TheRCF

I seem to be the odd person when it comes to the distance my computer gives.

My rides are very consistent. I have certain points where I know the computer will click over to the next digit and it will be the same every ride, within a few feet.

But I've come across riders who get bigger numbers for the same territory. In fact, when we had the Honolulu Century, I showed less than 96 miles (might have been about 94). This struck me as a pretty big error!

Some time ago, I measured my wheel circumference very carefully. I sat on the bike while rolling it. I put a piece of tape on the tire to mark the spot more exactly (instead of using the valve stem which would be further away, possibly making eyeballing the result less accurate, and I did it repeatedly to be sure the number was always the same. Since then, whenever I get new tires, I don't measure it - I just do a normal ride and see if the numbers click over at the same key points (all are at least 9 miles away).

But with the century difference and a couple other cyclists I've checked with getting different numbers, I'm wondering who is right.

I think I am, because of the care I took, but also because I have some mapping software and also checked online at Google's gmap pedometer and they give virtually identical results to mine.

But I really don't know if these mapping sources are that accurate or not. Anybody know? They give figures to the hundredths so if they are accurate, then my computer is right.

2manybikes

There are to many variables to get it exact. The list is long. Especially over 100 miles. Don't expect bike computers to match exactly. Don't expect car speedometers to match exactly either.
Who says a century ride is exactly 100 miles? It's close and then to a convenient place to stop. The people arrowing the century are unlikely to have anything more accurate than a car trip meter to measure the distance anyway. There's no reasonable way to get it perfect even if someone tried.

TheRCF

> There are to many variables to get it exact. The list is long. Especially over 100 miles. <

I don't know what variables you are referring too, but like I said, on my rides, I'll hit distances of 10, 15, or more miles almost exact - literally a variation of feet. This surprised me since I figured that on each ride, the exact path my wheel would follow on the same route would vary. But either those variances are really small or, on average, they cancel each other out

> Don't expect bike computers to match exactly. <

I wouldn't, but it seems a 5% difference or more is pretty big to me. And since it turned out I got within 0.03 over 8.6 miles (based on the map software which I hope is right), it seems it is possible to be much less than 5% with a little effort.

> Don't expect car speedometers to match exactly either. <

But there is a reason for that. Manufacturers consistently make them read high so that you can't gripe to the company for speeding because of inaccuracy. So they make it inaccurate in the opposite direction. How much varies, of course, depending on what the company does.

> Who says a century ride is exactly 100 miles? It's close and then to a convenient place to stop. <

Well, I guess that is true, but I do think a century should be AT LEAST a century. Just my opinion - and if it isn't, they should say so. I made it a point to ride a little before the start just in case. I wanted my century to really be a century.

> The people arrowing the century are unlikely to have anything more accurate than a car trip meter to measure the distance anyway. There's no reasonable way to get it perfect even if someone tried. <

Well, I can't agree with that part. For one thing, people running these things certainly are or have access to people who do long rides. A little care on measuring the wheel would give a far more accurate result than any car I know. Also, IF these maps are truly pretty accurate, that would take care of the measurement. Since the map almost exactly matches my numbers after careful tire measurement, I tend to think the maps are accurate.

And remember that many centuries are regular events over the same course. You only have to get it right once. But I suspect they sometimes do a sloppy job and it lasts forever!

But I could be wrong, which is why I'm asking for other people's experiences. Maybe mine is a coincidence.

supcom

Unless you have a known calibrated course against which you can test your computer, you have no way of knowing whether your computer is more, or less accurate than anyone else's. Repeatability is not the issue. Any computer working correctly should give repeatable results if the bike is ridden over the exact same path.

Mileages given in online mapping tools are not good sources of high accuracy. The road data used by these programs is made up of many straght-line segments that are not surveyed with high accuracy. They don't need to for road navigation, so they aren't. They are certainly close enough for cycling use.

BTW, I believe that my cycle computer is absolutely dead on 100% accurate under any and all conditions. When my computer reads 100 miles, I KNOW I have done a century ride. It's just fact until someone proves me wrong.

