GPS vs Bike Speedo?

(PeteCresswell) said:

My kneejerk is that the micro code in the GPS is smarter than just averaging out
the instant speeds... but in the back of my mind Occam is whispering....

Anybody have an explanation?

I don't, however I have the same device. Over a 90km ride, the variance
between the GPS recorded distance and my old cateye bike computer was
less than 300m. I was impressed!

--
..dt

(PeteCresswell) said:

- On instantaneous speed, the GPS almost always read between .5 and 1.5 mph
faster than the speedo. Sometimes it was less... but the vast majority of the
time it was more. One stretch that must've lasted 10 minutes I held the
speedo at exactly 12.8 mph. During that entire time the GPS read at least .5
mph higher, and sometimes more like 1 mph higher.

- But the average speed and distance covered were almost exactly the same on the
GPS and the speedo. Within a half mile total distance out of 24 miles and
within about .3 mph average speed.

[...]

Quoted message said:

Anybody have an explanation?

I believe the averages are taken using samples. The frequency of the samples
is - usually - adjustable. On my Garmin, I can use a "by time" method, a "by
distance" method and a proprietary method by Garmin, probably a mixture of
the two previous methods. In all cases it is possible to use different
frequencies. The default was a sort of average value, if I remember well
for the Garmin optimized method.

Now, let's assume the bike goes along a straight line. The sampling frequency
is irrelevant: two samples or N samples must give the same output.

Let's take the opposite example: a circular path. Here sampling frequency is
very important. If the circle has a circumference of 1 mile and the sampling
frequency is 1 mile, than the overall distance is zero and so is the average
speed. Increasing the sampling frequency will bring the calculated average
closer and closer to the real average, without reaching an exact match.

In conclusion, if the algorithm used by the GPS to compute the average speed
is based on the data points, then I'd expect this to be slightly below the
actual average speed. The error should be proportional to the sampling rate.

Try to increase the sampling rate and see if that helps.

Ciao,
--
Roberto Divia` Love at first sight is one of the greatest
Dep:PH Bat:53 Mailbox:C02110 labour-saving devices the world has ever seen
Route de Meyrin 385 ---------------------------------------------
Case Postale Phone: +41-22-767-4994
CH-1211 Geneve 23 CERN Fax: +41-22-767-9585
Switzerland E-Mail: [email hidden]

I read that subject as "GRS vs. Bike Speedo?"

GRS = Gay Roadie Sh^t in AM-B, so in that context "Bike Speedo" conjured
up...

Oh, never mind.

"(PeteCresswell)" <[email hidden]> wrote in message
news:[email hidden]...

Quoted message said:

Been playing around with a Garmin ForeTRex 201 GPS receiver.

On the ride to/from work today, I put the GPS next to my CatEye Velo.

Funny thing:
-------------------------------------------------------------------------
- On instantaneous speed, the GPS almost always read between .5 and 1.5
mph
faster than the speedo. Sometimes it was less... but the vast majority of
the
time it was more. One stretch that must've lasted 10 minutes I held the
speedo at exactly 12.8 mph. During that entire time the GPS read at
least .5
mph higher, and sometimes more like 1 mph higher.

- But the average speed and distance covered were almost exactly the same
on the
GPS and the speedo. Within a half mile total distance out of 24 miles
and
within about .3 mph average speed.
-------------------------------------------------------------------------

If the averages/totals weren't so close I'd just chalk it up to improper
setting
of the wheel diameter on the speedo.

If it weren't for that 10-minute stretch, I'd attribute it to latency.

If the total distances weren't so close, I'd suspect that somehow I was
selectively remembering the times the GPS read higher and ignoring the
ones when
it read lower.

My kneejerk is that the micro code in the GPS is smarter than just
averaging out
the instant speeds... but in the back of my mind Occam is whispering....

Anybody have an explanation?
--
PeteCresswell

Hi

I'd suggest that the instantanious fixes for the GPS 'jump around' a bit.
These are averaged out for the odometer reading (the unit probably uses a
longer filter). My understanding is that GPS receivers calculate speed by
measuring distance between fixes. It's likely these aren't filtered heavily
to give a more responsive speed display. Note that much software which
displayes recorded tracks from GPS units have a 'remove velocity spikes'
function, which would suggest that speed isn't that filtered, unlike output
position. The position that is displayed is not normally a fix position,
but derived using a filter.

