# General Cycling Discussion - How accurate are bike computers?

Bikeforums.net is a forum about nothing but bikes. Our community can help you find information about hard-to-find and localized information like bicycle tours, specialties like where in your area to have your recumbent bike serviced, or what are the best bicycle tires and seats for the activities you use your bike for.

View Full Version : How accurate are bike computers?

chris_pnoy
08-12-05, 04:43 PM
I was just wondering.

I went out on a ride and I adjusted my magnet so that it rides a tad bit further from the center and I managed faster speeds. I don't know if I was actually going faster or it was because I adjusted the magnet. So I would like to know how accurate are bike computers? Is there a set distance (from the center) so that it can properly use the circumference of the wheel to calculate the linear speed?

operator
08-12-05, 04:49 PM
Uh. It takes the same amount of time to pass the sensor irregardless of how far it's placed from the center.

chris_pnoy
08-12-05, 04:52 PM
Thats what I was thinking, but the best equations I could come up with in Physics didn't really help.

btw, the word is "regardless". I just had to say that because it already is without regards to by making it "regardless", "iregardless" would make it with regards to...

but anyway, thank you.

Ganesha
08-12-05, 05:05 PM
They are as accurate as you calibrate them to be (i.e. the number you put in when you setup the bike computer).

It's pointless the try to get them better then 2-3% percent, since I suspect the uncertainty of the circumference of the front wheel (due to pressure, temperature, how much weight is being transfered by the front/back wheels...etc), is much larger then 2-3% during a single ride.

Wildwood
08-12-05, 06:13 PM
They are as accurate as you calibrate them to be (i.e. the number you put in when you setup the bike computer).

If you want to get anal about this you could correct the circumference after pumping tires prior to a ride. Don't take the reading based on 1 revolution. Get a 50' tape measure and ride 10 or more revolutions before putting the single revolution value into the computer. Totally ridiculous, but your bike computer can be very accurate.

eubi
08-12-05, 06:14 PM
The speed of a rolling object is calculated by:

v = r x w where:

v = velocity of the center of the rolling object (length/time)
w = angular velocity of the wheel (radians/time).. Yeah, I know it's supposed to be lower case "omega", but I'm to lazy to find it.

Given that, you set the radius of the wheel by inputting it into the computer. My computer likes to have the wheel circumference in mm. I like to measure this directly by mounting my bike and rolling it so the tire rotates one revolution. Repeat several times and average. Your computer may be different.

The anglar velocity is calculated from the number of blips the computer sees from the magnet passing the sensor. No matter where the sensor is located radially on the wheel, it will generate the same number of blips, and the angular velocity is calcuated.

The computer now has everything it needs to calculate the speed of your bike.

So, The location of the sensor is not important, as long as it triggers the sensor on every rotation of the wheel.

skookum
08-12-05, 06:37 PM
Hmmm.
It seems that the orientation of the magnet could affect the speed readings.

AndrewP
08-12-05, 09:12 PM
That error is caused by magnet mounted on the spokes with the flat face pointed forwards instead of sideways facing the sensor. The error is not caused by the distance from the hub.

TheRCF
08-12-05, 09:22 PM
As far as accuracy, they are darn accurate in my experience. After very careful measuring (SITTING on the bike as I measure the distanced traveled/revolution), I have set up two different ones which had results indistinquishable over a distance of 12 miles. That is, starting from the same point, I would have it click over to 12 miles at the same place each time (in my case, this was a specific traffic light pole).

I have also found, much to my surprise, that rides day after day have the same result. I had expected some variation in the exact path which would be noticeable. Apparently there isn't much variation since it doesn't show up as fluctuating more than a matter of a few feet each ride.

I find it also doesn't change in any meaningful way based on air pressure - but keep in mind that I don't let it vary much. I set it at 120 lbs and when I pump it back up a few days later, it typically won't need more than 5 lbs or so. Certainly more difference in the pressure would show up more. But when I compared measuring the circumference while sitting on the bike compared to not sitting on it, there was a distinct measurement difference - not huge, but maybe a couple milimeters (I forget how much it was).

After taking a lot of care to get these measurements twice, I no longer take this approach. I know how consistent my ride distances are so I set it where I think it might belong and then, if my actual ride distance doesn't match what it used to be, I change it until it comes out the same as in the past.

skookum
08-12-05, 10:07 PM
That error is caused by magnet mounted on the spokes with the flat face pointed forwards instead of sideways facing the sensor. The error is not caused by the distance from the hub.

Yes of course. The point being that you do have to pay attention to magnet placement. Some of them are constructed so that you could easily rotate them 90 degrees and end up with a problem.

I thought I saw somewhere in Sheldon Brown's pages that you could calibrate your cyclocomputer to an accuracy of +/- 1%, but now I can't find it. Maybe I dreamed it.

