GPS Computer Speed Accuracy

[TerryTibbs10]

In comparison to bicycle speed computers which use a magnet by the wheel, how accurate are GPS computers for measuring ones speed in real time? For instance the wahoo bolt.

Thank you

 

[cosmicbike]

In my experience with Garmin 500 & 800 units, it depends on where you are. Generally speaking if you're out in the open it's fine, whereas in places with tree cover it's pretty hopeless.
Which is why I have speed/cadence on my bike now...

 

[13 rider]

When I first got my wahoo bolt I still key the cateye wireless magnet on for a couple of rides there was discrepancies between the 2 but marginal about 0.1 miles per hour average .

 

[TerryTibbs10]

When I first got my wahoo bolt I still key the cateye wireless magnet on for a couple of rides there was discrepancies between the 2 but marginal about 0.1 miles per hour average .

So in essence, even in real time, the speed reading is very accurate?

 

[13 rider]

So in essence, even in real time, the speed reading is very accurate?

Yes while running both I didn't spend to much time looking at both but When I did they should never more than 0.1 mph apart

 

[swansonj]

0.1 mph sounds like 1% or even 1/2%. I'd have thought that was well within the calibration error of a wheel-derived signal (i.e. how accurately you actually know the correct tyre circumference to program it with)? Whereas, a GPS system, if it's picking up an adequate signal to start with (@cosmicbike 's point), should on average be dead accurate. GPS: systematic error small, random error bigger; wheel-derived: random error small, systematic error non-zero?

 

[Drago]

Generally GPS is pretty accurate. Nevertheless, remember that GPS is recording your movement between 2 constantly recalculated points in 3D space, whereas a wheel speedo is measuring your distance along a constantly changing surface - the reason the results between the two sometimes differ is because these distances themselves can be different.

 

[HLaB]

Ive added a speed sensor to my garmin as I found in certain tree'd areas etc it could bounce around a bit (which was a bit disconcerting on a TT until I started pacing off power) but it seems pretty accurate over longer distances and in the open when compared to OT's. I can increase that accuracy with Glonass but I think that eats battery on a longer ride.

 

[TerryTibbs10]

Ive added a speed sensor to my garmin as I found in certain tree'd areas etc it could bounce around a bit (which was a bit disconcerting on a TT until I started pacing off power) but it seems pretty accurate over longer distances and in the open when compared to OT's. I can increase that accuracy with Glonass but I think that eats battery on a longer ride.

Is it e asy to add a sensor to these sorts of sensors?

 

[Ajax Bay]

In comparison to bicycle speed computers which use a magnet by the wheel, how accurate are GPS computers for measuring ones speed in real time?

When I first got my wahoo bolt I still key the cateye wireless magnet on for a couple of rides there was discrepancies between the 2 but marginal about 0.1 miles per hour average

You are comparing one readout with another. Is either accurate? How accurate is one? How accurate is the other?

So in essence, even in real time, the speed reading is very accurate?

Schrodinger's cat. How do you know what your speed is "in real time"? The reading on a screen? Which screen?

remember that GPS is recording your movement between 2 constantly recalculated points in 3D space, whereas a wheel speedo is measuring your distance along a constantly changing surface - the reason the results between the two sometimes differ is because these distances themselves can be different.

When you say 'constantly changing surface' I think you are describing the undulating road as opposed to the type of surface eg chip and seal, tarmac etc - in which case your point is well made. Btw, except in mountainous terrain the vertical difference on a ride makes minimal (%age) difference to the distance ridden.
In my experience my GPS produces distances that are within 0.1% of the distance calculated for the route on Ride with GPS. Does that mean my GPS is accurate? The distance will be the integral of the myriad speeds recorded (by GPS) during the ride. If that distance is 'right' over time, then can I assume the displayed speed is 'accurate'? [Answer: IMO yes, except under trees/overhead cover/tunnels.]
I make the assumption that both the RwGPS and GPS (Edge 500) distances are accurate and attempt to set my wheel circumference on my wired (magnet on spoke) Cateye to get as close to the GPS reading as possible - my Cateye will only allow centimetre integers so the best I can get is to within 0.4% (eg 400m 'out' in 100km).

Ive added a speed sensor to my garmin

The 'speed sensor' calibrates itself by assuming that the GPS speed being recorded is accurate. It's not calibrating itself against 'real time speed' - how could it? [Nonetheless useful if you need to constantly refer to speed and there's intermittent overhead obstruction.]

 

[Ajax Bay]

Ive added a speed sensor to my garmin

Is it e asy to add a sensor to these sorts of sensors?

The newer (ie nowadays) Garmins have a ANT+ sensor which elastics onto the (normally rear) hub which outputs the rate of rotation (of the wheel - ie rpm (say)). This is the same as the differential by time of the reading you'd get from a magnet on the spoke / sensor on the fork set up. I referred to how it calibrates itself in my post above. The other type of sensor you can add (leaving out power meters and the like) is a cadence sensor - which elastics onto the left crank. Some Garmins are sold 'bare' others come bundled with these sensors.

 

[HLaB]

The 'speed sensor' calibrates itself by assuming that the GPS speed being recorded is accurate. It's not calibrating itself against 'real time speed' - how could it? [Nonetheless useful if you need to constantly refer to speed and there's intermittent overhead obstruction.]

AFAIK it calibrates it off a GPS reading over a longer length to determine the correspondence between one wheel revolution distance and hence speed is calculated. Once calibrated it means a stable speed reading (easier for pacing) and a more accurate reading as a gps unit will draw a straight line between drop outs whereas the road may actually twist a bit and the speed/distance unit will feed back to the GPS. The difference in most cases will be negligible I think but on a sportive in London there was a lot of long tunnels with bends and it was a fair bit out on distance.
You can of course put a measured wheel circumference into the unit and eliminate the need for GPS entirely but why would you as its more complicated and probably less accurate

 

[Alien8]

Determining velocity by differencing adjacent position fixes will be significantly less accurate than solving for it directly from measured range-rates.

I don't know what your average bike receiver software will be doing, but I would like to think they all use range-rate, thus the route you take between position fixes will be irrelevant, as far as velocity determination is concerned.

Of course the data you see displayed will be filtered by some algorithm, and possibly aided by additional sensors, such as a wheel sensor.

All things being equal, I'd like to suggest that a GPS measured velocity will be typically more accurate than one measured by counting just wheel revolutions.

 

[Drago]

But as aforementioned, range-rate between two points on an undulating surface will differ between a line drawn in 3D space between the same two points. Even assuming both devices were 100% accurate the results will still differ, because they're measuring different things.

 

[Banjo]

GPS is extremely accurate but individual gps units are adjusted to slow down the time between updating to a useable rate so the number isn't constantly changing on the screen.