Forums > Windsurfing   Gps and Speed talk

Hope for a GPS alternative

Reply
Created by boardsurfr > 9 months ago, 1 Dec 2017
sailquik
VIC, 6068 posts
3 Dec 2017 8:36PM
Thumbs Up

Select to expand quote
AUS02 said..
Thanks Roo. Just tried it driving around in the car and it does show your speed (currently 55.05 knots), once above the set minimum, and also your average speed since starting (15.16 knots), elapsed time (00:54:39) and distance travelled (25.57 km), which is all what I was after, but the average speed is only in small font until you complete the 1hr. I wonder whether it would be worth making the change to have the average speed shown above the current speed, where the 'na' is in the image below? This would make it a lot easier to see while sailing? Not sure what happens once you complete your hr, perhaps it could 'flash' and keep tracking your average speed? Thoughts?




Quite an interesting hack to try to get around this issue.
However, doen't the 'Genie' reset every time you drop below the minimum speed set? If that is the case, every time you stop, crash or drop off the plane in an hour attempt, it will reset and start again? That could get quite annoying.

boardsurfr
WA, 2202 posts
3 Dec 2017 9:56PM
Thumbs Up


sailquik said..


It would be nice if I could fine some affordable waterproof BT earphones!


I tried some cheap BT earphones with "waterproof" claims that died after the first crash. But recently, I got these for $25:
https://www.amazon.com/gp/product/B06XB4Z58Q/
They have lasted through a few sessions with crashes and swims. The waterproofing looks better than on the other ones I tried. I also like the way they fit on my ear, they hold much better than most earphones I used for windsurfing (wired or wireless). My wife, who has smaller ears, says they don't fit her, though.

Another recent find on Amazon was this waterproof arm bag:
www.amazon.com/gp/product/B01HM9PFPG/
It has one of the best bands I have seen so far, and the attachment on the back should last longer than most cheap backs. I still double-bag the phones since it will leak sooner or later. You guys have better bags in Australia that are hard to get in the US, but this may be of interest to someone in the US.

boardsurfr
WA, 2202 posts
3 Dec 2017 10:17PM
Thumbs Up

Select to expand quote
sailquik said..
Back to Boardsurfr's original post.

Mat and I bought a couple of those Ublox GPS dongles about a year ago. I tested it on my computer and it seemed quite good.

The problem we had with connecting to to our phone was the one you mention in your Blog that I have highlighted below:

"A Samsung Galaxy J3 Luna Pro phone running Android. I bought it for $30 at Best Buy.
A USB GPS dongle with a U-blox 7 chip that I bought for $15 from Amazon (like this one).
A USB "On-The-Go" cable for about $5 (like this one) to connect the dongle to the phone (note that the phone must support USB OTG - not all phones do!)."

I remember that YOYO had the same idea well before that, but had trouble finding a phone that had USB OTG as well.

Things may have changed by now and obviously you have found at least one inexpensive phone that does support USB OTG in the Samsung Galaxy J3 Luna Pro. Now we need to see if it available with that spec in Australia at a reasonable price.


Finding a cheap phone with USB OTG support is still a bit of a challenge. The first time I played around with the USB dongle, I rooted the Galaxy J1 I had, but even that did not help. This time, I checked online what phones were available for cheap in store, and then did a Google search for the specs. Here's an example from gsmarc.com for the J3 Luna Pro:


The "With OTG" is what's needed. Sometimes is says "Mass Storage", which is the same thing. I like to see it on a couple of different sites, though.

The price of $105 for the J3 in Australia is similar to the regular price for the J3 in the US if you convert the currencies (~ $80 US). In Europe, I found some cheap phones with OTG for about 60 Euros. Here in the US, the big box electronic store always has a bunch of "prepaid" phones on sale. I got lucky finding something for $30, but finding a phone with USB OTG support for $50 is usually no problem. Check the size, though - the J3 just barely fits into my arm band bag with the cable attached. A phone with a 5.5 or 6 inch screen may not fit at all, unless you have an extra-large bag.

boardsurfr
WA, 2202 posts
3 Dec 2017 10:34PM
Thumbs Up

Select to expand quote
sailquik said..
Not in the standard settings. The problem as explained to me is that it would take a huge amount of processing and memory capacity to have a 1 hr running average calculation in real time.


