i bought a bell f20 bike computer and so far it has seemed pretty accurate on the first bike i installed it on. today i decide to fix a flat on my other mountain bike and swap the computer onto it and it no longer seems so accurate. the first bike i installed it on has a 24x1.75 back tire and a 26x2.00 front tire the second bike has a 26x2.00 back tire and a 26x2.30 front tire. i figured that since the second bike had a bit bigger of a rear tire i should have been able to beet the top speed of my first bike but it was 4km/h slower. i went down the exact same hill. all day it has seemed a bit slow. any idea's because it just don't seem right to me.
my top speed with the first bike is 54.7km/h and the top speed with the second bike is 50.4km/h

At the same speed you have fewer rpm's with the larger tire so the computer thinks you are going slower.

You need to do a roll-out measurement, i.e. the actual circumference of the tire, and input that number to get an accurate readout.

As with many things about bicycling, the late Sheldon Brown's web site covers this topic in full. Browse over to his Cyclocomputer Calibration page. Read on down beyond the chart to the Roll-out Test for Super Accuracy and better yet Using a measured course to fine-tune your setting. You can measure a course you ride using Google Maps or other online map sites.

Calibrated properly, the computers can be both quite accurate (=true measurement) and precise (=repeatedly measures the same course the same distance). I can tell when my tire pressure has dropped significantly when the distance measured increases by 0.02 miles over just a 3 mile course.

Also see Sheldon's Articles about Cyclecomputers page for more useful information.

If you have a GPS, you can use it to calibrate as well. Any GPS will do, as long as it can run on batteries and you can somehow transport it on your bike. Borrow a friend's Garmin 705, use the one out of your car, a handheld one, whatever. Doesn't need to be mounted nicely, in a jersey pocket would be fine.

Here's the procedure --

Set up your cyclocomputer. In particularly, you have to know exactly what number you entered into it (the units don't matter.)

Pick a route that is relatively straight (corners may reduce the GPS accuracy slightly) and has no tunnels (thick tree cover is OK if the GPS doesn't mind it), zero the GPS and your cyclocomputer, ride. The longer the better, but a few miles is plenty.

At the end, make exact note of your cyclocomputer mileage reading and your GPS mileage reading.

Once done, take your old cyclocomputer, and plug it into this formula

(new number) = (old number) * (GPS distance) / (cyclocomputer distance)

then reconfigure your cyclocomputer with the new number. Then run the test again (doesn't need to be the same path or distance) -- the distances between the two should be very close, within one percent. If you want you can do the formula again (remember, the (old number) changed -- use the most recent version) and get a tad closer and then test again.

Note that if you configure the cyclocomputer properly using the table they provide and your tire size, it should be within 5% or so (without any corrections on your part.) If you find it to be off more than that, either you got the tire size or configuration wrong or something else is wrong. In particular, if your cyclocomputer reads way too low, the odds are good that the sensor isn't reading every time the tire rotates and you may want to check that. If it reads too high and is wireless (especially if it goes way too high sometimes) it may be picking up interference from something.

i have not not got to measure the circumference of the tires yet but i think it is 1km/h slow for very 4km/h so instead of my top speed being 50.4 it would be more like 60.4km/h. i noticed when i was walking with my bike today that i was only going 4km/h and the first bike use to go about 5km/h when i was walking with it.

This kind of thumbnail math can reduce the error from 25% to ~10% or so, which still isn't very accurate. Your bike computer is capable of accuracy down to about 0.5% or better.

The best way that I've found to calibrate is to ride a short path of known distance, and then do the math. GPS works, or Google Maps, or gmap-pedometer.com, or bikely, or whatever. Pick a route a mile or so long. Look close at the plotted route to make sure you're biking exactly that. Ride it and note the distance on your bike computer. Then:

(new calibration number) = (current calibration number) * (gmap or gps distance) / (recorded distance)

a piece of chalk (or a crayon, or a sharpie, or whatever) and a tape measure are so much easier. Find a piece of pavement about 30 feet long. mark the pavement, put your valve stem straight down on the mark, and walk the bike in straight line for three (or more) complete revolutions. Mark the pavement there. Measure. divide by the number of revolutions. Convert that rollout into the units your computer expects. Takes ten minutes, if you ahve to scrounge to find the tape measure.

