Garmin Edge: Is this elevation deviation expected?

bicycle-forum.net
Promoting biking discussion.

Main

Date: 27 May 2007 18:14:18
From: Joseph
Subject: Garmin Edge: Is this elevation deviation expected?

I took my Garmin Edge out for my first run this afternoon. A modest loop;
9.1 miles in 43 minutes. Here are the elevation stats:

Start: 88 feet
End: 70 feet

This is odd. I started and stopped at the exact same spot (unless someone
moved my car ;-). So right there there's an 18 foot discrepancy. Not sure
whether this is within the expected deviation range.

Total Ascent: 293 feet
Total Descent: 306 feet

Again, the numbers are slightly off; there's only 13 feet additional
descent, 5 feet short of the expected 18 feet additional descent which would
take me from a start of 88 feet to an end of 70 feet. Again, not sure
whether this is in the expected deviation range. I understand the Edge uses
GPS + Barometric pressure to achieve accurate elevation statistics, but I
wonder what the expected accuracy is. To the foot? Within 10 feet, 20 feet?

If you are using this device, or something similar, I'd appreciate hearing
what degree of accuracy you get from your device, and what I should expect
from mine. If there's a problem with the device, I'm sure Garmin will make
good, it's brand new, but I just want to see if there is in fact a problem,
before I give them a call.

Thanks for your advice.

- Joseph -

Date: 27 May 2007 18:40:20
From: joe
Subject: Re: Garmin Edge: Is this elevation deviation expected?

Joseph wrote:

> I took my Garmin Edge out for my first run this afternoon. A modest loop;
> 9.1 miles in 43 minutes. Here are the elevation stats:
>
> Start: 88 feet
> End: 70 feet
>
> This is odd. I started and stopped at the exact same spot (unless someone
> moved my car ;-). So right there there's an 18 foot discrepancy. Not sure
> whether this is within the expected deviation range.
>
> Total Ascent: 293 feet
> Total Descent: 306 feet
>
> Again, the numbers are slightly off; there's only 13 feet additional
> descent, 5 feet short of the expected 18 feet additional descent which
> would take me from a start of 88 feet to an end of 70 feet. Again, not
> sure whether this is in the expected deviation range. I understand the
> Edge uses GPS + Barometric pressure to achieve accurate elevation
> statistics, but I wonder what the expected accuracy is. To the foot?
> Within 10 feet, 20 feet?
>
> If you are using this device, or something similar, I'd appreciate hearing
> what degree of accuracy you get from your device, and what I should expect
> from mine. If there's a problem with the device, I'm sure Garmin will make
> good, it's brand new, but I just want to see if there is in fact a
> problem, before I give them a call.
>
> Thanks for your advice.
>
> - Joseph -
I have a little over 13,000 miles on my edge and I see this kind of
(in)accuracy in the altitude regularly. Turning the unit on about 20
minutes before you ride will give time for the altimeter to calibrate a
little better, but I would not expect better accuracy than you are seeing.
You will see discrepancies between the difference in starting/ending
elevation and total ascent/descent due to the way it calculates the totals
and the effects of cumulative rounding errors. Changes in weather that
cause changes in barometric pressure will cause errors as will winds that
pass over the pressure sensor. If you look at the edge forum at
www.motionbased.com you will find several discussions of these issues. The
bottom line is that your unit seems to be functioning as expected based on
its design.

Date: 27 May 2007 19:48:14
From: Joseph
Subject: Re: Garmin Edge: Is this elevation deviation expected?

Hi joe:

> bottom line is that your unit seems to be functioning as expected based on
> its design.

OK, thank you!

