PDA

View Full Version : Edge 500, GSC-10, and Wheel Size



TEXMURPHY
01-29-2010, 10:30 AM
I have the Edge 500 + GSC-10 settings for a wheel size of 2096, which is roughly my roll-out calculation (2 revolutions divided by 2) and a "normal" setting for 700cc wheels plus 23 tires.
I have done about 140 miles over 4 rides with the 500 (parallel with my Edge 305). I notice there is a difference between the 500 reported distance and the gps calculated difference which is enough to throw off many stats for the rides. In attempting to manually estimate a correct wheel size I use the following formula <correct wheel size> = <gps distance> * 2096 / <Edge 500 distance>. Applying this to my 4 ride sample gives a consistent <correct wheel size> ~= 2086.

Edge 500 GPS <gps>*2096/<Edge500>
46.41 46.17 2085
41.17 40.98 2086
22.32 22.23 2088
30.36 30.20 2085
*Distances in miles
I will adjust my Edge 500 wheel size setting to 2086 and see what happens. But while the error is under 1%, it is noticeable. Interesting that the Edge 305 device distances for the ride are near the gps calculated distances from Edge 500 data. The Edge 500 reports accuracy of 12 ft in the open under clear sky and declining to 18+ ft under trees in overcast.

What are others finding / using for wheel circumference?

T_SMIT
01-29-2010, 12:34 PM
A circumference difference of 1 cm corresponds to a radial difference of about 1.5mm. When you do your rollout for manual measurement, are you loading your bike with a mass equivalent to yourself? You might be seeing the difference due to the tire footprint getting squished under your weight and reducing the actual rolling radius.

ROBERTDAINES
01-29-2010, 04:15 PM
GPS distance and distance measured using the GSC-10 will never be exactly the same, particularly using Smart Recording. Take a look at your track, you will see that the GPS track is not perfect, there will be cut off corners which will result in slightly reduced distance compared to the wheel sensor. This error will be reduced using 1 sec recording as more track points will be used. Set your wheel size to 2096 according to the chart and don't sweat the difference. The wheel sensor will be more accurate. Also, you will never be able to compensate by changing the wheel size as the GPS errors will be different every ride.

TEXMURPHY
01-29-2010, 04:39 PM
GPS distance and distance measured using the GSC-10 will never be exactly the same, particularly using Smart Recording. Take a look at your track, you will see that the GPS track is not perfect, there will be cut off corners which will result in slightly reduced distance compared to the wheel sensor. This error will be reduced using 1 sec recording as more track points will be used. Set your wheel size to 2096 according to the chart and don't sweat the difference. The wheel sensor will be more accurate. Also, you will never be able to compensate by changing the wheel size as the GPS errors will be different every ride.

Since wheel sensor should be more accurate, I am attempting to set wheel size based on characteristics of my wheels and ride style. When gps calculated distance is always different from the Edge 500 gsc-10 distance, then I feel I should adjust the wheel circumference setting.

While the difference between 1 second and smart recording was significant on the Edge 305 (and I believe the Edge 705), it is relatively insignificant with the Edge 500. The Edge 305 smart recording averaged about 5 seconds per trackpoint whereas the Edge 500 smart recording places a trackpoint less than every 2 seconds.

VBERNASCONI
01-30-2010, 10:19 AM
In my experience, for a 700x23 tyre, 8 atm, rider of 80 kg, 2085 mm of wheel circumference is a pretty accurate figure

EKUTTER
01-30-2010, 11:41 AM
One important thing to remember here is that the distance recorded does NOT necessarily equal the accumulated distances between recorded coordinates. The distance that gets recorded includes error correction that you can not account for simply by looking at the coordinates.

If the distance from the wheel vs GPS is always off by a similar amount, I would go with the GPS. In fact, after lots of rides, many on very accurately measured segments, my GPS distance almost always ends up with in a fraction of a percent of a wired bike computer. As long as you aren't doing winding mountain bike trails. The accuracy has been so good that I don't even use the wheel sensor since it can fluctuate by more just by changes in tire pressure.

One other test would be to have both your 305 and 500 go by GPS only and see how close they are. This would tell you if the 305 accuracy is just lower than the latest technology. If they are nearly dead on, you can be pretty sure it is your wheel size calculation.