The 201 is a slightly older generation of receiver. Try a journey with
more/less tree cover. If there is more/less variance in speed the above
could be true. It may also be possible to show the 'error radius' on the
display. You may see some correspondence between this and the instantaneous
speed differences.

If you can do a nice long journey in the open, you could 'calibrate' the
speedo with the correction from the GPS.

Happy trails

Skippy
E&OE

Roberto Divia said:

I believe the averages are taken using samples. The frequency of the
samples is - usually - adjustable. On my Garmin, I can use a "by
time" method, a "by distance" method and a proprietary method by
Garmin, probably a mixture of the two previous methods. In all cases
it is possible to use different frequencies. The default was a sort
of average value, if I remember well for the Garmin optimized method.

I'm fairly certain that the frequency of sample you are referring to
only affects the stored tracklog data, not the instantaneous data shown
on the screen.

Anthony

Anthony Jones said:

Roberto Divia said:

I believe the averages are taken using samples. The frequency of the
samples is - usually - adjustable. On my Garmin, I can use a "by
time" method, a "by distance" method and a proprietary method by
Garmin, probably a mixture of the two previous methods. In all cases
it is possible to use different frequencies. The default was a sort
of average value, if I remember well for the Garmin optimized method.

I'm fairly certain that the frequency of sample you are referring to
only affects the stored tracklog data, not the instantaneous data shown
on the screen.

I was talking about the average values, not of the instantaneous data.

In other words, the instantaneous values - where the measurement is not
dependent from the sampling - give you one number while the average and
total values are calculated using the sample points (which is logical,
if you want to avoid measurement errors and temporary fluctuations in
the point estimate).

Ciao,
--
Roberto Divia` Love at first sight is one of the greatest
Dep:PH Bat:53 Mailbox:C02110 labour-saving devices the world has ever seen
Route de Meyrin 385 ---------------------------------------------
Case Postale Phone: +41-22-767-4994
CH-1211 Geneve 23 CERN Fax: +41-22-767-9585
Switzerland E-Mail: [email hidden]

On Thu, 08 Jun 2006 21:29:06 -0400, "(PeteCresswell)" <[email hidden]>

Quoted message said:

Been playing around with a Garmin ForeTRex 201 GPS receiver.

On the ride to/from work today, I put the GPS next to my CatEye Velo.

Funny thing:

I think you have too much time on your hands. If you are single, look
for a second girlfriend. If you are married, look for the first.

That should cure this problem completely.

Roberto Divia said:

Quoted message said:

I'm fairly certain that the frequency of sample you are referring to
only affects the stored tracklog data, not the instantaneous data shown
on the screen.

I was talking about the average values, not of the instantaneous data.

In other words, the instantaneous values - where the measurement is not
dependent from the sampling - give you one number while the average and
total values are calculated using the sample points (which is logical,
if you want to avoid measurement errors and temporary fluctuations in
the point estimate).

Hmmm, I see your point now, and while I agree there is some logic in the
system you're suggesting, I still think your conclusion is incorrect.
The option to change the frequency of points recorded to the tracklog is
required because the GPS unit has limited storage. Why should changing
this option affect the other data presented on the screen? There is no
requirement for it to do so.

Anthony

Anthony Jones said:

Hmmm, I see your point now, and while I agree there is some logic in the
system you're suggesting, I still think your conclusion is incorrect.
The option to change the frequency of points recorded to the tracklog is
required because the GPS unit has limited storage. Why should changing
this option affect the other data presented on the screen? There is no
requirement for it to do so.

Having re-read your original message, I think I agree with your final
suggestion: "Try to increase the sampling rate and see if that helps." I
doubt that it will, but I guess there's only way to find out!

Anthony

I missed the first post so please excuse me if I am off topic.

I ran a test of GPS vs. a carefully calibrated bike speedo
(using the roll-out method with weight on the bike).

On a straight road the distance logged matched perfectly.
On a twisty XC trail the GPS recorded slightly less
distance. I assume this was due to the sample rate of my
Magellan GPS which would not pick up the full travel
distance around sharp turns.

The data shown on the screen of a gps is not instantly
updated. It takes time to do the calculations so the
position is typically updated every second for most consumer
GPS units.