1 hour calculations need a bit of memory, but not processing power. The straightforward implementation would have to save the traveled distance for the last hour; at 5 Hz, that's 18,000 points, or 72 kb of data. That's too much for most micro controllers, which have a total memory of somewhere between 2 kb and 64 kb. So Locosys concluded the watch can't do hours.

However, there are some practical shortcuts possible that enable a slightly less accurate calculation with a lot less memory. For example, saving the traveled distance every 10 seconds reduces the amount of data 50-fold, to 360 points. That can be saved in less than 2 kb of memory (which would not even need to be RAM, since updates happen only every 10 seconds). The effect on accuracy would be small - worst cast error is 10/3600, or 0.3%. For a 20-knot hour, the worst case error would be 0.06 knots. The typical error would be a lot lower.
The amount of additional processing would be minimal. 10 second values are already calculated 5 times per second. The one hour value would need to be updated only every 10 seconds.

Given a choice between no numbers for 1 hour and numbers that may be wrong by 0.3%, I'd rather have the slightly less accurate numbers.

Roo
765 posts
3 Dec 2017 11:34PM
Thumbs Up

Select to expand quote
AUS02 said..

Roo said..




AUS02 said..





boardsurfr said..








bartontb said..
In MHO, the GPSLogit app interest was killed by the GPSTC decision not to approve the phone solution.











I think that the non-approval "killed" GPSLogit is quite an overstatement. The fraction of speed surfer who want to hear their speeds while sailing is pretty darn small. I'm in that minority sometimes, but in most sessions, I don't want anything in or on my ears. Many windsurfers find convenience and waterproofing much more important. Several guys in our team used non-approved watches, and simply did not post, rather than bothering with waterproof armbands, before the GW-60 was released.

I simply think more choice is better. Some guys still like the Canmore best, despite its well-known shortcomings. Developing and testing prototypes is good fun, but if you can't buy the things in the end, it's rather pointless. A bit like having a GPSTC team where there's just one guy posting all the time.

As for why there are about 10-fold fewer users for GPSLogit than for Windsport Tracker, maybe it's differences in functionality. Or maybe it because of the difference in the text on the little green button:










Can either of these Apps calculate average speed while doing a 1hr (and if not could they)? You'd no doubt need a 'start' button to signify the start of the one hour and then it would track your average speed as you go and count down the hour or elapsed time. I think a reset button would be needed, so if you decide to restart the 1hr you can while sailing by just pushing the reset button.






GPSLogit can. You just set it to that in settings and it reads out on the screen. It also tells you via earphones once you have sailed for an hour and then drop below your speed threshold. I use it when I'm doing hour attempts.





Thanks Roo. Just tried it driving around in the car and it does show your speed (currently 55.05 knots), once above the set minimum, and also your average speed since starting (15.16 knots), elapsed time (00:54:39) and distance travelled (25.57 km), which is all what I was after, but the average speed is only in small font until you complete the 1hr. I wonder whether it would be worth making the change to have the average speed shown above the current speed, where the 'na' is in the image below? This would make it a lot easier to see while sailing? Not sure what happens once you complete your hr, perhaps it could 'flash' and keep tracking your average speed? Thoughts?



Once you complete the hour it changes NA to the hour speed, so it appears in large font.

AUS02
TAS, 1981 posts
4 Dec 2017 6:46AM
Thumbs Up

Select to expand quote
Roo said..


AUS02 said..



Roo said..






AUS02 said..







boardsurfr said..










bartontb said..
In MHO, the GPSLogit app interest was killed by the GPSTC decision not to approve the phone solution.













I think that the non-approval "killed" GPSLogit is quite an overstatement. The fraction of speed surfer who want to hear their speeds while sailing is pretty darn small. I'm in that minority sometimes, but in most sessions, I don't want anything in or on my ears. Many windsurfers find convenience and waterproofing much more important. Several guys in our team used non-approved watches, and simply did not post, rather than bothering with waterproof armbands, before the GW-60 was released.