Use a GPS to calibrate a $30 computer... Thats huh....expensive.

Just follows everyone elses advice and determine you wheel rollout then set your computers wheel size.

This page has a link to the manual if your missing it.
https://sitivi.blogspot.com/2009/04/b...-computer.html

Well, it's cheaper than having a course built with an accurately measured length!

But considering that lots of people already have GPSs for hiking or portable units in their cars, in their phones, or you could borrow the Garmin 305 from the cyclist next to you for a mile, and it'll give you more accurate results than pretty much anything else (though most of the methods given are already plenty accurate), it was worth mentioning, in spite of your "thats huh....expensive" claim.

The GPS method is not only more expensive it might be less accurate that measuring the tire/wheel diameter (under load). (Keep in mind that the tire pressure and the weight of the load contributes to the effective tire diameter.)

There is nothing that indicates that measuring the diameter is any less useful than other methods (ie, the GPS method is overkill).

Keep in mind that the bicycle isn't travelling in a perfect straight line.

Again, it's only more expensive if you have to actually buy a GPS -- and I started out "if you have a GPS" and continued with "borrow a friend's Garmin" ...

It's more accurate than measuring the circumference of your tire because it's real world. You're riding your bike. It's under load. Your tires are (hopefully) properly inflated (but do be aware that changing pressures will change your cyclocomputer readings, and calibrating now won't fix future changes.) And as an added bonus, if your cyclocomputer loses 1 out of 200 revolutions (or whatever), it takes that into account (though hopefully this isn't an issue, because unless it's constant (which is unlikely) it's going to mess with your accuracy badly.)

It's more accurate than using a section of road with a known distance because it's better able to correct for a non-perfectly straight path. (Still, the straighter the better, but most GPSs do update every second so as long as your turns are gradual it should be fine.) It's also rare that you actually do have a section of road with a distance known to 0.1% (five feet per mile) but a GPS can easily do that.

(I would suggest avoiding hills for the test, however -- your cyclocomputer will measure the straight line distance taking altitude changes into account, but GPSs are less accurate with regard to altitude changes and some might (maybe) ignore changes in altitude entirely when calculating your distance traveled, though I hope none actually do this.)

I'm sorry you're so strongly against this method of calibrating your cyclocomputer, but it's a perfectly valid and extremely accurate method and very easy to do. Yes, if you don't own a GPS and can't borrow one, then perhaps this method isn't for you, but I think most people can figure out if they fall into that category on their own.

That said, all of the ideas given in this thread are good enough and can get you within a few percent even without even being that careful -- and that's probably just fine for 99% of the people out there. And if you're careful, you can get within 1% with most of the ideas here -- and that's close enough that small changes in tires, tire pressure or load will probably introduce larger errors than that.

You're not going to get anywhere near 0.1% accuracy with a non-corrected GPS; worse, most GPS receivers don't even tell you things they could that would let you guess at their instantaneous accuracy. You're suggesting an incredibly complicated method which will have no benefit than a less complicated one. It's also hugely more work. So there are lots of reasons to object to it.

Most now have WAAS, which does provide corrections (though it's not needed so much for purposes of this, read on.) If you ride a few miles with the GPS on in mostly a straight line, the biggest limitation on it's accuracy of how far you took is how many digits in it's display, and the reason I'll mention shortly.
Actually, many do -- they give you an estimate of the error in your position. But more importantly, the error is likely to not jump around so much in a short period -- so if it's off by 15 feet now, it's likely to be off by approximately 15 feet (in the same direction) a mile down the road five minutes from now, so the estimate of how far you went will be off by just a few feet at most.
It's no more complicated than trying to measure how far your bike goes in three revolutions of the wheel -- and more accurate. Three revolutions worth of tire rotations works out to about 248 inches for 700c tires. Most tape measures aren't that long, so you'll have to measure out a few times and make marks. I'd estimate the error when all done at around 3" or so? If so, that's a bit over 1% error -- which is certainly acceptable, but a GPS could easily do 2 or 3 times better over a mile or two.