Anthony Jones said:

Roberto Divia said:

I believe the averages are taken using samples. The frequency of the
samples is - usually - adjustable. On my Garmin, I can use a "by
time" method, a "by distance" method and a proprietary method by
Garmin, probably a mixture of the two previous methods. In all cases
it is possible to use different frequencies. The default was a sort
of average value, if I remember well for the Garmin optimized method.

I'm fairly certain that the frequency of sample you are referring to
only affects the stored tracklog data, not the instantaneous data shown
on the screen.

Anthony

All GPS systems will read long in perfect satellite reception conditions due
to the zig zag effect produced by the underlying accuracy of the GPS system.
Under average conditions they come out with reasonable average speeds and
total distances because, when they frequently lose signal, which they do,
then, on any curved track they draw a straight line between the last point
they had a good signal and the current point.

My experience with my Garmin unit is that it achieves on average only an 85%
full signal availability so it is cutting quite a few corners on a long
ride. This seems to virtually compensate for the zigzag effect and I get
reasonable average speed and distance results compared to my standard cyling
computer. At any time during the early part of the ride, however, they can
differ quite markedly.

Do not forget, at present, GPS units are very good at telling you where you
are to within a few metres (10s of feet) but as such they are still not that
reliable as instantaneous speed and distance instruments.

Graham

Per Anthony Jones:

Quoted message said:

"Try to increase the sampling rate and see if that helps." I
doubt that it will, but I guess there's only way to find out!

Sampling rate has been maxed from the start.

"Helps" might be too strong a word.... There isn't that much riding on the
moment-to-moment accuracy as long as the numbers are comparable and don't jump
around randomly too much.

I use it mainly when paddling my surf ski to get an idea whether paddle
technique changes are helping raise speed/reduce effort.
--
PeteCresswell

Per - Bob -:

Quoted message said:

I think you have too much time on your hands. If you are single, look
for a second girlfriend. If you are married, look for the first.

That should cure this problem completely.

I was waiting for that one from the moment I hit "Send"...-)
--
PeteCresswell

"Roberto Divia" <[email hidden]> wrote in message
news:[email hidden]...

Quoted message said:

(PeteCresswell) said:

- On instantaneous speed, the GPS almost always read between .5 and 1.5
mph
faster than the speedo. Sometimes it was less... but the vast majority
of the
time it was more. One stretch that must've lasted 10 minutes I held
the
speedo at exactly 12.8 mph. During that entire time the GPS read at
least .5
mph higher, and sometimes more like 1 mph higher.

- But the average speed and distance covered were almost exactly the same
on the
GPS and the speedo. Within a half mile total distance out of 24 miles
and
within about .3 mph average speed.

[...]

Quoted message said:

Anybody have an explanation?

I believe the averages are taken using samples. The frequency of the
samples
is - usually - adjustable. On my Garmin, I can use a "by time" method, a
"by
distance" method and a proprietary method by Garmin, probably a mixture of
the two previous methods. In all cases it is possible to use different
frequencies. The default was a sort of average value, if I remember well
for the Garmin optimized method.

Now, let's assume the bike goes along a straight line. The sampling
frequency
is irrelevant: two samples or N samples must give the same output.

Let's take the opposite example: a circular path. Here sampling frequency
is
very important. If the circle has a circumference of 1 mile and the
sampling
frequency is 1 mile, than the overall distance is zero and so is the
average
speed. Increasing the sampling frequency will bring the calculated average
closer and closer to the real average, without reaching an exact match.

In conclusion, if the algorithm used by the GPS to compute the average
speed
is based on the data points, then I'd expect this to be slightly below the
actual average speed. The error should be proportional to the sampling
rate.

Try to increase the sampling rate and see if that helps.

Ciao,
--
Roberto Divia` Love at first sight is one of the
greatest
Dep:PH Bat:53 Mailbox:C02110 labour-saving devices the world has ever
seen
Route de Meyrin
---------------------------------------------
Case Postale Phone: +41-22-767-4994
CH-1211 Geneve 23 CERN Fax: +41-22-767-9585
Switzerland E-Mail: [email hidden]

The values displayed on the unit (not just the 'current' ones) are derived
seperately from the the track storage, IME. I've used Garmin kit on a
number of occasions and values such as trip and odometer distance, mean
speed etc are different from those derived once track data is uploaded to
another device.

I would seriously doubt if Nyquist comes into the values used for the
display. Just because the output is at 1PPS, and the storage is at per 50m
or whatever, does not mean that the speed and odometer readings are either
using this sample frequency, or that this is as good as it gets.