I simply think more choice is better. Some guys still like the Canmore best, despite its well-known shortcomings. Developing and testing prototypes is good fun, but if you can't buy the things in the end, it's rather pointless. A bit like having a GPSTC team where there's just one guy posting all the time.

As for why there are about 10-fold fewer users for GPSLogit than for Windsport Tracker, maybe it's differences in functionality. Or maybe it because of the difference in the text on the little green button:












Can either of these Apps calculate average speed while doing a 1hr (and if not could they)? You'd no doubt need a 'start' button to signify the start of the one hour and then it would track your average speed as you go and count down the hour or elapsed time. I think a reset button would be needed, so if you decide to restart the 1hr you can while sailing by just pushing the reset button.








GPSLogit can. You just set it to that in settings and it reads out on the screen. It also tells you via earphones once you have sailed for an hour and then drop below your speed threshold. I use it when I'm doing hour attempts.







Thanks Roo. Just tried it driving around in the car and it does show your speed (currently 55.05 knots), once above the set minimum, and also your average speed since starting (15.16 knots), elapsed time (00:54:39) and distance travelled (25.57 km), which is all what I was after, but the average speed is only in small font until you complete the 1hr. I wonder whether it would be worth making the change to have the average speed shown above the current speed, where the 'na' is in the image below? This would make it a lot easier to see while sailing? Not sure what happens once you complete your hr, perhaps it could 'flash' and keep tracking your average speed? Thoughts?





Once you complete the hour it changes NA to the hour speed, so it appears in large font.



Roo, yes, it did, and that was good (see below 17.02), but it would be better if you could see your 1hr average speed as your going, in the larger font. Interestingly, the average speed continued, and started to climb further (17.07), as I was still driving at the time.


Boardsurfr indicates that the phone can't calculate the hour as it is going along, but it does appear to be. Any stopped time while doing the 1hr saw the average speed dropping, in-line with how I thought it would. Accuracy of the 1hr using the GPS Logit isn't so important, as I think it would be just a guide while sailing, not for posting. I'm sure the software can then still select the best 1hr from the whole session for actual posting.

Sailquik, every time the travelling speed dropped below the minimum speed (of 20 knots in this instance) the display showed 'na', but the average speed still seemed to be working as expected and didn't appear to reset.

At this stage, it looks as though as having the average speed shown in a larger font is all that required to be able to see this better while actually sailing. This would just be an added feature, as you can already look at the GW60 (or other device) to see what speed your travelling at, but this would be giving you your average speed for a 1hr as well, with an easy reset if the wind drops for a while and then comes back in again and you're ready to start again!

Roo
765 posts
4 Dec 2017 4:01AM
Thumbs Up

Additionally the speedtalker will read out your 1 hour speed to you once you have sailed for an hour so no need to look at the display.

AUS02
TAS, 1981 posts
4 Dec 2017 8:13AM
Thumbs Up

It's pretty handy to see your average speed as you're sailing, so you know whether it's worth continuing after a bit of a lull in speed. Often you think you're going fast with a few fast runs, but then at the end of it all, you end up with an unexpectedly slow 20knot 1hr, due to all the slower parts of the hour, which you conveniently forget to notice!
Having the average shown as you're sailing would give a true indication of 1hr performance during the run. If that small font is showing average speed, just wondering if it can be made larger so visible while sailing. If not, that's fine, it will still be useful.

boardsurfr
WA, 2202 posts
4 Dec 2017 7:13AM
Thumbs Up

Select to expand quote


AUS02 said..
Boardsurfr indicates that the phone can't calculate the hour as it is going along, but it does appear to be.




No, I was talking about the watch, not the phone. I outlined a rather simple way how 1-hour speeds could be calculated on the GW-60, even given the very limited memory it has for calculations. A similar approach would work for the nautical mile.

John340
QLD, 3045 posts
4 Dec 2017 10:37AM
Thumbs Up

Select to expand quote
boardsurfr said..




AUS02 said..
Boardsurfr indicates that the phone can't calculate the hour as it is going along, but it does appear to be.