- Joseph -

"joe" <no@thanks.com > wrote in message
news:135k5nl11f9l579@corp.supernews.com...
> Joseph wrote:
>
>> I took my Garmin Edge out for my first run this afternoon. A modest loop;
>> 9.1 miles in 43 minutes. Here are the elevation stats:
>>
>> Start: 88 feet
>> End: 70 feet
>>
>> This is odd. I started and stopped at the exact same spot (unless someone
>> moved my car ;-). So right there there's an 18 foot discrepancy. Not sure
>> whether this is within the expected deviation range.
>>
>> Total Ascent: 293 feet
>> Total Descent: 306 feet
>>
>> Again, the numbers are slightly off; there's only 13 feet additional
>> descent, 5 feet short of the expected 18 feet additional descent which
>> would take me from a start of 88 feet to an end of 70 feet. Again, not
>> sure whether this is in the expected deviation range. I understand the
>> Edge uses GPS + Barometric pressure to achieve accurate elevation
>> statistics, but I wonder what the expected accuracy is. To the foot?
>> Within 10 feet, 20 feet?
>>
>> If you are using this device, or something similar, I'd appreciate
>> hearing
>> what degree of accuracy you get from your device, and what I should
>> expect
>> from mine. If there's a problem with the device, I'm sure Garmin will
>> make
>> good, it's brand new, but I just want to see if there is in fact a
>> problem, before I give them a call.
>>
>> Thanks for your advice.
>>
>> - Joseph -
> I have a little over 13,000 miles on my edge and I see this kind of
> (in)accuracy in the altitude regularly. Turning the unit on about 20
> minutes before you ride will give time for the altimeter to calibrate a
> little better, but I would not expect better accuracy than you are seeing.
> You will see discrepancies between the difference in starting/ending
> elevation and total ascent/descent due to the way it calculates the totals
> and the effects of cumulative rounding errors. Changes in weather that
> cause changes in barometric pressure will cause errors as will winds that
> pass over the pressure sensor. If you look at the edge forum at
> www.motionbased.com you will find several discussions of these issues.
> The
> bottom line is that your unit seems to be functioning as expected based on
> its design.

Date: 27 May 2007 22:39:42
From: Ron Hardin
Subject: Re: Garmin Edge: Is this elevation deviation expected?

I generally get 20 feet _consistency_ between 2 eTrex's
http://home.att.net/~rhhardin4/odyssy23.jpg

(which is different from accuracy ; these are ones without barometers.)

Over a few hours, the barometer could confound things, since
the barometric pressure changes. You'd come back from an airplane
flight after a few hours and find the altimeter is a hundred feet
off from where it was when you left.
--
Ron Hardin
rhhardin@mindspring.com

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

Date: 27 May 2007 19:24:49
From: Joseph
Subject: Re: Garmin Edge: Is this elevation deviation expected?

Hi Ron,

Thanks for your advice. Perhaps barometric pressure changes account for that
discrepancy. Although my understanding is that the Edge uses GPS to
recalibrate the barometer at fixed intervals in order to minimize
dscrepancies caused by barometric shifts. Anyone out there with an Edge? Is
this discrepancy to be expected?

Here's another thing which is weird. I uploaded my run to MotionBased
(motionbased.com) and they put my elevation gain at +455 / -468. This is way
off from what my device reports; +293 / -306 (although the difference
between the two is constant at 13 feet). I'm 99% sure the device is correct
and MB is wrong. Here's what I think the problem is. My route takes me
*under* the NYS Thruway approach to the Tappan Zee. If MB is using GPS
mapping to place me *on top* of the Thruway at those coordinates, then this
improperly pops me up and then back down after I cross under. On the way
back, my route takes me across a ramp running *over* the NYS. Again, if MB
places me on the road surface, that'll drop me down and then pop me back up
once I cross over. Also, I run along a trail through the woods on a shoulder
of the cliff between (beneath) Route 9 and the Hudson. If MB is improperly
placing me at the top of the cliff, on Route 9, this would account for a
dscrepancy. So I think that MB is just flat out wrong about this. Has anyone
else ever uploaded their stats to MB and seen this sort of discrepancy? Is
there any way to tell MB not to use its own elevation calculation based on
its own GPS data, but rather to go with just exactly what is reported from
the device?

Thanks for your advice.

- Joseph -

"Ron Hardin" <rhhardin@mindspring.com > wrote in message
news:4659FA88.7A48@mindspring.com...
>I generally get 20 feet _consistency_ between 2 eTrex's
> http://home.att.net/~rhhardin4/odyssy23.jpg
>
> (which is different from accuracy ; these are ones without barometers.)
>
> Over a few hours, the barometer could confound things, since
> the barometric pressure changes. You'd come back from an airplane
> flight after a few hours and find the altimeter is a hundred feet
> off from where it was when you left.
> --
> Ron Hardin
> rhhardin@mindspring.com
>
> On the internet, nobody knows you're a jerk.

Date: 27 May 2007 19:38:28
From: Joseph
Subject: Re: Garmin Edge: Is this elevation deviation expected?

