Electronics, Lighting, & Gadgets - confused about how to calibrate bike speedo?

I have a bell 4 function dig speedometer and am trying to figure out what i should enter for the wheel factor.

My front tire says 26 x 1.95

In the manual that came with speedo theres a chart and the closest i can get to a 26" wheel diameter with a 1.95 width is the following

26" (650A) (= wheel factor of 2073) ..or...
ATB 26" X 1.75 (=2045) .. or...
ATB 26 X 2 (650B) (=2099)

( I have no idea what "ATB" or the "A" or "B" stands for, i was hoping someone who knows bikes could discipher?)

In the manual it says that if your wheel size is not listed in the chart, you just need to convert wheel size to mm, then multiply by pie.
But how does wheel width factor into this?

Also, what math are they doing to arrive at those wheel factors?
I can understand the first one (26 x 2.54 x 10 x pie = 2074.71), but have no idea how tire width factors into the wheel factor?

Also, the wheel diameter 26.6" (700 X 25C) is highlighted in bold in the manual for computer, as if that is a common wheel size or something..

Thanksfor any clarification, i'm totally confused.:(

Your speedometer is nothing more than a counter/timer. When you calibrate your speedometer you are effectively entering the circumference of the tire. Your tire circumference is the distance traveled for 1 revolution of the wheel. By counting the distance as a function of time you get your speed.

The most accurate way to calibrate your speedometer is to use the roll-out method. This is more accurate than just entering tire size since the circumference of your tire is subject to manufacturing variations, your weight and the corresponding pressure you fill your tires to.

By the way, the width of your tire doesn’t factor at all in your computer’s calibration.

Start by filling your tires to the pressure you plan to ride with (tire with sensor mounted should be accurately filled).
Measure the distance that you travel for at least 1 revolution as accurately as possible. Suggest that you use the valve stem as an aid to determine revolution as well as mark distance. Best accuracy will be obtained if you can sit on the bike (since your tire will deform as a result of your weight) as well as using a very long tape measure and average several revolutions.

If you measured distance in inches multiply by 25.4 to get millimeters. This distance is the same as your tire circumference.

26" (650A) (= wheel factor of 2073) ..or...
ATB 26" X 1.75 (=2045) .. or...
ATB 26 X 2 (650B) (=2099)

You are probably worrying more than you need to.

The two widest spaced numbers there are about 2.5% different.

This means that if you put in the largest number (2099) and you should have used the smallest (2045) you will have about a 2.5% error. If this is the case, the impact of the error is that when the computer says 20.0 you are actually riding at about 19.5 Or if you ride a century of exactly 100 miles on the computer you really only rode 97.5 miles... Nothing to worry about unless you are going for a world record or have other needs that dictate extreme accuracy.

If you are concerned about the greatest possible accuracy, then the best way is the roll out method, but most people don't need that level of accuracy. If it is that critical you should also recalibrate occasionally to accommodate tire wear.

The width of the tire matters because the width is the same (or similar) to the additional radius caused by the height of the tire on the wheel... this changes the effective radius and therefore circumference which is used for speed calculations. For example, with tires for the same 26" rim, a 1.5" tire will have a shorter outer circumference than a 2" tire.

For your 26 X 1.95, I would put in a number slightly less than 26 X 2 since this is a difference in radius of .05 ... probably something like 2091... This should be accurate enough. 2099 - (2 * 3.14 * .05 * 25.4)

What LD said
if you want to be exact, your 26" (nominal tire) rim is actually less than 26", probably somewhere around 22" - 23". The reason I can't give you a precise answer is that the 26" designation is used for all types of actual rim sizes and I'm not familiar enough to tell you which one you have, but my best guess is a 559mm. Take that number, add the 1.95" twice, so you have 559mm+2*(1.95in*25.4mm/in)=658.06mm tire rolling diameter * pi (I love to eat pie, too) = 2067mm tire rolling circumference.
The 650A and 650B wheel sizes are old french size wheels that you may never see.