No, I was talking about the watch, not the phone. I outlined a rather simple way how 1-hour speeds could be calculated on the GW-60, even given the very limited memory it has for calculations. A similar approach would work for the nautical mile.


Peter,
Send your suggestions to Locosys. If the firmware changes are easy they may include in an a future upgrade.

boardsurfr
WA, 2202 posts
4 Dec 2017 10:10AM
Thumbs Up

Select to expand quote
John340 said..
Send your suggestions to Locosys. If the firmware changes are easy they may include in an a future upgrade.


Sorry but I will not. My interactions with Locosys have been far less than pleasant. For example, it took several months just to get them to say that a broken screen repair would cost about as much as buying a new watch. The watch I eventually received as a warranty replacement for another broken watch (this one had stopped charging) is also partially broken (no screen backlight, GPS button very hard to press). Nor am I impressed with their implementation of the alpha algorithm, which works for me maybe one out of 10 times, even with speed genie set below my minimum jibe speed (not that setting speed genie that low makes any sense).

I rather spend my energy looking into alternatives than telling Locosys that their "it cannot be done" statement is wrong. Anyone who's taken an "Algorithms 101" class should have been able to come up with the an workaround like the one I outlined. Lots of guys figured out how to get alpha 500 numbers on a Suuonto watch, which imposes much stricter resource limitations.

Fortunately, the tests with the u-blox USB dongle are very encouraging. I did a driving test today with one dongle and 2 GW-60s, and the initial analysis is that the dongle performed better than the watches in this test. Not because it recorded at 10 Hz (which it did, and which worked fine), but because it had fewer missing data points that either of the two watches. Looking at 5x10 numbers (allowing multiple per leg since I was driving), many numbers were very close (within less than 0.1 knots). When the numbers differed more, either one or both of the GW-60s had missing data points, or a GW-60 with a high +- number in GPSResults was the outlier (due to raw SDoPs around 1.5). This will require more number crunching and more tests, but I am very encouraged from what I have seen so far. Sailquik and Roo have said similar positive things about the u-blox GPS chips - maybe they are just better than whatever is in the GW-60s (Sirf 4s?). I've already ordered more dongles and another phone for more side-by-side tests. This time I found a Kyocera Hydro for $43 on Amazon - with IP57, it might even survive if a bit of water gets into the bag.

decrepit
WA, 11828 posts
4 Dec 2017 12:23PM
Thumbs Up

Select to expand quote
boardsurfr said..
>>>>Sailquik and Roo have said similar positive things about the u-blox GPS chips - maybe they are just better than whatever is in the GW-60s (Sirf 4s?). I've already ordered more dongles and another phone for more side-by-side tests. This time I found a Kyocera Hydro for $43 on Amazon - with IP57, it might even survive if a bit of water gets into the bag.


I've had no missing pts with my GW60s. My driving tests, were with them mounted on the roof.
If you have them inside the car, the difference could be due to a better antenna in the dongle rather than the chip?

But a very encouraging result, I'm a bit tempted to join you.

sailquik
VIC, 6068 posts
4 Dec 2017 5:09PM
Thumbs Up

Select to expand quote
decrepit said..
I've had no missing pts with my GW60s. My driving tests, were with them mounted on the roof.
If you have them inside the car, the difference could be due to a better antenna in the dongle rather than the chip?

But a very encouraging result, I'm a bit tempted to join you.


Yes, same here. Absolutely no missing points at all in all my GW-60 tests and usage.

Still looking for a cheap Android OTG phone to try the USB Ublox7 dongle on. :-)

watermonster
18 posts
4 Dec 2017 4:09PM
Thumbs Up

Over here (Holland) someone is making his own device forum.windsurfing.nl/viewtopic.php?f=62&t=13890429 meanwhile it's approved to use on gps-speedurfing.com and maybe interesting to get some ideas.

sailquik
VIC, 6068 posts
4 Dec 2017 8:31PM
Thumbs Up

Select to expand quote
watermonster said..
Over here (Holland) someone is making his own device forum.windsurfing.nl/viewtopic.php?f=62&t=13890429 meanwhile it's approved to use on gps-speedurfing.com and maybe interesting to get some ideas.