> there any way to tell MB not to use its own elevation calculation based on
> its own GPS data, but rather to go with just exactly what is reported from
> the device?

I found this option under Tune Up and disabled it. Now MB shows 'Corrections
Disabled' but still shows elevation gain of +457 / -469 which is way off
from what the device itself reports both on the device itself and in the
Garmin Training Center software. How come MB is so wildly off on this? (I
even deleted the upload and re-uploaded, but this didn't change anything.)

Thanks,

- Joseph -

"Joseph" <joseph@nospam.com > wrote in message
news:cro6i.822$0A1.31@newsfe12.lga...
> Hi Ron,
>
> Thanks for your advice. Perhaps barometric pressure changes account for
> that discrepancy. Although my understanding is that the Edge uses GPS to
> recalibrate the barometer at fixed intervals in order to minimize
> dscrepancies caused by barometric shifts. Anyone out there with an Edge?
> Is this discrepancy to be expected?
>
> Here's another thing which is weird. I uploaded my run to MotionBased
> (motionbased.com) and they put my elevation gain at +455 / -468. This is
> way off from what my device reports; +293 / -306 (although the difference
> between the two is constant at 13 feet). I'm 99% sure the device is
> correct and MB is wrong. Here's what I think the problem is. My route
> takes me *under* the NYS Thruway approach to the Tappan Zee. If MB is
> using GPS mapping to place me *on top* of the Thruway at those
> coordinates, then this improperly pops me up and then back down after I
> cross under. On the way back, my route takes me across a ramp running
> *over* the NYS. Again, if MB places me on the road surface, that'll drop
> me down and then pop me back up once I cross over. Also, I run along a
> trail through the woods on a shoulder of the cliff between (beneath) Route
> 9 and the Hudson. If MB is improperly placing me at the top of the cliff,
> on Route 9, this would account for a dscrepancy. So I think that MB is
> just flat out wrong about this. Has anyone else ever uploaded their stats
> to MB and seen this sort of discrepancy? Is there any way to tell MB not
> to use its own elevation calculation based on its own GPS data, but rather
> to go with just exactly what is reported from the device?
>
> Thanks for your advice.
>
> - Joseph -
>
> "Ron Hardin" <rhhardin@mindspring.com> wrote in message
> news:4659FA88.7A48@mindspring.com...
>>I generally get 20 feet _consistency_ between 2 eTrex's
>> http://home.att.net/~rhhardin4/odyssy23.jpg
>>
>> (which is different from accuracy ; these are ones without barometers.)
>>
>> Over a few hours, the barometer could confound things, since
>> the barometric pressure changes. You'd come back from an airplane
>> flight after a few hours and find the altimeter is a hundred feet
>> off from where it was when you left.
>> --
>> Ron Hardin
>> rhhardin@mindspring.com
>>
>> On the internet, nobody knows you're a jerk.
>
>

Date: 29 May 2007 03:47:04
From: Booker C. Bense
Subject: Re: Garmin Edge: Is this elevation deviation expected?

In article <1Eo6i.3352$Ka2.558@newsfe12.lga >, Joseph <joseph@nospam.com> wrote:
>> there any way to tell MB not to use its own elevation calculation based on
>> its own GPS data, but rather to go with just exactly what is reported from
>> the device?
>
>I found this option under Tune Up and disabled it. Now MB shows 'Corrections
>Disabled' but still shows elevation gain of +457 / -469 which is way off
>from what the device itself reports both on the device itself and in the
>Garmin Training Center software. How come MB is so wildly off on this? (I
>even deleted the upload and re-uploaded, but this didn't change anything.)
>

1. Motion Based is worthless for elevation data. I've had several
email debates with the developers there and they don't seem to
really understand the problem[1]. Getting good stats out of noisy
data requires more sophistication than they seem to be able or
want to do. From what I can tell, they are reading just the
GPS data and not the smoothed by barometric data that's
actually part of the device. Motion based is a small website
bought out buy Garmin and I get the impression that there is a
lot of the right hand not talking to the left in terms of
getting data out of the device.

2. Upgrade the software on your device, the newer versions do a
much better job of tracking elevation in a reasonable way.

In general the Edge is a very nice toy, but it does occasionly
generate non-sensical data[1] and the software is still a work in
progress. In general GPS does not do a particularly good job in
getting elevation data correct, that's why elevation GPS have a
barometer. But nothing is perfect.