JavaMan
08-12-05, 10:37 PM
A switch??? I'm totally surprised any manufacturer would use a switch when all that is required for the sensor is a coil of wire. The moving magnet generates a voltage pulse in the coil. I thought that's how all bike computers worked. A switch with a moving element would be terribly unreliable due to so much shock and vibration. Think about it, don't just believe everything you see on the internet.

khuon
08-12-05, 10:43 PM
A switch??? I'm totally surprised any manufacturer would use a switch when all that is required for the sensor is a coil of wire. The moving magnet generates a voltage pulse in the coil. I thought that's how all bike computers worked. A switch with a moving element would be terribly unreliable due to so much shock and vibration. Think about it, don't just believe everything you see on the internet.

I can't vouch for all but most of the ones I've seen use a reed-relay switch. Actually with some models you can actually hear the clicking as the switch activates.

samp02
08-13-05, 04:24 AM
For exact measurement you can use the roll out method described in every manual I have ever read.

Eggplant Jeff
08-13-05, 07:18 AM
If your previous placement of the magnet was right on the edge of where the sensor was able to pick it up, it is possible that it was only triggering the sensor 90% of the some or so... resulting it it displaying 90% of your actual speed. When you adjusted it, you may have gotten it in a better location (relative to the sensor) so it now triggers the sensor 100% of the time, thus displaying a higher speed (which is your actual speed all along, it was just reading slow before).

Dunno about that computer but mine requires the magnet to pass no more than 2mm from the sensor, and when I was fiddling with it I noticed it would read low speeds if it wasn't quite close enough.

 for the switch debate: Mine does not make any audible noise, however it is a wireful (not wireless) Cateye Velo 2 cheapo \$14 at Performance. Would it make sense that wireless ones might use a switch, since they need to trigger a transmitter, whereas wireful ones can use the inductive coil, since the wires can simply carry the signal up to the computer?

psuaero
08-13-05, 02:03 PM
In my experience with several different models they can be very accurate if you just take the time to set it up instead of using the book numbers.

I just calibrated my new computer (Astrale 8) after having using the "stock" setting of 2096mm for a 700x23 road tire for about a week. I rolled out my tire for 6 complete revolutions on the sidewalk with myself on the bike. It turns out the reading should have been 2104mm (according to the book that is a 700x25 tire). A difference of 8mm is only 20ft/mile but it adds up slowly. Now I should be right on!

chris_pnoy
08-13-05, 08:19 PM
This is interesting. Maybe I too should recheck the calibration of my computer. Thank you all guys.

phidauex
08-14-05, 12:29 PM
Many motorcyclists use the Sigma computers for their motorcycles, because they are dead on accurate up to 199 kph! People have done very careful weighted rollout tests to calibrate them, then gone on big long rides only to have the Sigma odometer perfectly match the GPS odometer.

Makes you wonder why car and motorcycle speedos aren't that accurate... Oh yeah, its because they are purposely optimistic, to make you slow down. ;)

I think that some computers have a reed switch, and others have a hall effect sensor (which is preferable to a coil of wire for magnetic field sensing). If I were designing a bike computer, I'd definately use the no-moving-parts hall effect sensor instead of a faliable reed switch, but the switch is probably a little cheaper.

peace,
sam

jur
08-15-05, 07:23 AM
Does tyre pressure affect the accuracy? The way I see it, not. The contact surface of the tyre does not become longer when pumped hard, I think. Infinitesimally perhaps.

Mercian
08-15-05, 08:08 AM
To calibrate my speedo, use non-stretch masking tape (at least I don't pull it tightly), rolled around the outside circumference of a fully inflated tire, make a mark and then measure.

It's helpful to do this a couple of times, but rarely has the difference been more than a milimeter, and this is usally due to stretch of the maksing tape, stretch of the tape measure, or whatever other errors creep in, but it is usually more accurate than a roll out on the garage floor. After all that precision, find that sometimes a particular brand of speedo won't allow the exact measurement to be entered==you know, the result has to be rounded to the nearest "5mm".

TheRCF
08-15-05, 01:57 PM
Does tyre pressure affect the accuracy? The way I see it, not. The contact surface of the tyre does not become longer when pumped hard, I think. Infinitesimally perhaps.

Don't know how much difference that makes, but I do know that measuring the distance of a single revolution made a difference if I was sitting on the bike or not. Wish I could remember how much, but I can't - but since my computer lets me enter to the nearest mm, I do know it had to be at least one digit difference. True, that's not much. Even if off by 2, that's only about 1/1000 difference.

It seems to me that if your pressure is lower, that your weight would press down on it more, making a bigger difference, but I haven't actually checked that.

Big Tommy C
08-15-05, 02:06 PM
On mine, I like to just wrap a tape measure around the front wheel and get the circumference that way.

As far as sensors, my CatEye Mity 8 definitely has a relay that clicks, as I can hear it when I spin the wheel (though not when I ride, it is pretty quiet). My Planet Bike Protege on the mountain bike I have never heard, it could be either a different type of sensor or it might just be quieter.