Get some bright orange poster paint. paint a thick dot on the wheel. Ride the bike when the paint is still very wet. There will be marks on the ground. Measure the distance between the marks, center to center. That's your wheel size.

Your "26-inch" wheel is most likely a current mountain bike (aka ATB) rim or an ISO 559.

Using that as the diameter plus 2x tire width (it's safe to assume bike tires are pretty much circular in cross section) gives a diameter of 559 + 2*(1.95*25.4) = 658 mm The circumference will be Pi times 658 = 2067.

Get some bright orange poster paint. paint a thick dot on the wheel. Ride the bike when the paint is still very wet. There will be marks on the ground. Measure the distance between the marks, center to center. That's your wheel size.

This would actually be a fairly accurate rollout. Rollouts are not accurate without the weight of the rider.
Accuracy would be improved if several marks could be made and an average of the separation used as the wheel circumference.
Interpolating between the 26 X 1.75 (2045) and 26 X 2 (2099) would be good enough, probably better than most people's rollouts.
A good way to confirm your calibration would be to ride known distances. I use surveyed section road intersections that are exactly 1 mile apart. But I agree with Little Darwin, it's really not worth worrying about.

Al

but at what tire pressure will the rolling radius be correct?

but at what tire pressure will the rolling radius be correct?

Where ever it was when you did the roll-out calibration. Usually you set the pressure to your desired level and calibrate it there.

And, yes, we are agonizing over minutiae unless you are trying to set officially recognized records.

"650A" and "650B" are names that refer to specific rim bead seat diameters. I would guess that "ATB" means all-terrain bike.

This page probably has more than you ever want to know about various rim/tire size designations :)http://sheldonbrown.com/tire-sizing.html#isoetrto

A quick way to get a good approximation for tire circumference is to simply add twice your tire width to your rim's bead seat diameter and multiply that result by PI. For 26" mountain bikes, rim bead seat diameter is 559mm. So 26x1.95 would have a circumference of about (2*(1.95*25.4)+559)*3.14159 = 2067mm. Subtract about 5mm or so if you want to account for the fact that the tire compresses during riding.

The most accurate way to calibrate your speedometer is to use the roll-out method.

Wrong.

The most accurate method is to ride a measured distance with a rough calibration, then compare the distance on the computer with the known distance. Divide the known distance by the computer's distance, and use the result of that division to multiply your computer's wheel circumference setting. The new value is the correct wheel circumference.

Of course, this value varies slightly with bike load (the bike's weight, your weight plus the weight of all your clothes, stuff and the stuff on the bike), tyre pressure (which decreases with time and also depends on temperature differences) and other minor factors.

Using this method on my newest bike, I now have an error of just over 0.1% for distance. Pretty damn good, if you ask me. That's about 600 ft wrong after a century!

Wrong.

The most accurate method is to ride a measured distance with a rough calibration, then compare the distance on the computer with the known distance. Divide the known distance by the computer's distance, and use the result of that division to multiply your computer's wheel circumference setting. The new value is the correct wheel circumference.

Of course, this value varies slightly with bike load (the bike's weight, your weight plus the weight of all your clothes, stuff and the stuff on the bike), tyre pressure (which decreases with time and also depends on temperature differences) and other minor factors.

Using this method on my newest bike, I now have an error of just over 0.1% for distance. Pretty damn good, if you ask me. That's about 600 ft wrong after a century!

Actually, neither this method nor the paint dot method is particularly accurate because you can not ride a perfectly straight line. Although logic would suggest that small movements off a perfectly straight line would make little difference, that logic is incorrect.