TheRCF

> Unless you have a known calibrated course against which you can test your computer, you have no way of knowing whether your computer is more, or less accurate than anyone else's. <

Well, that brings us back to the question of how accurate these maps are (Topo USA and the online gmap pedometer). Both give results within 0.01 of each other for about an 8.6 mile course which is also within 0.03 of my computer. If the maps are very accurate, than so is my computer. But it would be nice if some other people, maybe using GPS, especially on a section of straight road, have compared that with map software, just to be sure.

> Repeatability is not the issue. <

That was something I learned and just mentioned for completeness. But before experiencing it myself, I did expect a little variation - not a lot, but more than I get.

>Mileages given in online mapping tools are not good sources of high accuracy. The road data used by these programs is made up of many straght-line segments that are not surveyed with high accuracy. <

I can see that. In fact, when using the gmap stuff, you can't just mark the start and end on a given road because it just results in a straight line. You have to create these little straight lines yourself and I did a bunch of them to minimize inaccuracies. This may account for the small difference between my map software and it - but it was only 0.01 difference anyway!

The question isn't really whether it is perfect, but is it extremely close? I would be very happy if it was within 1%. I have a feeling it is since all three of my results (two maps, my computer) are all so close. That would be a big coincidence!

> BTW, I believe that my cycle computer is absolutely dead on 100% accurate under any and all conditions. When my computer reads 100 miles, I KNOW I have done a century ride. It's just fact until someone proves me wrong. <

LOL!

slvoid

Just do a damn roll out riding rather than pushing your bike.

TheRCF

That's how I did mine - I measured the distance while sitting on the bike and one of the reasons I think mine is pretty darn accurate.

But as I talk to people who get different results, they ALL are convinced theirs are correct. It's also darn tough to get them to say how they got their calibration numbers. Mostly they just restate that they know they are right, it is the same as their friends, etc.

But just today I got one to tell me that he used the suggested figures in the instructions. Well, of course that isn't going to be as accurate, yet he spent days telling me he was positive his was correct, but all he did was accept someone else's numbers who couldn't know what tires he had on his bike!

Retro Grouch 

6% short seems about right to me for a course that was measured using a car odometer. Every time you go around a curve the car takes a sligltly wider path than you do with a bicycle or while running.

huhenio

Unless you are racing on a track ... accuracy will suffer. Do not dwell on it: your bike odometer is probably more accurate than your car

Derailed

My wife's computer, compared to my own, is consistently within 1% of mine when we ride together. For example, if my computer says 10.0 miles, hers will be 9.9 or 10.1 (if not dead on). This holds for longer rides as well.

I calibrated both using the rollout method, with the rider of the bike sitting on the bike for the measurement. I aplogize if you've already stated clearly that you do this, but I use a long enough tape measure that I can get 7 or 8 revolutions for the measurement. I run it all the way down our driveway and then do the rollout. This reduces errors that may come up if you do just one revolution due to the exact position of where you start and stop.

Is it possible your magnet is not properly placed, or there is a malfunction in your computer that causes it to not count every revolution? There can of course be some error due to tire wear, but this will be small. Tire pressure is likely a more significant factor, but you could place bounds on this fairly easily: deflate your tires about as far as you could imagine letting them go before you're sure you'd add air, do a rollout, and then pump them up as far as you could imagine yourself going and do another rollout.

An interesting test would be to install two computers on your bike, and give each the same calibration. They should record exactly the same distance unless one is not counting every revolution.

If you've done a careful rollout, you're using the same tires at reasonable pressure, and your computer is installed and working properly, I don't think you'll find a more accurate way to measure the distances of your rides.

Good luck!

2manybikes

The diameter of the wheel changes if you change the pressure or the weight on the bike.

Same for a car. Not to mention changing tire sizes.

You can't get three roll outs to be exactly the same if you spent your life on it. What was your tolerance variation on your rollouts? + or - .005 ?

Tire diameter of a marked size varies from manufacturer to manufacturer.