Already mentioned on the previous page.



boardsurfr
WA, 2202 posts
4 Dec 2017 10:33PM
Thumbs Up

Select to expand quote
sailquik said..

decrepit said..
I've had no missing pts with my GW60s. My driving tests, were with them mounted on the roof.
If you have them inside the car, the difference could be due to a better antenna in the dongle rather than the chip?

But a very encouraging result, I'm a bit tempted to join you.



Yes, same here. Absolutely no missing points at all in all my GW-60 tests and usage.

Still looking for a cheap Android OTG phone to try the USB Ublox7 dongle on. :-)


It's easy to not see that points are missing, unless you specifically look for missing points. The GW-60 often drops points in crashes. It very easy to see in the "Trackpoints table" in GW-60 if you sort by time. Here is an example from one of Andrew's tracks from ka72 (his first GW-60 track - the GW-52 track has even more missing points):




boardsurfr
WA, 2202 posts
4 Dec 2017 11:00PM
Thumbs Up

It's rare to see missing points in the middle of a run, but I have seen them on occasion. Usually it's just one point, and GPSAR will still use the section (GPSResults will not use it for 2 an 10 second, maybe more, with default settings).

For driving, I had the GPS units on my dashboard in my Nissan NV2500 high roof (not that easy to mount stuff to the roof, and it's cold here). That means some GPS satellites towards the back may have been blocked. The GW-60 produced data most of the time, but also dropped data points a bunch of times. I would have expected the dongle - phone combo to drop more points since it recorded at 10 Hz, but it ended up dropping way fewer points (and always just single points).

I think this is just another example of the watch handling bad reception poorly. We've seen similar things with underhand grips (although that usually just gives lower accuracy, not dropped points) and with surfing on a mountain lake, where nearby peaks obscured satellites.

It impossible to say if the problems are related to the GPS chip, the small antenna, or other design aspects. But examples where the watch produces crazy speeds when just re-acquiring the signal after crashes are more common, which points towards the chip (or firmware), not the antenna. Also, the antenna in the USB dongle cannot be much larger than the one in the watch, and so far seems to be ok.

boardsurfr
WA, 2202 posts
4 Dec 2017 11:12PM
Thumbs Up

While we're talking about GW-60 artifacts, here's another good one from a recent session where I used 2 GW-60s:


Here, the speed from one of the watches just dropped to 0 a bunch of times (the red line). The other wach actually also shows one drop to 0 at the beginning of the first run. For the rest of the session after the section shown, both watches behaved well.

This is fortunately rare, I recall seeing it in 2 or 3 windsurf sessions (out of about 120 with the watch). But it has happened with different watches. I recall someone else reporting a similar problem, too.


yoyo
WA, 1646 posts
5 Dec 2017 12:31AM
Thumbs Up

I found the ublox dongle I got a couple of years ago would often record 8 points out of 10 when set on 10 Hz. Also not having a battery ment I had to reset it to 10Hz each time I started it up from the default 1Hz it would revert to.
It seems to have a bit more filtering than phone internal GPS which I think cuts the noise down. At least the position does not jump around as much.

boardsurfr
WA, 2202 posts
5 Dec 2017 5:25AM
Thumbs Up

Select to expand quote
yoyo said..
I found the ublox dongle I got a couple of years ago would often record 8 points out of 10 when set on 10 Hz. Also not having a battery ment I had to reset it to 10Hz each time I started it up from the default 1Hz it would revert to.
It seems to have a bit more filtering than phone internal GPS which I think cuts the noise down. At least the position does not jump around as much.


yoyo, how exactly did you record from the dongle? A PC running u-center, or ..?

The dongle I have has a capacitor rather than a battery to keep the RAM powered. It will hold the settings for some time (hours?), but reset to defaults after a while. The idea is that whatever app does the recording would set the rate at the start. What rate works best remains to be determined.

sailquik
VIC, 6068 posts
5 Dec 2017 10:37AM
Thumbs Up

OK. To qualify my statement about missed points:

I have had none during runs with the GW-60 when comparing results from it to other GPS that I have been running at the same time.