_ Booker C. Bense

[1]- Or even really want admit that it exists. It's way more fun
to add fancy new features than make old boring ones work
correctly.

[2]- Max speeds well over 100 mph...

Date: 29 May 2007 00:49:23
From: Joseph
Subject: Re: Garmin Edge: Is this elevation deviation expected?

Hi Booker,

> 2. Upgrade the software on your device, the newer versions do a
> much better job of tracking elevation in a reasonable way.

I've got all the latest software updates applied. I think the Edge does a
pretty good job, except for the slight altitude discrepancy when ending a
ride at the same point as its starting point. BTW, this would be the easiest
thing to solve via software for a discrepancy within a certain limit of
tolerance (parameterized). If the ending point is within 50 ft
(parameterized) of the starting point, and the discrepancy is within the
defined range of tolerance, prompt the user:

Do you want to resolve starting and ending altitudes?

If the user answers Yes, then shift the two altitudes toward each other by
50% of the discrepancy, and re-calibrate all altitudes measured along the
route by the correct pro-rated percentage. The pro-rated adjustment
percentage would be largest toward the beginning and end points of the ride,
and would taper down to zero at the midpoint. This wouldn't appreciably
change the workout, but it would leave the entire route in a state which
makes sense once all is said and done.

It seems so simple. I wonder why they're not doing this.

- Joseph Geretz -

"Booker C. Bense"
<bbense+rec.bicycles.misc.May.28.07@telemark.slac.stanford.edu > wrote in
message news:f3g7no$aae$1@news.Stanford.EDU...
> In article <1Eo6i.3352$Ka2.558@newsfe12.lga>, Joseph <joseph@nospam.com>
> wrote:
>>> there any way to tell MB not to use its own elevation calculation based
>>> on
>>> its own GPS data, but rather to go with just exactly what is reported
>>> from
>>> the device?
>>
>>I found this option under Tune Up and disabled it. Now MB shows
>>'Corrections
>>Disabled' but still shows elevation gain of +457 / -469 which is way off
>>from what the device itself reports both on the device itself and in the
>>Garmin Training Center software. How come MB is so wildly off on this? (I
>>even deleted the upload and re-uploaded, but this didn't change anything.)
>>
>
> 1. Motion Based is worthless for elevation data. I've had several
> email debates with the developers there and they don't seem to
> really understand the problem[1]. Getting good stats out of noisy
> data requires more sophistication than they seem to be able or
> want to do. From what I can tell, they are reading just the
> GPS data and not the smoothed by barometric data that's
> actually part of the device. Motion based is a small website
> bought out buy Garmin and I get the impression that there is a
> lot of the right hand not talking to the left in terms of
> getting data out of the device.
>
> 2. Upgrade the software on your device, the newer versions do a
> much better job of tracking elevation in a reasonable way.
>
> In general the Edge is a very nice toy, but it does occasionly
> generate non-sensical data[1] and the software is still a work in
> progress. In general GPS does not do a particularly good job in
> getting elevation data correct, that's why elevation GPS have a
> barometer. But nothing is perfect.
>
> _ Booker C. Bense
>
>
> [1]- Or even really want admit that it exists. It's way more fun
> to add fancy new features than make old boring ones work
> correctly.
>
> [2]- Max speeds well over 100 mph...

Date: 29 May 2007 11:16:34
From: Ron Hardin
Subject: Re: Garmin Edge: Is this elevation deviation expected?

If they extended their Kalman filter (assuming they use one) there could
be no discrepency, just a single estimate of position consistent at
all times. It's not new technology or anything.
--
Ron Hardin
rhhardin@mindspring.com

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

Date: 29 May 2007 03:55:57
From: Booker C. Bense
Subject: Re: Garmin Edge: Is this elevation deviation expected?

In article <f3g7no$aae$1@news.Stanford.EDU >,
Booker C. Bense <bbense+rec.bicycles.misc.May.28.07@telemark.slac.stanford.edu > wrote:
>
>1. Motion Based is worthless for elevation data. I've had several
> email debates with the developers there and they don't seem to
> really understand the problem[1]. Getting good stats out of noisy
> data requires more sophistication than they seem to be able or
> want to do.

FWIW, I've looked at the raw data their software takes out of the
device and uploads to the web site and it's very noisy. Exactly
what you expect from a GPS alone.