If the person arrowing the century has a 2.5% error and you have a 2.5% error it could be a 5% error. It's accumulation of all the various errors that add up.

Volunteer to arrow some centuries for a bike club and learn how it is done by experience. You're assuming a lot with no information.

What they said about maps and altitude.

Some bike computers only calibrate in increments of 5mm some go down to 1, how do you deal with fractions? Some bike computers just have settings for tire sizes and nothing in between. They know it can't be perfect. How do you know the arrower was not using one of those. One of my bike computers has a setting for 700 x 20 and then it jumps up to 700 x 25. I have used 700 x 23 And also 700 x 26 on that bike. It only calibrates in increments of 5. **********???

Did you go straight down the road with no wobbles? Did you measure the difference? Did the person arrowing not pull over or add to the mileage by turning around for a short way, because they missed arrowing an important corner? How in the world do you know? It happens. If you miss arrowing a corner in the car and have to go back how do you reset the car trip meter by 300 feet?

The list is endless. The more you know the longer the list.

Derailed

Even if your computer only allows increments of 5mm for the circumference, this is not a big deal: for reference, say the circumference of your inflated, weighted tire is 2,150 mm (typical for a 700x30 tire). +/- 5mm on that is only +/- 0.2 %. Fractions of a mm, of course, are even less significant. Furthermore, that circumference implies a radius of ~342 mm; suppose you do your rollout when your tires are fully inflated: to exceed 1% error, your tire radius would have to decrease by more than 3 mm. If you're riding on road tires, I doubt you'd tolerate that much deflation before you'd add air.

My point is that even with all the sources of error, if you do a careful rollout, your measurement should be within 1% of the actual distance you ride (perhaps greater if you use mtn bike tires, or don't mind riding when your tires are severely deflated).

eubi

...or use a GPSr

supcom

You seem to be obsessed with this. The conventional wisdom is doing a rollout with tires properly inflated and your weight on the bike is the simplest method to get a reasonably accurate calibration. If you had a know accurate test mile along a highway, you could use that and possibly get a slightly better accuracy, but it's not really worth the trouble.

As far as other people go, what difference does it make that their computers are calibrated differently than yours?

sngltrackdufus

If you can get it to +/- .01 mi., regardless of how you calibrate it that is good. Both of my computers register that on a measured mile & i don't give it any more attention.Of course i might be losing/gaining a mile or so every 100mi. but hey, life's just not fair.

NomadVW 

Yes, where any given reading may be off as little as 10-12 feet on the move (and not always in the same direction will it be off) or as much as ... well.. a lot. This is a good answer if you live in flat-to-the-horizon areas, but will only by chance be as accurate as a well-tuned cyclocomputer for most of us. That's not to say my MTB doesn't have one on the front, but for road biking, I prefer to have my measurements guaranteed to continue working even when I go around that street between the houses or around the corner along the side of the mountain.

VW

ghettocruiser

Maybe you’re holding a strait line and your buddies are waving all over the road.

You can't use a GPS to accurately measure a short distance. But on a long ride your 3m accuracy (i.e. 3m error) for instance, becomes pretty minimal, as in is non-cumulative. Taking the GPS on a long ride is the way to go for calibrating. Just check the GPS internal tracklog file afterwards (most of them have this now) to make sure it didn’t lose the signal when you went under a bridge or something.

There is a new Garmin unit that uses GPS signals AND a wheel magnet to automatically give a composite readout. On my list one of these days, if it’s not too huge.

TheRCF

> The diameter of the wheel changes if you change the pressure or the weight on the bike. <

I know - which is why I sit on the bike while doing a rollout. As for pressure, it seems it would have to be fairly significant - far more than I would allow. I've ridden my bike after just pumping it up and after it was not inflated for a couple days. I know it loses a little pressure, but based on my rides, the difference is hardly noticeable, if noticeable at all.

> Not to mention changing tire sizes. <

Yes. That's why I don't rely on those recommended setting in the instructions.