Practical example (I'm not going to buy into the maths and theory and all that good stuff, we're talking pushbikes, not property boundaries):
Years ago, I served time in a survey instrument supply firm (between jobs using the things). One of our products was a measuring wheel - wheel on a stick, counter attached. Common and useful gadgets. We sent one off to a local council run by one of life's compulsive fiddlers. This bloke decided to 'check' his wheel's accuracy (he was bored and curious, not being nasty). Anyway, he rang up suggesting the wheel was over 10% out. We were concerned ... as you'd expect. So I laid out a 100m steel tape, pulled it tight and straight, then ran the measuring wheel along it. Just walking normally, doing my best to keep it on a straight line (and succeeding pretty nicely I would have thought) but not overly worrying about it - this produced that 10% error. However, taking it very slowly and being ultra careful in keeping that wheel on the steel tape, gave near enough to zero error. Me mate up athe council ran the same test and got similar results. Moral, it takes very little deviation from a straight line to introduce noticable errors in distance measurement.

The most accurate measurement is going to be from a rollout over one revolution (less chance for error to affect the result) ... after you've found out some way of allowing for rider weight, tyre wear, position of tongue in mouth, etc. And what happens when you strap your laptop and lunch onto your bike (therefore increasing the weight and flattening out the tyres more)?

Hey, it's a pushbike computer - such discussions are interesting over the morning coffee and sticky bun, but in the real world ... ?

Richard

Actually, neither this method nor the paint dot method is particularly accurate because you can not ride a perfectly straight line. Although logic would suggest that small movements off a perfectly straight line would make little difference, that logic is incorrect.


Riding a known distance becomes a much more accurate measurement if the computer's sensor is on the rear wheel. I can dial mine in to a 0.5% accuracy easily if I want to go to the trouble.

Actually, neither this method nor the paint dot method is particularly accurate because you can not ride a perfectly straight line. Although logic would suggest that small movements off a perfectly straight line would make little difference, that logic is incorrect.

Practical example (I'm not going to buy into the maths and theory and all that good stuff, we're talking pushbikes, not property boundaries):
Years ago, I served time in a survey instrument supply firm (between jobs using the things). One of our products was a measuring wheel - wheel on a stick, counter attached. Common and useful gadgets. We sent one off to a local council run by one of life's compulsive fiddlers. This bloke decided to 'check' his wheel's accuracy (he was bored and curious, not being nasty). Anyway, he rang up suggesting the wheel was over 10% out. We were concerned ... as you'd expect. So I laid out a 100m steel tape, pulled it tight and straight, then ran the measuring wheel along it. Just walking normally, doing my best to keep it on a straight line (and succeeding pretty nicely I would have thought) but not overly worrying about it - this produced that 10% error. However, taking it very slowly and being ultra careful in keeping that wheel on the steel tape, gave near enough to zero error. Me mate up athe council ran the same test and got similar results. Moral, it takes very little deviation from a straight line to introduce noticable errors in distance measurement.


I guess you didn't think things through before you posted. Precisely BECAUSE you can't ride in a perfectly straight line, "my" method is the most accurate, since it averages out and includes the general error that a non-straight movement causes. And it's not anywhere near 10%. The difference between my initial setting based purely on ETRTO wheel+tyre size and the corrected setting based on a 40 mile ride was 1.2%.

Wow. I',m now speedo calibration scientist. ha ha
Thanks for the help guys.

Wow. I',m now speedo calibration scientist. ha ha
Thanks for the help guys.

Now we want a ride report on your next ride to the nearest 1/100 of a mile!

:D

I guess you didn't think things through before you posted. Precisely BECAUSE you can't ride in a perfectly straight line, "my" method is the most accurate, since it averages out and includes the general error that a non-straight movement causes. And it's not anywhere near 10%.

+1

I guess you didn't think things through before you posted. Precisely BECAUSE you can't ride in a perfectly straight line, "my" method is the most accurate, since it averages out and includes the general error that a non-straight movement causes. And it's not anywhere near 10%. The difference between my initial setting based purely on ETRTO wheel+tyre size and the corrected setting based on a 40 mile ride was 1.2%.

Actually “your” method is not really more accurate. If you did the roll out "properly" you should have arrived at the correct value for your circumference with a lot less effort. Your method can be made more accurate if your initial circumference guess is closer to the true value, or if you do the same ride multiple times and perform your scaling correction for the circumference after each ride.