ROBERTDAINES
01-30-2010, 03:34 PM
One important thing to remember here is that the distance recorded does NOT necessarily equal the accumulated distances between recorded coordinates. The distance that gets recorded includes error correction that you can not account for simply by looking at the coordinates.

What kind of error correction is used? How does the GPS know what to correct? AFAIK, the GPS draws straight lines between track points; it has no way of knowing what should be in between those points. The closer the track points, the more accurate the distance will be. If you have documented information that says otherwise, I would love to read it. If there is correction involved, this does not show up when looking at the track on the map. I am interested in learning more. Thanks.

EKUTTER
01-30-2010, 04:18 PM
What kind of error correction is used? How does the GPS know what to correct?
At least the Edge 305 actually used to contain negative distance segments. The data that gets recorded is accumulated distance since the start. If you analyzed the distance, you would come across data points that had a total distance shorter than the previous data point. I am not sure if this is still the case or if they do this error correction in a different way. In any case, the unit doesn't save anywhere near all the GPS coordinates it calculates but probably does use all of them to calculate distance. I am not sure of the exact rate but it is clearly more often than the smart recording rate and probably more frequent than every second. Also, there are formulas out there for calculating the direct distance between two coordinates and the value you get will not match the recorded distance.

On a slightly different topic, you only know, within a second, of the time each data point was calculated, since the granularity of the time stamps is only per second. I noticed last week with my FR310 data, shortly after I hit the lap button, the speed I would calculate from duration/distance fluctuated wildly between data points where as the speed recorded from the speed sensor was consistent. So if I was riding at 15mph, the recorded speed stated that but if you calculated the speed from the segment duration and segment distance, the speed ranged from 1.5mph to 80mph for 5 or 6 segments. I suspect what is happening is some of the segments are for a fraction of a second while others are slightly longer than a second. But the data file rounds (or truncates) the time to the nearest full second. (in this case, no GPS was involved as I was on the trainer with GPS turned off)

The point of this is that Garmin does some undocumented stuff in calculating the data and it is dangerous to jump to any conclusion about how it does that. I bet that if you had two identical units and set one to record smart and one to record every second, the recorded distances would basically be the same even though the number of recorded data points would be way different. It would be pretty bad if the number of data points they record effected the recorded distance. More it impacts the ability to do post analysis of the data. Given the raw data points, there is no way to recreate the totals, like distance, for the ride.

TEXMURPHY
01-30-2010, 04:51 PM
The Edge 305 distance calculation was based on device known accuracy. As such Edge 305 distances reported for laps, etc. are different from post-processed gps trackpoint distances.
The Edge 500 with the GSC-10 uses the GSC-10 plus wheel circumference to do its distance (and speed) calculations. As a result it is important for the wheel circumference to be as close to real as possible.
As I have noted, the Edge 305 distance calculation is closer to the Edge 500 gps trackpoint post-processed distance, then it is to the Edge 500 device reported distances with an inappropriate wheel circumference setting.

Weather in the mid-Atlantic is preventing any further realworld test at the moment. Although I may do an indoor test with Edge 500 and Edge 305, both with gps off to compare their non-gps distance calculations.

DW152
01-30-2010, 05:16 PM
So if I was riding at 15mph, the recorded speed stated that but if you calculated the speed from the segment duration and segment distance, the speed ranged from 1.5mph to 80mph for 5 or 6 segments. I suspect what is happening is some of the segments are for a fraction of a second while others are slightly longer than a second. But the data file rounds (or truncates) the time to the nearest full second. (in this case, no GPS was involved as I was on the trainer with GPS turned off)


This particular issue is a bit different, and is caused by a discrete number of wheel revolutions per recording interval:

https://forums.garmin.com/showpost.php?p=21503&postcount=5




On a slightly different topic, you only know, within a second, of the time each data point was calculated, since the granularity of the time stamps is only per second. I noticed last week with my FR310 data, shortly after I hit the lap button...

Keep in mind that with smart recording, the 310 has a problem that when the user hits the lap button, it doesn't necessarily immediately store a trackpoint. Instead, it waits until it thinks it should be recorded.

EKUTTER
01-30-2010, 06:56 PM
This particular issue is a bit different, and is caused by a discrete number of wheel revolutions per recording interval:

https://forums.garmin.com/showpost.php?p=21503&postcount=5

Keep in mind that with smart recording, the 310 has a problem that when the user hits the lap button, it doesn't necessarily immediately store a trackpoint. Instead, it waits until it thinks it should be recorded.

