Page 2 of 3 FirstFirst 1 2 3 LastLast
Results 8 to 14 of 18

Thread: Navi GT-31

  1. #8
    Junior Member
    Join Date
    Feb 2013
    Location
    Cape Cod, USA
    Posts
    25
    If you compare any two GPS devices, you will get almost identical results for most runs most of the time. But the general problem with positional data is that they occasionally tend to have spikes. One common cause is a change of the number of satellites tracked. For an example, look at this blog post:
    http://boardsurfr.blogspot.com/2011/...oid-phone.html

    It is possible to get GPS runs that are practically identical with positional data and Doppler data. But it is just as possible that the positional data show crazy spikes that are completely absent in the Doppler data. I have seen tracks from several guys who thought they had broken 40 knots based on positional data (e.g. Garmins), where the speed through the entire session was below 30 knots, with just one or two "40 knot" spikes.

    One advantage of the Doppler data is that they also give you precision estimates. For the GT-31, they are typically in the 1-2 knot range. Therefore, a speed of 40.4 knot is not really different from a 40.0 knot speed. For record claims, the error is generally subtracted from the top speed, so give a speed that has actually been reached with 95 percent confidence. That is after careful examination of data from 2 GPS units; data where the two units disagree by too much are generally thrown out as bad. I think GPS Results can do that automatically.

    The GT-31 is far from perfect, but with proper analysis, the Doppler speeds it gives are significantly more reliable and accurate than those of Garmins and other position-only units. Newer units like the Suuonto watch or even smart phones could be useful, but only if they would generate data files that contain the doppler speed information and the error estimates. Many of them (including the Suuonto watches) are useless because they do not actually log speeds in 1-second intervals, but in larger intervals. For accurate analysis of speed in turns (i.e. alpha 500), you definitely want a 5 Hz or 10 Hz GPS chip.

  2. #9
    Senior Member
    Join Date
    Nov 2009
    Posts
    3,738
    I assume that Locosys have done pretty well with the GT11 and GT31 especially when the GT 31 is almost the default for water based speed events. As far as I can see it is only a rumour that production of the GT 31 has been stopped, but if that is true, is it not more likely that Locosys have developed a replacement/updated model? Regardless, the number of GT31 in use in Windsurfing means it will remain an accepted GPS for a few seasons yet.

  3. #10
    Junior Member
    Join Date
    Feb 2013
    Location
    Cape Cod, USA
    Posts
    25
    Quote Originally Posted by mikerb
    As far as I can see it is only a rumour that production of the GT 31 has been stopped, but if that is true, is it not more likely that Locosys have developed a replacement/updated model?
    The "rumour" is based on several letters that Locosys has sent out to different people. The Sirf3 chip inside the GT-31 is ancient - a newer version has come out about 5 or 6 years ago. Apparently, production of the Sirf3 chip has now stopped.

    The market for windsurf GPS units is simply too small to matter to any decent sized manufacturer. Even if every person who ever posted to the GPS sites (GPS3 and GPSTC) bought a GT-31, that's still a lot less than a million dollars in total revenue (which then has to be split up between manufacturer and resellers). That's why Locosys never updated the GT-31. For every speedsurfer, you have hundreds or thousands of runners, hikers, and triathletes; that's why the Suuonto watches have great technology to compensate for arm movements when walking, running, or swimming, but no doppler or 1-Hz recording for speedsurfing.

    Over the past few year, several technologically savvy speedsurfers have created GPS prototypes that were clearly better than the GT-31. For example, units based on the ublox GPS chip run circles around the GT-31 with respect to accuracy and recording rate. But the units have never become commercially available because the market is so small.

    We have been rather lucky to find someone who has already developed a commercially successful GPS unit dedicated to a niche market - the FlySight (for wingsuit jumping etc.). The FlySight developer happens to be a very nice guy who is interested in creating a new units specifically for speedsurfing and similar sports. Prototypes are currently being developed, with a lot of input from the "gurus" who decide which GPS units are officially accepted at the speedsurfing web sites.

    Quote Originally Posted by mikerb
    Regardless, the number of GT31 in use in Windsurfing means it will remain an accepted GPS for a few seasons yet.
    That's a safe bet.