If your method works for you, then stick with it.

Actually “your” method is not really more accurate. If you did the roll out "properly" you should have arrived at the correct value for your circumference with a lot less effort. Your method can be made more accurate if your initial circumference guess is closer to the true value, or if you do the same ride multiple times and perform your scaling correction for the circumference after each ride.

If your method works for you, then stick with it.

You didn't understand what I wrote, I take it.

A rollout will give you an accurate reading for a perfectly straight distance. The problem is that we don't ride like that. My method accounts for how we actually ride.

And the initial setting used with my method doesn't affect accuracy at all. I could've entered my wheel size as being 35" or 15" and still end up with an accurate calibration at the end.

Actually, I did understand what you initially wrote. The method that you are discussing is a method of successive approximations that will eventually lead you to the correct answer. Try doing the math and setting up the equations numerically, then maybe you will understand. Your method is still based on performing a rollout where you contend that your distance measurement is more accurate...it really isn't since as you, as well as several previous posters have indicated the difficulty of measuring a linear distance. Your method will eventually converge on the correct solution if you followed what I wrote.

No, now I KNOW you didn't understand.

There is no convergence to speak of. You simply set it approximately correct the first time to get a usable value to modify after the calibration ride, and what you set this to doesn't matter.

Then, you ride a known distance (a long distance - in my case I use rides of about 30-60 miles) and use my calculation above to find the correct circumference setting.

My method, as I've described it, provides the best possible calibration straight away after the calibration ride, and will provide a much more accurate calibration than any form of simple rollout can ever hope to accomplish.

CdCf: I am curious as to how you know that your 30+ mile "known" course is the actual distance you're using for your calculation. I'm not trying to be argumentative, I just wonder how you know where your starting line and finish line are on a course that long.

The only idea I have is on a highway with mile post numbers on it, and you start at one mile post sign and end at one 30+ miles down the road.

I get my distance data from our national road database (in digital form). The distances are measured very accurately. I can get distance data down to +/- 5 m (15 ft), if I want to. But I settle for the nearest 100 m (300 ft). That is, for a distance over 50 km (~30 miles), for example, I use the first decimal. So 50.0 in this case. That allows an accuracy down to around 0.2%.

How about using one of those GPS devices on several rides to get the distance, then doing th emath for your correction ?

CdCf: I am curious as to how you know that your 30+ mile "known" course is the actual distance you're using for your calculation. I'm not trying to be argumentative, I just wonder how you know where your starting line and finish line are on a course that long.

The only idea I have is on a highway with mile post numbers on it, and you start at one mile post sign and end at one 30+ miles down the road.

These can be very inaccurate, depending on where you live and the specific road.
I've driven I-84, along the Columbia River. The total distance between Umatilla & PDX seems pretty close, but in between, I've seen the MP's vary by 2 miles vs my odometer.

As far as GPS, what if you ride in a big circle and end up where you started? It seems to me that they would only be accurate if you rode in a perfectly straight line. Of course I've never owned a GPS, so I could be....

GPS has a fairly poor accuracy. You'd be far off.

I get my distance data from our national road database (in digital form). The distances are measured very accurately. I can get distance data down to +/- 5 m (15 ft), if I want to. But I settle for the nearest 100 m (300 ft). That is, for a distance over 50 km (~30 miles), for example, I use the first decimal. So 50.0 in this case. That allows an accuracy down to around 0.2%.

How do they calculate the distances? If they are measuring down the centre of the road and you are riding at the edge what happens on a turn? You will travel less or further than the measurement depending on the direction of the turn. Might be pretty hard to find a route that turns to the left the same amount that it turns to the right.

I've noticed using a gps to make maps (http://www.openstreetmap.org/) that a cyclists route down a road is very difficult to describe. Corners are cut, pot holes and traffic are avoided and many other deviations. These various deviations vary greatly for different roads and often even a different ride down the same road.

