Accurate calibration of cycle computers/cyclometers

"Mark Leuck" wrote

I use White-out on the concrete garage floor

Yes. The weight of the rider on the bike changes the
wheel circumference by a small amount. Some people
resort to putting a spot of paint or other material on the
ground, riding through it, and then measuring the distance
between the paint imprints as the wheel rotates.

Where a surveyed accurate known distance course is
available, such as perhaps a running track or a highway
speedometer test zone, you can simply ride the course, note
the mileage, and apply a correction to the cyclometer setting
in ratio to the actual vs. reported distance. Repeat as
desired.

I generally don't bother with more than a simple single
rotation measurement when I get a new cyclometer
or a new tire. I recently rode through a five-mile
marked speedometer test zone and found even my
relatively casual setting was within 2%. I haven't
bothered to adjust the setting...

When this subject has come up before, some people
mention use of GPS distance measurement or automobile
odometer readings for comparison. Both are subject to
errors, of course. GPS constructs a route by "connecting
the dots" of its sampling period and doesn't account for
the curves it cuts out. Some are better than others.
Automobile odometers are often off by several percent
and are subject to tire size changes and inflation differences.

For most of us, close is good enough. Even a single
wheel rotation measurement is more accurate than using
the wheel/tire size tables that come with most cyclometers.

Jon Meinecke

I've used the measured course method as told by Sheldon Brown

http://www.sheldonbrown.com/cyclecom...l#measuredmile.

I use the same course, a twisty section of road about 4 miles long I
regularly ride. I've plugged the formula into the excel spreadsheet
and calibrate my bikes. I use the mile markers on the side of the
road, and during testing, I keep the speed down so I can stay about
the same distance from the edge of the road, no cutting corners.

Very curious thing, the small wheels provide a very repeatable result.
I run the route and collect the data, enter it into the spreadsheet
and adjust the computer and retest to find more accuracy. 16 and 20"
wheels work very well.

26 and 700 wheels get a bigger error than started with. I've no idea
why.

My GPS, a Garmin Vista mounted on the handle bars, has almost no
error, .01over the 4 miles, so long as we don't drop satellite
contact.

Jon Meinecke wrote:

When this subject has come up before, some people
mention use of GPS distance measurement or automobile
odometer readings for comparison. Both are subject to
errors, of course. GPS constructs a route by "connecting
the dots" of its sampling period and doesn't account for
the curves it cuts out. Some are better than others.


True, I've seen as much as 10% error between two GPS receivers
measuring the same twisty route (when subject to variable tree cover
etc.).

However, they can be used to get a very much more accurate measurement
if you can use them to create mile markers over a straight line.

Choose a time when there are many satellites visible in a location
without buildings or tree cover. Halt at one end of the route, allow
some time to average the position and waypoint that location. Note
the display of estimated position error (this isn't totally reliable
but you can at least try to ensure it's only a few feet). Then move
to the other end of the straight stretch, again stand still to get
the best out of the position averaging and use the function which
measures distance to the first waypoint. This value should be
accurate to within 2x the claimed accuracy of the receiver, which
is far better than the odometer function. You might want to repeat,
tweaking your location for an exact mile etc.

It's also possible to use projected waypoints to get an exact mile but
the principle is similar - you're using the comparison of two averaged
locations rather than the sum of a sequence of estimated movements.

It's always best to use a straight line - mile markers on a twisty
route are not necessarily reliable unless you follow the same line
with the bike as the original marker took.

-adrian

wrote

Very curious thing, the small wheels provide a very
repeatable result.
[...]

26 and 700 wheels get a bigger error than started with.
I've no idea why.

We can speculate...

Cyclometers work by counting pulses or switch closures
from the wheel sensor. It is possible that they may miss a
pulse or see one pulse as two. With larger wheels, the
resulting error in computed distance is larger than with
smaller wheels. However, smaller wheels would result
in a higher pulse rate and perhaps more opportunity to
skip or miscount! %^P Hmmm....