> You can't get three roll outs to be exactly the same if you spent your life on it. What was your tolerance variation on your rollouts? + or - .005 ? <

Well, don't know what you mean by "exact", but my results were so close to the same that I couldn't really see any difference.

> Tire diameter of a marked size varies from manufacturer to manufacturer <

This is why I worked real hard to get one set of tires correct. Then I would actually ride 10-20 miles and notate certain spots where the number on the computer changed. Since I found it was so consistent, I could then take new tires and, rather than calibrate it with a rollout, just do my normal rides and see if the numbers changed the same places - then adjust the calibration if necessary.

> If the person arrowing the century has a 2.5% error and you have a 2.5% error it could be a 5% error. It's accumulation of all the various errors that add up. <

Sure - but the pressure issue doesn't apply since I top the tires off before testing or measuring any important ride. The difference in tires or variation within a tire line doesn't matter because I would have already made the adjustments.

> You're assuming a lot with no information. <

Maybe. Which is why I'm posting. I did ask a guy who said he set up the century - and he didn't recall how he did it!

> Some bike computers only calibrate in increments of 5mm some go down to 1, how do you deal with fractions? <

Mine goes down to 1 and any fraction of that is so small, it really doesn't matter. After all, you could only be off by half a mm - what is that, 1/4000th?

> Some bike computers just have settings for tire sizes and nothing in between. They know it can't be perfect. How do you know the arrower was not using one of those. <

Well, I don't, but that would just mean he would more likely have larger errors.

> Did you go straight down the road with no wobbles? <

I don't think this matters as much as people may think, especially if someone has the magnet on the rear wheel (because it tends to not move out of a direct path as much as the front wheel). I did a ride for 10-12 miles where I deliberately weaved all over the lane (when there was no traffic). The difference was only 0.1 or 0.2 for the whole trip as I recall.

> Did the person arrowing not pull over or add to the mileage by turning around for a short way, because they missed arrowing an important corner? <

Could be, but, personally, I wouldn't be laying out the route at the same time as putting up arrows. I would have laid out the route first, well in advance.

TheRCF

If it can be done, I prefer more accuracy over less accuracy. If I do a century, I want to be sure it is really a century. Other people I ride with often talk about how fast they did this or that ride. Just this week, I found one of them was likely 10% off! So, if he was going 20 mph, he was really going about 18 mph. I don't know about you, but I find a speed increase of even a couple of 10ths to be pretty significant progress, so it is important to know what your numbers really are.

TheRCF

I agree. But if I can get it closer, I'll go for it. That's why I'm interested in the accuracy of these maps. If they are accurate, especially over straight lines, then I can use them to my advantage.

TheRCF

That sounds pretty cool.

MMACH 5

Go for a ride pushing one of these...

MMACH 5

This is true as long as you don't go up and down too many hills. The GPS can't account for the extra vertical distance, (i.e. - If you fall from a 40 story building, you'll travel about 400 feet and be moving at a rather high rate of speed; the GPS will show a distance and speed of zero).

geraldatwork

Your bike computer, especially the way you calibrated it, is probably more accurate than they way the people running the century measured it. For example if they used a car to measure it makes sense that it came out shorter as many car companies have their speedometers very slightly inflated. One of the reasons for doing this is it eliminates a few miles the manufacturer is responsible for on warrantees. For example when your car shows 36,000 miles it really has only about 34-35,000 miles. they save millions of dollars. Recently when my car hit 36,800 miles I had some transmission problems that eventually came out to around $1500. My dealer wouldn't cover it saying it was way past the 36,000 miles. I threatened to have it verified and said I would sue them. They seemed to know what I was referring to and said would get back to me. The next day said it would be covered under warrantee. Check your car against your bike or one of those speed checkers you will find your speedometer is off by a few percent. Oh BTW you didn't do a century. Go back and re-do it with the extra 6 miles.

spinnaker

Not necessarily true. GPS devices can measure changes in altitude (be it not hugely accurate). It would depend on the GPS if it used this change in altitude to compute it's speed. My guess though is that most (if not all) do not use it. But sorry I really don't feel like testing my GPS.