Top speed (really?)

[G3CWI]

I often see people publishing their top speeds here. But how accurate are they? I use Endomondo on my iPhone and the reported top speed is always faster than any speed I can find in the workout (sometimes by a large amount). I think that in my case the reported top speeds are often the result of slight GPS errors and poor averaging and thus I ignore them. The ones on the Endomondo graph are much more modest but far more believable. The effect is far more obvious when walking where the top speed is often completely ridiculous.

 

[Shaun]

Personally the only top speed I would trust is the one from my bike comp. The one that's getting its data directly from my bike. I've always considered GPS to be a good guide, and useful for the purpose of plotting / positioning, but I'm not sure it can account for subtle changes in speed, gradients, etc. as well as a directly-attached-to-the-wheel bike comp.

Having said that, I've never owned a high-end GPS device - they may be more accurate than I'm giving them credit for.

 

[ASC1951]

I don't know about 'high end', but I've tested my Garmin 800 against motorway mileposts in the car and it is very accurate on both speed and distance. Less so for ascent/descent, but that is inherent in the way the triangulation works.
Unlike a GPS, bike comps are only accurate if you set them up correctly for tyre size; but I found both my Octos and the Garmin as accurate on speed as any sensible person would ever need.

 

[MontyVeda]

When i had a bike comp... i was always dubious about the top speed (and overall distance). The first one required me to measure the outside circumference of the tyre in mm, which i guess was far more accurate them my 2nd bike comp, where one selected 26".

With the first comp, if my measuring was a mm or two out... how much effect would that have on the speed reading? And the second one, being an average 26" that doesn't take in to account the type of tyre so the assumed circumference could be a good inch out.

Either way... i find not having such data any more rather liberating.

 

[srw]

I don't know about 'high end', but I've tested my Garmin 800 against motorway mileposts in the car and it is very accurate on both speed and distance.

There is no inconsistency between this statement and the statement that instantaneous high-speed readouts might be wrong. Over a mile you've got dozens or hundreds of readings (depending on your settings). If one of a single pair of readings is out by a few feet (which is well within tolerance) you could get a speed difference of several mph. And, ironically, the more "accurate" your GPS is - i.e. the more readings it takes - the bigger the speed error.

With the first comp, if my measuring was a mm or two out... how much effect would that have on the speed reading? And the second one, being an average 26" that doesn't take in to account the type of tyre so the assumed circumference could be a good inch out.

As it happens I have the instructions for one of my bike computers on the table in front of me. It tells me that a 700 x 25c tyre has a circumference of 2135mm, so an error of 2 mm in circumference will result in a distance/speed error of 0.1%.

Which reminds me, I've got a bit of Pythagoras to do to test my theory about the difference between a bike computer distance and a GPS distance. (The computer measures the hypoteneuse, the GPS measures the horizontal side of a right-angled triangle. By looking at the height climbed reported by the GPS I should be able to work out whether the discrepancy is reasonable or whether my computer is miscalibrated.)

 

[JtB]

I use Endomondo, and the combination of winding country lanes / GPS sample rate results in plots where most of the corners are shown as cut. As such the final distance and hence my overall average speeds reported are less than the actual ones.

 

[Rob3rt]

Accurate enough.

 

[Night Train]

The computer on my Brompton is a cheap Aldi one. I was never really sure it was accurate but the speed read the same as Arch's computer when at around 10-12mph. The real test was when I did my half century. The computer told me I did 52miles and Arch's calculations on the map was within 1/2 mile, so I reckon my computer is close enough.

Besides, the only one I am competing against is myself so as long as the top speed, or distance, is higher then last time then I am happy.

 

[Moon bunny]

There is no inconsistency between this statement and the statement that instantaneous high-speed readouts might be wrong. Over a mile you've got dozens or hundreds of readings (depending on your settings). If one of a single pair of readings is out by a few feet (which is well within tolerance) you could get a speed difference of several mph. And, ironically, the more "accurate" your GPS is - i.e. the more readings it takes - the bigger the speed error.



As it happens I have the instructions for one of my bike computers on the table in front of me. It tells me that a 700 x 25c tyre has a circumference of 2135mm, so an error of 2 mm in circumference will result in a distance/speed error of 0.1%.

Which reminds me, I've got a bit of Pythagoras to do to test my theory about the difference between a bike computer distance and a GPS distance. (The computer measures the hypoteneuse, the GPS measures the horizontal side of a right-angled triangle. By looking at the height climbed reported by the GPS I should be able to work out whether the discrepancy is reasonable or whether my computer is miscalibrated.)

Which reminds me,I have a life to live.

 

[rvw]

We (srw and I) are also absolutely certain that our GPS is occasionally dead wrong on speed. It has registered us doing 0.00 mph on a number of occasions when we have been freewheeling downhill - but overhanging trees have got in the way of the signal.

 

[HLaB]

If you've got a cadence/ speed sensor fitted with the garmin I think it'd be very accurate indeed.