It is perfectly understandable that there will be occaisional missed points when the watch is submerged or covered during crashes (I rarely have any these days, or at least none that are caught on camera ) or water starts.

We have had a couple of instances like your example of a block of missed points a year or two back, but they were with GT-31 and GW-52. So far no others have been drawn to my attention from the GW-60.

In that example, the same person had that missing block occurring from both the GT-31 and GW-52 at almost the same time, (but not exactly). I think there was another similar instance in the same area from another sailor at around that time. I also had a similar missing block from my GW-52 in the same month, but the other GPS's I was wearing at the time recorded everything. I have not had any similar instances drawn to my attention since. In all cases, the missing blocks were isolated to one part of the session and for the rest there were no gaps. We looked as closely as we could to all the possible causes we could think of, but those issues are still a mystery to us. GPS system error?? Alien signal blocking?? Nefarious secret squirrel government experiments??

If any other users come accross similar missing blocks, please tell Mike and I about it and send the data to us. The more data we have, the more chance we will get an understanding of the loss.

I almost always wear at least two GPS, usually 3 and occasionally 4 or more when testing something, and I run all my tracks through GPS-Results (and often also RealSpeed and GPSAR-Pro). I would definitely notice missing points in runs, unless they were present at the exact same time in all my worn devices, and even then I think there would be a very high chance I would see them.

As reported previously, when testing the Ublox M8 BT dongles at 18Hz with my cheap Alcatel phones, I was getting regular missed points, (often almost exactly one every second). I put this down to a processing or communication overload or limitations in the cheap phones. Manfred and Roo don't report this problem using more expensive, higher spec phones. My solution was to revert to 10Hz recording where there are no missed points. I suspect this is a better setting anyhow as the GPS can use 2 GNSS (Gps and Glonass) in this mode.

There is no doubt that satellite reception with the GW-60 is not as consistently strong as with the GT-31 and GW-52 due to it's smaller antenna and the way it is normally worn. Mike and I have done extensive comparisons between the wrist worn GW-60 and head mounted GW-52 (and GT-31. Ublox) and the GW-60 pretty consistently has higher error estimates, especially when used underhand grip. It is usually quite easy to see the loss due to underhand orientation in the data, as this invariably occurs on alternate tacks. Fortunately, the speed average differences between the head mounted GPS and the wrist mounted GPS are usually acceptably small. But there is no doubt that wearing the GPS on the wrist windsurfing, especially with such a small antenna, is not tecnically the ideal solution. It is just the trade off we are currently stuck with while most sailors prefer this form for convenience, and I must say I have been surprised at how well it does most of the time.

I ran the USB dongle on my laptops at 10Hz using U-centre, and I didn't notice any regular missing points when I did it outdoors with good reception. But I did not do exhaustive testing since I could not connect it to any of the Android phones I had.

yoyo
WA, 1646 posts
5 Dec 2017 10:23AM
Thumbs Up

Select to expand quote
boardsurfr said..

yoyo said..
I found the ublox dongle I got a couple of years ago would often record 8 points out of 10 when set on 10 Hz. Also not having a battery ment I had to reset it to 10Hz each time I started it up from the default 1Hz it would revert to.
It seems to have a bit more filtering than phone internal GPS which I think cuts the noise down. At least the position does not jump around as much.



yoyo, how exactly did you record from the dongle? A PC running u-center, or ..?

The dongle I have has a capacitor rather than a battery to keep the RAM powered. It will hold the settings for some time (hours?), but reset to defaults after a while. The idea is that whatever app does the recording would set the rate at the start. What rate works best remains to be determined.


I had quite a few different GPS analyzing programs on my computer back then. May have been one of Ublox more detailed setup ones. The current Ucentre for Android is very basic.
Computer was fast but old, connection long and Ublox dongle cheap Chinese one ($10 I think). Newer stuff may not have same problem with update rate. Interesting the dropped points did not affect Results but did ActionReplay ( or was it the other way around?)

I now have nice Android with Nougat and OTG but still prefer to use the old $20 Alcatel with Logit and internal GPS.
Good enough for me. At end of session I just press the screen and it is all there.

boardsurfr
WA, 2202 posts
6 Dec 2017 1:43AM
Thumbs Up

Select to expand quote


yoyo said..