More expensive, if you have to buy a GPS, I can agree with. (Not that tape measures are free, of course.) Beyond that, "hugely more work" ? "incredibly complicated method" ? No way. It does have a trivial formula to use, but you had "convert that rollout into the units your computer expects" so it's not like a calculator wouldn't be used in either case. (I guess we could factor in the price of a calculator too.)

If "new number = old number * gps distance / cyclocomputer distance" is too hard, perhaps you should just take your bike to the bike shop, they'll get it pretty close for you (probably by either looking the number up or measuring how long it takes to do a revolution, as that would beat riding the bike for them.)

(Hell, Sheldon Brown gave the exact same formula, in the guise of "Using a measured course to fine-tune your setting" -- my method is exactly the same, except that you used the GPS to turn any course into a measured course.)

You haven't presented any support that it is any better than other methods!

Why "three revolutions"?

You argument is based on guesses!

And the path that the bicycle takes is likely going to be longer than the accurate path the GPS measures. Bicycles wobble all the time.

And here you present the reasons why the GPS method is overkill!

Nope, I roll my wheel out once, I have a tape measure that has MM on the tape.

The procedure is really simple. I use the valve stem for the tire marker. Stretch out tape, sit on bike, roll one revolution, read tape, enter read measurement into computer. Done. It is that simple.

Measuring multiple revolutions reduces the inaccuracy from getting the start and end points on the wheel wrong in measuring the rollout.

Let's say you've got a wheel that's exactly 2200 mm in circumference. You measure, and you get the start and end points off by 1% each, in opposite directions, so you measure it as 2244 (or 2156). Your computer is going to be off by about two percent. If you measure three revolutions, and you're off by 1% of a wheel circumference, you'll measure 6644 mm, and your computer is going to be off by about 0.6%.

This is a pretty standard error reduction technique, used when the human error is roughly constant, and measuring a big quantity (or time) causes it to be a smaller fraction of the total.

Very good point.
huh, if you could use like a GPS to measure over a few miles, you'd be VERY accurate

i still have not had a chance to do a roll-out measurement yet because it has been so crappy out and i have been feeling a bit sick but i think i have it fixed. i moved the magnet and pickup closer to the hub and it seems to have corrected it. now when i walk my bike it says 5km/h. i don't have a GPS unit so that is not an option. my guess is based on the fact that the average person walks at 5km/h

This technique assumes that there is no appreciable error (variation in distance traveled) between the endpoints.

It's not that easy keeping a bicycle perfectly straight for one wheel rotation. It would be harder to do that for three.

I suspect that the largest contribution to variation in the measurement is keeping the bicycle straight, not running the tape measure.

You could measure 3 separate wheel revolutions and take the average.

This indicates that it was missing counts of some wheel rotations.

it was not missing count on a rotation. it caught every rotation but i think the magnet and pickup need to be a certain distance from the center of of the hub and the installation were not very clear on where it is to be placed. plus my front hub is a bit bigger than most bikes which could affect where the magnet sits.

The distance from the hub center DOES NOT make a difference to the computer. The wheel circumference does. Do the roll out measurement.

When I was looking to see how accurate my computer was, I rode a measured 25mi bike trail out/back and used the 50.0mi to 'calibrate' my computer. After a couple trips doing this, my computer ticked over to 50.0 within 20 feet of the '0' mark. I figured that was accurate enough.

i did the roll out measurement and got 81 & 1/4 inch. i did th roll out test on the first bike i installed it on and got 80 & 3/4 inch. i think the distance that the magnet is from the hub does make a difference even though it should not.