I don't think that fully explains it. I was only seeing this huge fluctuation for 5 or 6 data points following the press of the lap button. In the above scenario, you would expect to see it throughout the file. There is still a variation of a couple percent throughout, but not the huge swings I was referring to.

One thing I should have mentioned was that I was also using a power meter so was generally getting every second recording. It seems more like what is happening is the processor starts to lag behind after a lap press in its calculations as it occasionally would go up to three seconds per data point following a lap press. I did account for this in my speed from duration/distance calculations. After a few data points it seems it would catch up and then behave normally until the next lap press.

Also note that I wasn't relying on any distance or time data from the lap information, but rather just noticing that the distance vs speed of the recorded data points were more inconsistent just after pressing the lap button.

TEXMURPHY
01-31-2010, 04:52 PM
...weather in the mid-atlantic is preventing any further realworld test at the moment. Although i may do an indoor test with edge 500 and edge 305, both with gps off to compare their non-gps distance calculations.

I did a trainer ride with GPS=OFF for both Edge 500 and 305.
Interesting to note that the distance at end of 30 minute session with no stops was 0.02 miles (~32 meters) different. That distance change slowly increased over the course of the workout.

Distance over time thanks to OldManBiking's Dot Racing SportTracks Plugin (http://www.zonefivesoftware.com/SportTracks/Plugins/plugin_detail.php?id=92)
http://i47.tinypic.com/1f8sd2.jpg

http://i50.tinypic.com/35kohi1.jpg

http://i45.tinypic.com/iymvb6.jpg

EKUTTER
01-31-2010, 06:41 PM
Probably stupid question, but you did have the wheel size set to the same value, didn't you? As you point at, it is very interesting that the difference was a slow creep rather than just an error over a couple of data points.

Interesting how the graph was so much smoother on the 500. Any idea how the recording intervals compared? Was the Edge 305 set to every second?

TEXMURPHY
02-01-2010, 08:59 AM
Probably stupid question, but you did have the wheel size set to the same value, didn't you? As you point at, it is very interesting that the difference was a slow creep rather than just an error over a couple of data points.

Interesting how the graph was so much smoother on the 500. Any idea how the recording intervals compared? Was the Edge 305 set to every second?
Wheel size set same on both Edge 500 and Edge 305
* Edge 305 set to "Smart recording", Edge 500 default to Smart since "PowerMeter=NO"
* Circumference = 2096
* Autopause = 2.0 mph
* Autolap = 0.50 miles
* Since GTC will NOT allow identical start time activities, the start time of the Edge 500 was set back 1 sec (otherwise times unaltered)
Note that SportTracks does not have this restriction and the distance over time graph previously posted used unaltered timing.
* Procedure: Start timer, wait for both 500 & 305 to autopause, then start pedaling.
* No breaks in pedaling over course of test.
Attached xls is from combined uploads of tcx xml exports from GTC.

1st 1500 meters Edge 500 / 1000 meters Edge 305 below:
http://i48.tinypic.com/21l1xj8.jpg

EKUTTER
02-01-2010, 11:19 AM
Have you by any chance tried this without the half mile auto lap? The FR310 had an issue (at least when doing a workout) where every lap, the distance would get slightly off. People doing workouts with short distance intervals were noticing they were getting out of sync with their friends that had a FR305 with the same workout. They have since fixed this in the latest firmware for the FR310. But it seems like the Edge500 and FR310 have a lot in common. I suspect the underlying hardware (and thus software) might be nearly identical.

TEXMURPHY
02-02-2010, 04:30 PM
I have decided to turn off all autolapping for now. Most of my lap usage was for post ride analysis and I can use SportTracks and its plugins to better look at lap sections.

Autopause and the gsc-10: I have autopause set for 2 mph. The Edge 500 uses the gsc-10 for speed and distance and so takes longer to detect motion than the Edge 305. In fact it looks like it will take 2 or more wheel revolutions to compute motion and speed. So 2*<2086> or 4+ meters will not be recorded as distance traveled. Taking all pauses into account, this can add up. I am considering changing my autopause from speed="2mph" to "When Stopped".