Interesting the dropped points did not affect Results but did ActionReplay ( or was it the other way around?)


Other way around. GPSResults is stricter and applies filters by default. ActionReplay has filtering as a separate step.

I agree that the phone is perfectly fine for casual sailing and for speed feedback. Just for competition, the variation between devices is too large, and there are just too many different phones. Quite often, manufacturers even change the internals of a phone from year to year, but keep the name the same (or almost the same).

boardsurfr
WA, 2202 posts
6 Dec 2017 2:03AM
Thumbs Up

Select to expand quote
sailquik said..
OK. To qualify my statement about missed points:

I have had none during runs with the GW-60 when comparing results from it to other GPS that I have been running at the same time.

It is perfectly understandable that there will be occaisional missed points when the watch is submerged or covered during crashes (I rarely have any these days, or at least none that are caught on camera ) or water starts.

We have had a couple of instances like your example of a block of missed points a year or two back, but they were with GT-31 and GW-52. So far no others have been drawn to my attention from the GW-60.

In that example, the same person had that missing block occurring from both the GT-31 and GW-52 at almost the same time, (but not exactly). I think there was another similar instance in the same area from another sailor at around that time. I also had a similar missing block from my GW-52 in the same month, but the other GPS's I was wearing at the time recorded everything. I have not had any similar instances drawn to my attention since. In all cases, the missing blocks were isolated to one part of the session and for the rest there were no gaps. We looked as closely as we could to all the possible causes we could think of, but those issues are still a mystery to us. GPS system error?? Alien signal blocking?? Nefarious secret squirrel government experiments??

If any other users come accross similar missing blocks, please tell Mike and I about it and send the data to us. The more data we have, the more chance we will get an understanding of the loss.

I almost always wear at least two GPS, usually 3 and occasionally 4 or more when testing something, and I run all my tracks through GPS-Results (and often also RealSpeed and GPSAR-Pro). I would definitely notice missing points in runs, unless they were present at the exact same time in all my worn devices, and even then I think there would be a very high chance I would see them.

As reported previously, when testing the Ublox M8 BT dongles at 18Hz with my cheap Alcatel phones, I was getting regular missed points, (often almost exactly one every second). I put this down to a processing or communication overload or limitations in the cheap phones. Manfred and Roo don't report this problem using more expensive, higher spec phones. My solution was to revert to 10Hz recording where there are no missed points. I suspect this is a better setting anyhow as the GPS can use 2 GNSS (Gps and Glonass) in this mode.

There is no doubt that satellite reception with the GW-60 is not as consistently strong as with the GT-31 and GW-52 due to it's smaller antenna and the way it is normally worn. Mike and I have done extensive comparisons between the wrist worn GW-60 and head mounted GW-52 (and GT-31. Ublox) and the GW-60 pretty consistently has higher error estimates, especially when used underhand grip. It is usually quite easy to see the loss due to underhand orientation in the data, as this invariably occurs on alternate tacks. Fortunately, the speed average differences between the head mounted GPS and the wrist mounted GPS are usually acceptably small. But there is no doubt that wearing the GPS on the wrist windsurfing, especially with such a small antenna, is not tecnically the ideal solution. It is just the trade off we are currently stuck with while most sailors prefer this form for convenience, and I must say I have been surprised at how well it does most of the time.

I ran the USB dongle on my laptops at 10Hz using U-centre, and I didn't notice any regular missing points when I did it outdoors with good reception. But I did not do exhaustive testing since I could not connect it to any of the Android phones I had.





Andrew, I posted your data as a little warning about definite statements like "I never had". Disproving negatives is far easier than proving them.

But I agree that missing data points in GW-60 windsurfing speed runs are at least rare. Even if you'd have a few dropouts in an hour, the chance that they appear in one of the top 5 10-second runs is very small. Actual chances are even smaller since dropouts are linked to poor reception, and reception does not change much in a short speed run. However, I have seen at least one example of a dropout in the middle of a jibe. Since we are moving hands, body, and rig in a jibe, and the GPS may switch which satellites it is using since the body will affect the signal differently in the turn, the chances of a drop in the middle of the jibe are much higher. Just like the chances that the sailor will drop.