  4. #11
    Senior Member Capie's Avatar
    Join Date
    Jan 2013
    Location
    Cape Town
    Posts
    842
    Quote Originally Posted by boardsurfr View Post
    Many of them (including the Suuonto watches) are useless because they do not actually log speeds in 1-second intervals, but in larger intervals.
    Not true in relation to my Suunto watch. Mine (Ambit 2S) records speed every second.

    The way I understand it, the chipset has a minimum interval but it can't record anything if it doesn't have a signal. The trouble with smartphones and watches is not the frequency (which is actually 0.7 seconds in the case of the iPhone) but the fact that their aerials aren't optimal for GPS reception. So the iPhone doesn't always have as good a signal to work with as dedicated GPS units and waits till it gets one before recording data. http://forums.mtbr.com/gps-hrm-bike-...ld-771627.html

    It seems that what you're saying is that the accuracy of doppler helps when you're looking at short sample times. So, for windsurfing, 1-second peak speed and perhaps gybes where you may only have 4 or 5 pieces of data to analyse. It sounds like the difference is negligible for most of the standard measurement, such as 500m runs particularly if the software is good at identifying errors.

    If doppler is more accurate, why do other applications where speed is more critical not use doppler? I'm asking that seriously, not trying to be obtuse.
    Last edited by Capie; 8th January 2014 at 11:06 PM.
    My Boards: 2016 Fanatic Falcon TE 129, 2014 Patrik Slalom 115 vII, 2014 Patrik Slalom 92l, 1992 Windsurfer One Design, 2012 Fanatic Freewave 85l
    My Sails: North Sails Warp f2016 , North Hero, North Volt

  5. #12
    Junior Member
    Join Date
    Feb 2013
    Location
    Cape Cod, USA
    Posts
    25
    Quote Originally Posted by Capie
    Not true in relation to my Suunto watch. Mine (Ambit 2S) records speed every second.
    Do you actually have data files from sessions longer than an hour that have the speed for every second? I'd love to see one. The only examples with data every second I have seen so far were from short sessions (20 minutes or so).

    Quote Originally Posted by Capie
    It seems that what you're saying is that the accuracy of doppler helps when you're looking at short sample times.
    No, I am not saying that. Errors are more pronounced at short sample times, by they do not disappear if you look at larger numbers. If you look at bad positional data, you'll see zig-zags and sudden jumps. Both lengthen the apparent distance, and thus increase speed.

    Anyway, the most common used measure for top speed is the 5 x 10 second average. Even one spike in these 5 x 10 seconds can have relatively large effects. For the GPS Team Challenge, one of the measures is the 2 second top speed.

    Quote Originally Posted by Capie
    If doppler is more accurate, why do other applications where speed is more critical not use doppler? I'm asking that seriously, not trying to be obtuse.
    Not sure which "other applications" you are referring to - but keep in mind that most GPS units actually use Doppler measurements to improve the accuracy of their track points. Check out this image of tracks from the iPhone and 3 GPS devices (from this blog post):

    You can see the jumping of positions and the effect of filters quite well. The filters try to keep the positional data from jumping too much, but sometimes, they over-do it, and then have to make huge adjustments. I had one session where a GPS unit I tested reported speeds of several hundred (!) knots multiple times (and that unit used a very modern GPS chip - but perhaps it was defective). The tester who created the tracks shown above was walking a reasonably straight line - I think the image illustrates quite well how track points can over-estimate distance and therefore speed. The positional error is constant, independent of speed; so the faster you move, the lower the relative error will be. At highway driving speeds, estimating speed from positions is usually accurate enough.

    The "positional" data generally are created by some proprietary algorithm that filters and adjusts the calculated coordinates based on doppler speed and previous coordinates. Some GPS chipsets that do not support writing doppler data directly still use the doppler data to calculate speed, and report speeds that are different from the speed calculated from the coordinates alone. But how these speeds are calculated is basically a trade secret. Doppler data have the theoretical appeal that they are much closer to "raw" data.

    That said, the "positional" data reported by the FlySight GPS are a lot more accurate than the data from the GT-31. Here's an example from a recent session:
    Click image for larger version. 