Motion Based has some very interesting features, but elevation tracking
is not one of them. In general I have found them to be at best
within 20% of the true elevation gain and often they list double
the actual elevation gain.

_ Booker C. Bense

Date: 29 May 2007 10:36:10
From: Roger Zoul
Subject: Re: Garmin Edge: Is this elevation deviation expected?

Booker C. Bense wrote:
:: In article <f3g7no$aae$1@news.Stanford.EDU >,
:: Booker C. Bense
:: <bbense+rec.bicycles.misc.May.28.07@telemark.slac.stanford.edu >
:: wrote:
:::
::: 1. Motion Based is worthless for elevation data. I've had several
::: email debates with the developers there and they don't seem to
::: really understand the problem[1]. Getting good stats out of noisy
::: data requires more sophistication than they seem to be able or
::: want to do.
::
:: FWIW, I've looked at the raw data their software takes out of the
:: device and uploads to the web site and it's very noisy. Exactly
:: what you expect from a GPS alone.
::
:: Motion Based has some very interesting features, but elevation
:: tracking
:: is not one of them. In general I have found them to be at best
:: within 20% of the true elevation gain and often they list double
:: the actual elevation gain.

That's kind of depressing to hear since I just bought one of these things to
get this kind of data on my rides. I do see some nonsensical data in my
history file on the first century I did with it.

Date: 29 May 2007 18:54:22
From: Booker C. Bense
Subject: Re: Garmin Edge: Is this elevation deviation expected?

In article <135oejfste6t6af@news.supernews.com >,
Roger Zoul <rogerzoul2@hotmail.com > wrote:
>Booker C. Bense wrote:
>:: In article <f3g7no$aae$1@news.Stanford.EDU>,
>::
>:: Motion Based has some very interesting features, but elevation
>:: tracking
>:: is not one of them. In general I have found them to be at best
>:: within 20% of the true elevation gain and often they list double
>:: the actual elevation gain.
>
>That's kind of depressing to hear since I just bought one of these things to
>get this kind of data on my rides. I do see some nonsensical data in my
>history file on the first century I did with it.
>
>

Use the Garmin Software on your computer, it's doesn't have all
the features of the Motion Based site, but the numbers are at
least believable. There is a version for OS X as well now.
Given the time and delays it took to arrive[1] it's am amazingly
underfeatured piece of software, but it does work.

_ Booker C. Bense

[1]- Over a year late...

Date: 29 May 2007 14:51:56
From: Joseph
Subject: Re: Garmin Edge: Is this elevation deviation expected?

> :: Motion Based has some very interesting features, but elevation
> :: tracking
> :: is not one of them. In general I have found them to be at best
> :: within 20% of the true elevation gain and often they list double
> :: the actual elevation gain.
>
> That's kind of depressing to hear since I just bought one of these things
> to get this kind of data on my rides.

Take heart, he's not indicting the device itself, but rather what
MotionBased does to the statistics which are uploaded. As far as the Edge
itself, I just got this last week and I've used it twice and I am very happy
with it so far. I think it's a great device.

- Joseph Geretz -

"Roger Zoul" <rogerzoul2@hotmail.com > wrote in message
news:135oejfste6t6af@news.supernews.com...
> Booker C. Bense wrote:
> :: In article <f3g7no$aae$1@news.Stanford.EDU>,
> :: Booker C. Bense
> :: <bbense+rec.bicycles.misc.May.28.07@telemark.slac.stanford.edu>
> :: wrote:
> :::
> ::: 1. Motion Based is worthless for elevation data. I've had several
> ::: email debates with the developers there and they don't seem to
> ::: really understand the problem[1]. Getting good stats out of noisy
> ::: data requires more sophistication than they seem to be able or
> ::: want to do.
> ::
> :: FWIW, I've looked at the raw data their software takes out of the
> :: device and uploads to the web site and it's very noisy. Exactly
> :: what you expect from a GPS alone.
> ::
> :: Motion Based has some very interesting features, but elevation
> :: tracking
> :: is not one of them. In general I have found them to be at best
> :: within 20% of the true elevation gain and often they list double
> :: the actual elevation gain.
>
> That's kind of depressing to hear since I just bought one of these things
> to get this kind of data on my rides. I do see some nonsensical data in
> my history file on the first century I did with it.
>
>
>