This thread is rapidly turning into a "how many angels can dance on the head of a pin" kind of agonizing over minutiae.

Set the damn cyclometer to 2065 (or 206 if it's only got three digits) and be done with it. The calibration will be accurate within less than 1%.

How do they calculate the distances? If they are measuring down the centre of the road and you are riding at the edge what happens on a turn? You will travel less or further than the measurement depending on the direction of the turn. Might be pretty hard to find a route that turns to the left the same amount that it turns to the right.

I've noticed using a gps to make maps (http://www.openstreetmap.org/) that a cyclists route down a road is very difficult to describe. Corners are cut, pot holes and traffic are avoided and many other deviations. These various deviations vary greatly for different roads and often even a different ride down the same road.

Again, BECAUSE the bike doesn't move in a straight line on the road, this method is better, because it takes into account, automatically, all these variations. If I ride a measured course that's 30.00 miles, my wheels may have moved 31.62 miles, but that's irrelevant if you're interested in how far you've actually travelled. If you know you're going to ride from city A to city B, and the distance between them is 74.19 miles, you want your computer to show 74.19 miles when you're there. Not 76.22 or anything else.

Or maybe that's just me...

If you are riding around the Sacramento area, the Yolo Causeway is exactly 3.11 miles from levee to levee.

How would it make a difference whether the tire was compressed or not when you measure it? It still has the same circumference on the surface, it just compresses and releases.

Measure the straight rollout for one or two revolutions on the floor, and just plonk that in. You aren't going to be judged on the exact exact precision of the cyclometer.

How would it make a difference whether the tire was compressed or not when you measure it? It still has the same circumference on the surface, it just compresses and releases.


Because the tyre isn't a rigid disc, it will deflect under load, and in doing that, its effective radius will decrease. It is the radius between the hub and the contact point (under load) that determines the effective wheel circumference.

I thoroughly agree with HillRider that this thread is definitely getting out of hand. I also agree with CdCf that he is able to calibrate extremely accurately for a know distance given that you were somehow able to ride in a perfectly straight line. The problem I have with CdCf’s method is the statement that it is the most accurate method. Unfortunately, for a long distance you can’t ride in a perfectly straight line, and while it does take into account how we ride, that does not make it necessarily the most accurate method. I would venture to guess you have actually ridden farther because you have not ridden in a perfectly straight line. But, as HillRider has stated it is definitely good enough!

I thoroughly agree with HillRider that this thread is definitely getting out of hand. I also agree with CdCf that he is able to calibrate extremely accurately for a know distance given that you were somehow able to ride in a perfectly straight line. The problem I have with CdCf’s method is the statement that it is the most accurate method. Unfortunately, for a long distance you can’t ride in a perfectly straight line, and while it does take into account how we ride, that does not make it necessarily the most accurate method. I would venture to guess you have actually ridden farther because you have not ridden in a perfectly straight line. But, as HillRider has stated it is definitely good enough!

I'm only interested in the actual road distance I've travelled. That my wheels have rolled a greater distance is unimportant. It's the true distance that counts. If I set out to travel between city A and city B, which are 20 miles of road apart, and I ride a zig-zag pattern all over the road, all the way there, I still haven't actually gone past city B, even if my actual mileage turned out to be 30 miles for the ride...

In practice, though, the difference between the straight line along a road, and the actual path of the bike, is tiny. 1-2% at most.

I'm not disagreeing with you that there is anything wrong with calibrating your distance with your method. From my point of view however I am the opposite...I want to know how far I have actually travelled, not what the true distance between two points are. Isn't an accurate calibration based on how far you have actually travelled? We are splitting hairs over arguing this fact, and I wouldn't have even responded to your initial reply, but to imply that your method is the most accurate is really not a true statement. Unfortunately, how you get from point A to point B is path dependent.

This is my last post to this thread, because it just isn't worth arguing over a trivial error.

This is turning into a giant circle-jerk.

