Altimeter Errors, Calibration Problems



Status
Not open for further replies.
D

Dan

Guest
I have been using a CycloMaster (CycloSport) 436M for a few months. It is similar to the HAC4 but
lacks the heart monitor capability. I did the Death Ride this year (www.deathride.com) and collected
a full data set for the ride (http://www.geocities.com/danmerrick/deathride2003/). I only measured a
total climb of about 14,500 feet as opposed to the advertised 16,000 feet. a large error that seemed
unlikely. I have worked as an engineer in data acquisition and analysis for many years so I decided
to check for errors. By comparing my data elevations and topo elevations (www.topozone.com) at 13
points along the ride, I found that the error varied with elevation. I set up a spreadsheet and did
a linear regression analysis and came up with a correction (or calibration) factor. My first
analysis of the data after applying the factor came up with a climb total of 15,900 feet - good
enough for me.

So, I guess my point is - don't trust altimeter data until you have done a calibration check.
 
Dan wrote:

> I have been using a CycloMaster (CycloSport) 436M for a few months. It is similar to the HAC4 but
> lacks the heart monitor capability. I did the Death Ride this year (www.deathride.com) and
> collected a full data set for the ride (http://www.geocities.com/danmerrick/deathride2003/). I
> only measured a total climb of about 14,500 feet as opposed to the advertised 16,000 feet. a large
> error that seemed unlikely. I have worked as an engineer in data acquisition and analysis for many
> years so I decided to check for errors. By comparing my data elevations and topo elevations
> (www.topozone.com) at 13 points along the ride, I found that the error varied with elevation. I
> set up a spreadsheet and did a linear regression analysis and came up with a correction (or
> calibration) factor. My first analysis of the data after applying the factor came up with a climb
> total of 15,900 feet - good enough for me.
>
> So, I guess my point is - don't trust altimeter data until you have done a calibration check.

Or, get one that's built into a GPS and therefore self calibrating :).

David
 
I've experienced the same thing with an Avocet Vertech watch. Its altimeter is extremely accurate --
it matches an aircraft altimeter in level flight (I'm a pilot) --but has more lag in climb and
descent. This, combined with the fact that its accumulator looks only at 10-ft increments, may
explain why the Avocet logs fewer vertical feet than you actually climbed.

A friend uses a topo database to plan road rides. Assuming his data to be accurate (why not?),
my Avocet watch shows about a 40% deficit on flattish-to-rolling routes vs. 20% on seriously
hilly routes.

Mike Yankee

(Address is munged to thwart spammers. To reply, delete everything after "com".)
 
Mike Yankee writes:

> I've experienced the same thing with an Avocet Vertech watch. Its altimeter is extremely accurate
> -- it matches an aircraft altimeter in level flight (I'm a pilot) --but has more lag in climb and
> descent. This, combined with the fact that its accumulator looks only at 10-ft increments, may
> explain why the Avocet logs fewer vertical feet than you actually climbed.

Not so. It logs less altitude gain because it rejects dither from atmospheric change and undulating
terrain less than 10 meters. It rejects things like railway underpasses although it take account of
the bottom of the dip and records any gain beginning at 10 meters above this previous low. This
means that on a flat road that undulates as much as +-5 meters, no accumulation... a feature that
riders in the mountains (and that's what it's for) prefer.

This method is used in both the Vertech and AVO50 bicycle altimeter (no longer available).

> A friend uses a topo database to plan road rides. Assuming his data to be accurate (why not?),
> my Avocet watch shows about a 40% deficit on flattish-to-rolling routes vs. 20% on seriously
> hilly routes.

Jobst Brandt [email protected] Palo Alto CA
 
[email protected] wrote:
> Not so. It logs less altitude gain because it rejects dither from atmospheric change and
> undulating terrain less than 10 meters. It rejects things like railway underpasses although it
> take account of the bottom of the dip and records any gain beginning at 10 meters above this
> previous low. This means that on a flat road that undulates as much as +-5 meters, no
> accumulation... a feature that riders in the mountains (and that's what it's for) prefer.