Name:	Speedgraphs.jpg 
Views:	728 
Size:	23.0 KB 
ID:	9382
    You can see a lot of spikes in the GT-31 positional speed data (blue, upper part) that are not present in the FlySight data (red) and the doppler data (lower part). The FlySight positional speeds appear to be as accurate as the GT-31 doppler speed data. I guess this is probably because the FlySight uses a better filter for the positional data that heavily relies on doppler data; the 5-fold higher recording rate probably helps a lot, too. The high recording rate allows it, in effect, to use the doppler data as accurate accelerometer data.

  6. #13
    Senior Member Capie's Avatar
    Join Date
    Jan 2013
    Location
    Cape Town
    Posts
    842
    Quote Originally Posted by boardsurfr View Post
    Do you actually have data files from sessions longer than an hour that have the speed for every second? I'd love to see one. The only examples with data every second I have seen so far were from short sessions (20 minutes or so).
    I'm confused. I thought you said the Suunto was "useless" because it did not log data in 1-second intervals. Now you're saying you have seen data that shows it recording 1-second intervals in short sessions?

    Certainly my Suunto doesn't record 3,600 points per hour during a windsurfing session. The way I sail, it would be a miracle if any GPS device managed to maintain a perfect signal every second including while I was upside down under water. My watch doesn't give up after 20 minutes or hit some hardware limit. It sometimes loses signal but it always tries to record speed every second. I doubt it would be any different with a GT-31.

    I leave my data on Movescount and don't bother analysing it with other software so I honestly can't tell you how many times it didn't record every second. When I checked, most of the points were recorded every second. For my purposes, Movescount is more than sufficient. I have analysed a couple of tracks with GPS Action Replay just to see if the sky really was falling down, i.e. that my Suunto is so inaccurate as to be "useless". I was satisfied that this is not the case.

    The 1-second and 2-second max speed outputted by GPSAR is always higher than the max speed the watch records (which I understand to be a 2-second adjusted speed). The last track I looked at, the watch (and Movescount) recorded 25.5 knots max. GPSAR had 5, 1-second bursts over 26 knots with the highest at 27.77 knots and an average of around 26.4 knots. For 2-second runs, the average of 5 runs on GPSAR was 25.5 knots with some 2-second splits higher than that. I'm pretty sure the 27.77 knots measurement was a spike. The speeds before and after that one were around 25 knots so I doubt I went 2 knots faster for just one second. Some of the other 1-second speeds are probably also spikes and in any event, maintaining a speed for 1 second is in my view not really something to be proud of. The average of the two second runs was exactly what the watch gave me so for my purposes, being told simply that my max was 25.5 knots is perfect.

    I will, when I have a chance, download the files and let you have a look.
    Quote Originally Posted by boardsurfr View Post
    You can see the jumping of positions and the effect of filters quite well.
    I would think that the jumps in those images have a lot to do with the fact that he's walking (low speed) in a city between buildings that interfere with the signal.

    As you note, positional data can give a very accurate reading. It's interesting that you mention the Flysight because that's one of the applications I was referring to. In paragliding for example, where stall speed is critical, a lot of guys using GPS units that use positional data only, including smartphones!

    At the end of the day, I think there's a little bit of disdain for GPS watches and smartphones from the speedsailing community because they like the idea that what they are doing is very high-tech and inaccessible to mere mortals with just the phone in their pocket. Not saying this is you, boardsurfr. Clearly you recognise some of the issues with the GT-31. However, I think that information we get from the speedsailing community needs to be passed through this filter. There is a tendency to overstate the problem with other devices and to overstate the GT-31's abilities. You're guilty of that too when you say that the Suunto is "useless".