The roll-out and the ride method both have their points. What you do is this: You roll out the wheel, then do a KNOWN 1 mile course (I was lucky enough to get to use a high school's running track for this, and I was even told what lane the measurement of the track was based off of), so I did a full mile, compared to the odometer.

Then 5 mi, compare to the odometer. Then 10, then 15, then 20. Every time, I made very minute changes to the settings to get a proper result.

Then I proceeded to do some club rides, and compare the actual ride time vs MPH (our ride is a precise 20mi ride based on riding a proper line, and taking the turns properly). It was accurate within 15 seconds.

That's the thing cdcf has a point...rollout can be innacurate if a person does not ride straight, and measuring based on city streets can be innacurate based on the way you would ride (swerving glass on the road for example will skew it). It has to be based on multiple rides, and multiple adjustments.

The manual for a computer is only good for a generalized setting. For precision measuring, you need to compensate for non-linear travel skewing the measurements.

You are probably worrying more than you need to.

The two widest spaced numbers there are about 2.5% different.

This means that if you put in the largest number (2099) and you should have used the smallest (2045) you will have about a 2.5% error. If this is the case, the impact of the error is that when the computer says 20.0 you are actually riding at about 19.5 Or if you ride a century of exactly 100 miles on the computer you really only rode 97.5 miles... Nothing to worry about unless you are going for a world record or have other needs that dictate extreme accuracy.

If you are concerned about the greatest possible accuracy, then the best way is the roll out method, but most people don't need that level of accuracy. If it is that critical you should also recalibrate occasionally to accommodate tire wear.

The width of the tire matters because the width is the same (or similar) to the additional radius caused by the height of the tire on the wheel... this changes the effective radius and therefore circumference which is used for speed calculations. For example, with tires for the same 26" rim, a 1.5" tire will have a shorter outer circumference than a 2" tire.

For your 26 X 1.95, I would put in a number slightly less than 26 X 2 since this is a difference in radius of .05 ... probably something like 2091... This should be accurate enough. 2099 - (2 * 3.14 * .05 * 25.4)

OK i used LD's method because i really dont think i need absolute accuracy. HOWEVER, I am curious how inaccurate this method will be over the course of say, 1mile, and also compared to 100miles.
As i previously said, my wheel measuremnt is 26" with 1.95" width.

The closest preset measurment i could find was ATB 26" X 2 (650B) which according to chart gives a wheel factor of 2099. So I entered 2099.
So, correct me if wrong but would that mean that in 1 mile the margin of error would be +/- 0.1 mile?
And then in 100 miles we're talkin +/- 1 mile right?
Thanks for any more thoughts on this.

OK i used LD's method because i really dont think i need absolute accuracy. HOWEVER, I am curious how inaccurate this method will be over the course of say, 1mile, and also compared to 100miles.
As i previously said, my wheel measuremnt is 26" with 1.95" width.

The closest preset measurment i could find was ATB 26" X 2 (650B) which according to chart gives a wheel factor of 2099. So I entered 2099.
So, correct me if wrong but would that mean that in 1 mile the margin of error would be +/- 0.1 mile?
And then in 100 miles we're talkin +/- 1 mile right?
Thanks for any more thoughts on this.

The best way to tell, is find a place that has an exact measurement, for example say a track and field track, where they have it marked for certain distances, like 100m. Pump your tires to your normal pressure, put the bike on the start line, use chalk to mark the tire, and have someone else count the number of times the wheel goes around, as you ride over that distance. Go past the distance until you have a full revolution, measure from the end to your last revolution. Now you have the number of revolutions over a distance. Say it's 50cm past the 100m and you have 32 revolutions.

So 100.50/32 = 3.140625m the wheel factor should have a measurement on it, 2099 sounds like millimeters, so 3140.625 would be your measurement, you would need to decide if 3140 or 3141 is your prefered factor, 3140 would leave you with a reading that is too long, 3141 would be a little short, by .625mm per revolution. Over 100.5km you would have 32000 revolutions, leaving you out by 625mm, or roughly 25" over 62.45 miles. close enough for most people:D .