One could design a cyclometer (and perhaps they do)
that used frequency sampling and averaging over a time
increment to reduce the effect of skipped/miscounted
pulses....

In any case, it seems unlikely that such miscounts could
account for a significant error, though. What size is the
deviation/variation you see over what distance?

In my five-mile surveyed markers route, I saw the same 2%
error for each of the miles. This was with a 650c wheel
mounted sensor. I also have bikes where the sensor is
on 16" and 20" wheels and haven't noticed significantly
more or less accurate distance readings for the route I
commonly ride.

Jon Meinecke

On Thu, 28 Jul 2005 12:02:24 -0500, "Jon Meinecke"
wrote:


In any case, it seems unlikely that such miscounts could
account for a significant error, though. What size is the
deviation/variation you see over what distance?

Working from memory here because my notes are about 100 miles away..

When set to the printed suggestions tat came with the computer I was
about half mile off over the measure course.

After running the numbers through the spreadsheet, I was over a mile
off. Suggested number was something like 82.2 inches and corrected
number was 100.

I guessed at a new number, based on the fact the 100 seemed to go the
wrong direction with the error, and set the new number at 75 and I've
not had a chance to recheck, since this current calibration is a work
in progress, not to mention, I changed from a large tire on the wheel
to a narrow tire yesterday while doing some maintenance, so I will
have to start all over again.

Mike

wrote
On Thu, 28 Jul 2005 12:02:24 -0500, "Jon Meinecke"
wrote:

What size is the deviation/variation you see over what distance?

I may have misinterpreted your comments on differences in
repeatability differences between cyclometer sensors on small
wheels and big wheels. I thought you were saying you saw
more variance in the cyclometer measured distance over
the same course when riding with larger diameter wheels.

When set to the printed suggestions tat came with the computer
I was about half mile off over the measure course.

The correct setting is equal to the old setting times the actual
distance divided by the distance as reported with the old setting.
Since it is a simple ratio, it doesn't matter what the setting number
unit is (circumference in cm, mm, in, diameter in whatever...)

A = actual distance in miles
D = distance as reported with S setting
S =setting (that generated D)
C = correct setting

A / D = C / S

C = S * A / D

For instance, if the actual distance is 10 miles, the reported
distance is 10.5 and the setting was 82.2 then the correct setting
is 78.3.

In my case, my cyclometer shows I've ridden 1.02 miles for an
actual 1 mile course. I don't recall the exact number, but say my
current setting is 1995, the corrected setting would be 1956...

I changed from a large tire on the wheel
to a narrow tire yesterday while doing some maintenance, so I will
have to start all over again.

The known measured course need only be long enough to show
the error within the precision of the cyclometer. Shorter may be
better, as then there may be less variance, wandering on the
course. Ride it once and apply the formula.

I've never found the suggested setting numbers in cyclometer
instructions to be very accurate. In fact, I've always found them
to be too large for a given wheel/tire size. That makes sense,
if they're going to err, err on the side that makes people think
they're riding faster and further! %^)

I've also noticed the price scanning errors are most often in
favor of the store.... %^P

Jon Meinecke

On Wed, 27 Jul 2005 15:33:38 -0500, "bent_pilot"
wrote:


"Mike" wrote in message
hlink.net...
Dex wrote:
Accurate calibration of cycle computer

I measure the distance convered by several revolutions of the wheel
that I am calibrating. Make a scratch mark on the surface with the
valve stem at the bottom of the wheel. Make several revolutions of
the wheel and make another scratch mark with the stem at the bottom
again. I usually use 10 wheel revolutions. Measure this distance and
record it. If the distance was measured in feet, you usually have to
convert it to centimeters to use it in setting your wheel
circumferance. First, find your wheel circumferance by dividing the
distance by the number of wheel revolutions. Then multiply this
answer 30.48 -- this will be the number you use to set the wheel

Where does this number come from? The conversion I use is:
millimeters = Inches * 2.54

This is the number (30.48) that I think I have been using to convert
feet to cm.

-Dex