Perhaps it would make sense to discuss how missing points should be handled. Currently, GPSResults and GPSAction Replay treat them very differently (except for 1 hour) - Action Replay will still use regions with missing points, while Results will not. With higher-herz data, the potential impact of a missing point becomes less and less. If a point is missing at 10 Hz, we still have 9-fold more data then at 1 Hz! Interpolating single missing points seems more reasonable than excluding the region from analysis. Maybe not for 2 second speed, but definitely for 1 hour and nautical mile, and I think also for alpha 500. A nautical mile at 5 Hz has 500-800 data points at 5 Hz; not using a region because one point in the middle is missing makes no sense.

Te Hau
479 posts
6 Dec 2017 3:21AM
Thumbs Up

"A nautical mile at 5 Hz has 500-800 data points at 5 Hz; not using a region because one point in the middle is missing makes no sense"........
This is a common problem with GPSSS.
Often, the mile or alpha are posted as a zero.
On checking it's always a dropped data point.
Load onto KA72 and problem solved.

raymondw
47 posts
6 Dec 2017 6:04AM
Thumbs Up

A dropped point is mostly an issue with the GPS chip or the device.
Not the signal itself.
I've tested the whole year with 4 to 6 people on my homebuild Gyro devices.
Gigabytes(!!) of data where analyst with GPS-Results, custom program from Manfred and a homebuild python script.
If you stay within HW limits you will not loose a data point.

And I've also send an email to Locosys to help them or (re)program my GW.
No response at all...
Fiddled around a bit and created a semi-custom GW FW, but not a game changer...

boardsurfr
WA, 2202 posts
6 Dec 2017 11:59AM
Thumbs Up

Select to expand quote
raymondw said..
A dropped point is mostly an issue with the GPS chip or the device.
Not the signal itself.


Perhaps, except that the GW-60 (and the 52) will drop data points if the signal is bad enough. I've seen dropped data points in driving tests in just one of two watches, and sometimes in both watches at the same time. The watches are typically about 4-6 cm from each other, so I would not expect a lot of differences. Just tilting one watch 60 degrees relative to the other can make a big difference in SDoP. In a tests on a ferry, I saw great SDoPs, very smooth data, and very high correlation when I was out in the open. In a partially covered section (free view of ~ half of the sky), SDoPs went way up, and there were way more little spikes, which were different for the two watches (random noise).

You're using a u-blox chip and a micro controller in your device, right? How is the plan to manufacture and sell it coming along?

yoyo
WA, 1646 posts
6 Dec 2017 4:02PM
Thumbs Up

From boardsurfr blog "I started with some public domain software for USB communication on Android, and quickly got it to read the dongle data. I used the u-center software on Windows to tell the dongle that is should send out data in the format that includes the speed accuracy data (using "ubx" instead of "NMEA" messages); while there, I also told it to send data at a higher rate (5 Hz like the GW-60, instead of the default 1 Hz). The hacked app simply reads the data, shows them on the screen to show it's busy, and write them to a "ubx" file."

where did you find the apk to read ubx?

GNSS uses NMEA

raymondw
47 posts
6 Dec 2017 5:56PM
Thumbs Up

Select to expand quote
boardsurfr said..

You're using a u-blox chip and a micro controller in your device, right? How is the plan to manufacture and sell it coming along?


Everyone can buy a Locosys.
I build to order for the real speedfreaks ;)
My device is used in an electric rally car and I'm busy with a LeMans prototype (the car!)

Loosing data or having an invalid data point are 2 different things.
I accept an invalid point, but will not accept a vanished point...

In my device every vanished data point was due to poor programming and my understanding of electronics.
All others where just invalid due to poor reception (weather, water etc)

sailquik
VIC, 6068 posts
7 Dec 2017 12:11AM
Thumbs Up

Select to expand quote
raymondw said..
I build to order for the real speedfreaks ;)

Does this mean you are now accepting orders Raymond?



Subscribe
Reply

Forums > Windsurfing   Gps and Speed talk


"Hope for a GPS alternative" started by boardsurfr