    I do find it puzzling that no one in the speedsailing community has tried to compare the latest iPhone or Samsung, latest Suunto Ambit and the GT-31. I have an iPhone 5S and an Ambit. I guess I could borrow a GT-31 and do a test although frankly I don't really care too much about it. I'm already convinced that the Suunto is accurate enough for me.
    My Boards: 2016 Fanatic Falcon TE 129, 2014 Patrik Slalom 115 vII, 2014 Patrik Slalom 92l, 1992 Windsurfer One Design, 2012 Fanatic Freewave 85l
    My Sails: North Sails Warp f2016 , North Hero, North Volt

  7. #14
    Junior Member
    Join Date
    Feb 2013
    Location
    Cape Cod, USA
    Posts
    25
    Quote Originally Posted by Capie
    Certainly my Suunto doesn't record 3,600 points per hour during a windsurfing session.
    And that is the problem. My assessment of the Suuonto is based on (a) what another very disappointed Suuonto watch owner had posted on the Seabreeze forum, and (b) looking at the tracks on Movescount, where the apparent recording rate goes down as the session length goes up.
    BTW, the GT-31 does actually record 3600 points per hour, and the FlySight records 18,000 points per hour, even with wet jibes.

    Quote Originally Posted by Capie
    I think there's a little bit of disdain for GPS watches and smartphones from the speedsailing community because they like the idea that what they are doing is very high-tech and inaccessible to mere mortals with just the phone in their pocket.
    That analysis in incorrect. Anyone is quite welcome to use whatever he likes for his own purposes. But if you want to post your speeds to the speedsurfing sites, you need to follow their rules. Those rules are challenged many times a year by someone claiming that their device is "good enough". Typically, these claims are made on a very cursory analysis of very few tracks, or perhaps just the speed results from one or a few sessions. Often, those who make these claims make very clear that they have a very limited understanding of GPS technology. There are plenty of discussions online on the various windsurf forums and the GPS3 forum where the same issues were discussed years ago.

    The speedsurfers who decide which GPS units are acceptable for the GPS web sites are also involved in official record attempts and their validation, which adds a whole new level of stringency. For GPS3, that means that they carefully evaluate each record claim, which means comparing traces from two GPS units worn during the attempt. They also are part of leading speedsurfing teams, and thus share the experience of the entire teams (which may have a dozen 40-knot speedsurfers). They have all seen examples where someone got 40-knot speeds from a Garmin or even a GT-31 display due to spikes, even when the same sailors in reality had a hard time getting close to 30 knots. That experience can be demotivating, to say the least.

    A couple of years back, I bought an Android phone so that I could hear my speeds while sailing, using the GPS Speed Talker app. That was based on a suggestion by a speedsurfing guru - actually a guy who was involved in the development of the GT-31. Great suggestion - this is definitely very useful for improving your speed. But within one or two sessions, I got several examples where the accuracy was bad, and my top speeds were overstated by several knots. So I'd never think about using data from this phone to post to the GPSTC or GPS3.

    There are many speedsurfers who would love an alternative to the GT-31 that is at least as accurate. It does not matter whether the device calls itself a phone, wrist watch, or sports GPS. Promising new devices are closely monitored, and often bought and tested (I spent $500 last year on alternative GPS units I wanted to test). Usually, the results are very frustrating. WinTec has several units that are cheap, and use a GPS chip that's much better than the Sirf3 in the GT-31. But they chose a file format that does not include all the necessary data. The Suuonto watches (and their GPS pod!) is even more frustrating. They have great GPS chips, and they can logs tons of data. The chip can give doppler speeds, but Suuonto has chosen not include doppler data in their files. Requests to do so by several speedsurfers have fallen on deaf ears. That's always frustrating, but even more so when it comes from a company that usually listens to customers, and releases firmware updates on a regular basis. One of their engineers could probably modify the firmware within less than one day...

    I would think that the jumps in those images have a lot to do with the fact that he's walking (low speed) in a city between buildings that interfere with the signal.
    The walking has no effect on the positional accuracy - it only makes it easier to see. Signal interference from buildings will have some effect, but if you look at the blog post and scroll down to some other images, you can see very similar data in areas where the reception was clean (a big parking lot and close to the river).

    Anyway, enjoy your GPS watch, smart phone, or whatever you choose to use to record your sessions. Most of the time, any of these devices will give you a good idea of your approximate top speed, especially if you look at your data a bit closer. My "useless" comment was only with respect to posting data to GPS-speedsurfing.com or the GPS Team Challenge. But for your own entertainment, session records, improvement, etc., the Suuonto watch should be just fine.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •