more. I suppose I could keep tweaking the wheel circumference until I got even closer results but the phone GPS is only to 1 decimal place while the watch is 2.

So I'm calling it a win given I only paid -¥4,000 total for the speed and cadence sensors, about a third of what Garmin or Wahoo wants. Now I just have to figure out what to make of the data! The distance is seems to be just trivia point, while the cadence and speed might be more useful metrics for improving and finding an endurance sweet spot.
 
Well at some point you'll know cadence by feel. The number loses its importance.
 
Back
Top Bottom