Previous investigation of GPS suggests that the unit may take 4 fixes per
second and feed this into a filter to output a position expected at the time
it is to be output, but only 1 per second. Clever stuff in the 19080s but
pretty easy these days.

Skippy
E&OE

On Fri, 09 Jun 2006 17:02:09 -0400, "(PeteCresswell)" <[email hidden]>

Quoted message said:

I use it mainly when paddling my surf ski to get an idea whether paddle
technique changes are helping raise speed/reduce effort.

Dear Pete,

http://www.surfskis.com

Er, sort of a time-trial kayak, right?

Cheers,

Carl FOgel

Per [email hidden]:

Quoted message said:

http://www.surfskis.com

Er, sort of a time-trial kayak, right?

No, that would be a flat water racing kayak.

Most surf skis are designed to be paddled/raced in the open ocean. Their
special functionality is that then can ride the non-breaking ground swell.

Part of the skill/luck in an ocean race is jumping from swell-to-swell and
riding each until it fades. Top racers can average 9.5 mph over long
distances - like 30+ miles - if the race is "downhill" i.e. going in the same
direction as the swell.

A subset of surf skis - and actually the original surf skis - are designed to
let a lifeguard get out through breaking surf to rescue somebody in trouble and
then get back to the beach. I think these originated in Australia. They're
different from ocean skis in that they have more rocker - for turning - and a
bow plane/snitch/hammerhead in the front to keep the nose from submerging when
taking off on a wave.
--
PeteCresswell

(PeteCresswell) said:

Been playing around with a Garmin ForeTRex 201 GPS receiver.

On the ride to/from work today, I put the GPS next to my CatEye Velo.

Funny thing:
-------------------------------------------------------------------------
- On instantaneous speed, the GPS almost always read between .5 and 1.5 mph
faster than the speedo. Sometimes it was less... but the vast majority of the
time it was more. One stretch that must've lasted 10 minutes I held the
speedo at exactly 12.8 mph. During that entire time the GPS read at least .5
mph higher, and sometimes more like 1 mph higher.

- But the average speed and distance covered were almost exactly the same on the
GPS and the speedo. Within a half mile total distance out of 24 miles and
within about .3 mph average speed.
-------------------------------------------------------------------------

If the averages/totals weren't so close I'd just chalk it up to improper setting
of the wheel diameter on the speedo.

If it weren't for that 10-minute stretch, I'd attribute it to latency.

If the total distances weren't so close, I'd suspect that somehow I was
selectively remembering the times the GPS read higher and ignoring the ones when
it read lower.

My kneejerk is that the micro code in the GPS is smarter than just averaging out
the instant speeds... but in the back of my mind Occam is whispering....

Anybody have an explanation?
--
PeteCresswell

The GPS should always be accurate for total distance and will be
perfect for error correction, especially if you have a wireless
speedometer. The situtation you described is precisely what one would
and should expect. The current speed of a GPS system is going to have
some latency in terms of what the current display indicates. This is
not the same thing as being inaccurate. There is just less precision on
the smaller resolution data whereas the larger the distance the more
accurate it gets. The speedo is the opposite in that it has almost no
latency (well, very little and we are used to it) and any errors in
measuring get multiplied over the distance it reports. What I would do
is compare the distances and assume that the GPS is as accurate as you
can get. Not any difference and then see if you find a consistent trend
that you can correct in the speedometer.

Per (PeteCresswell):

Quoted message said:

- On instantaneous speed, the GPS almost always read between .5 and 1.5 mph
faster than the speedo.

Quoted message said:

but in the back of my mind Occam is whispering....

Anybody have an explanation?

Occam finally returned my call.

I have two bikes with different diameter tires.

Rode the second bike today and found the GPS consistently *slower* than the
speedo. It was consistently *faster* on the first bike.

Turned out that somewhere along the line I had inadvertently swapped the speedos
so that one bike had a speedo that was calibrated for a too-small tire and the
other for a too-large tire.

Swapped them back, re-calibrated the speedos just to make sure (turns out
there's about 1 cm diff between rolling a fat-tire bike unloaded and rolling it
with a rider's weight....). Now the instantaneous speeds of the GPS and the
speedo are reasonably close on both bikes. The total distances are right on
the money and the average speeds over a very short (<.5 miles) trip are within a
tenth of an MPH.
--
PeteCresswell