The actual vertical climb is infinite if you get down to infinitely small detail.
--
Ron Hardin [email protected]

On the internet, nobody knows you're a jerk.
 
MikeYankee wrote:

> I've experienced the same thing with an Avocet Vertech watch. Its altimeter is extremely accurate
> -- it matches an aircraft altimeter in level flight (I'm a pilot) --but has more lag in climb and
> descent.

This doesn't mean that it gives accurate altitudes, just that it accurately measures atomspheric
pressures. Barometric altimeters, including those in aircraft, can easily be off by hundreds of feet
if the atomspheric conditions don't match the 'standard atmosphere model' and the unit hasn't been
recalibrated periodically to compensate for atmospheric changes.

> This, combined with the fact that its accumulator looks only at 10-ft increments, may explain why
> the Avocet logs fewer vertical feet than you actually climbed.

It will log less if you're on a road that tends to go up and down repeatedly but less than 30' each
time. This usually isn't an issue on rides in hilly areas.

>
> A friend uses a topo database to plan road rides. Assuming his data to be accurate (why not?),
> my Avocet watch shows about a 40% deficit on flattish-to-rolling routes vs. 20% on seriously
> hilly routes.

Not clear what method he uses with the topo database, but in my experience using a computerized topo
map and drawing the route on it tends to give substantial overestimates of climbs. Looking closer at
the topo data showed that it was including lots of little ups and downs along sections of road where
I knew we were constantly ascending (or constantly descending). The topo data may have been valid if
we had hiked that section before the road was built, but the road builders flatten out lots of the
little irregularities in the land and reduce the total amount of climbing needed.
 
Peter wrote:
> This doesn't mean that it gives accurate altitudes, just that it accurately measures atomspheric
> pressures. Barometric altimeters, including those in aircraft, can easily be off by hundreds of
> feet if the atomspheric conditions don't match the 'standard atmosphere model' and the unit hasn't
> been recalibrated periodically to compensate for atmospheric changes.

Opening the window changes the indicated altitude by a couple hundred feet, in the primitive planes
I used to fly. Air motion does the same thing on a bicycle except there's no window to test it with.

Stand on the windward and opposite side of a mountain peak and see.

Anything that changes the direction of the air motion indicates a pressure difference due to
air motion.

A pure GPS altitude is unaffected by wind, whatever else may be true about it. It varies over a
hundred feet over a day, but not quickly. So probably for cumulative computations it's better.
--
Ron Hardin [email protected]

On the internet, nobody knows you're a jerk.
 
> The actual vertical climb is infinite if you get down to infinitely small
detail.

Yep, I agree. When I first looked at my data I decided that the low value for total climb was
because the unit measures steps in altitude and it had simply missed some of the small steps.
However, the ride is a series of 5 big climbs and descents. No flat segments and no rolling terrain
so this should not have been a problem. I had forgotten to set the altitude at the start so I
expected an offset error of a couple hundred feet due to changes in barometric pressure. The offset
I calculated to correct was different for the high and low points which led me to the conclusion
that there was a calibration error or the atmospheric pressure model was wrong. Atmospheric pressure
does not vary exactly linearly with elevation and temperature and this secondary effect may not be
accounted for by the designers. My ride varied from about 5100' to 8700' and I found that the
measured elevation change was about 93% of the actual.
 
uh..duh...which one should I buy? i don unnerstan alla tiknikal tawk
 
I never trust any reading on any computer untill I have calibrated it. That is, tested it's read out
against a measurement of known accuracy.

May you have the wind at your back. And a really low gear for the hills! Chris

Chris'Z Corner "The Website for the Common Bicyclist": http://www.geocities.com/czcorner
 
Status
Not open for further replies.

Similar threads