Sound like a loose wire in your phone. My old iPhone had similar problems with the GPS, and later with WiFi.

The three dongles I got all work fine (no real surprise if 9 out of 10 work). So do the two phones. Everything together still cost less than one GW-60. But I've had only test drives and a few windsurf sessions so far. In today's session, the dongle numbers where closer to one GW-60 than the numbers from the second GW-60 which was on the other arm. In yesterday's session that included longer swims, the dongle was prone to speed artifacts during the swims, but all these had very high SDoP values, so they can easily be filtered out. May be a while until I get another chance for windsurf tests, the forecast looks bad. We probably jinxed it when we did not bring the light wind boards on our trip for the first time in 3 years...

Still no wind here, so more driving around. The dongle still compared very well with the GW-60, typically giving numbers in the speed categories that are within 0.1 knot of the GW-60. In once case where the differences were larger, the GW-60 appeared to be at fault - another example that it does not handle poor GPS signals that well. Numbers and details at boardsurfr.blogspot.com/2018/01/speed-deltas.html

Very interesting test. From what I have read patch antennas give the best antenna reception for your buck when flat mounted. Bigger the better, though even small ones are way better than watch antennas. The vk172 has a middling 18x18x2mm antenna. I had thought of putting one on the nose of my board with a 60mm ground plane for another 3 dB gain but that may be a bit of overkill given it will have a perfect Sky view when sailing. However I need to get another OTG phone first. Office Works wants almost 200 for a Moto E4 in Oz so that not an option.
Do you have trouble with having the VK172 holding the 5hz setting set on Windows with UCentre? Or have you written an Android program to change the default update rate?

The dongle keeps whatever settings I set in ucenter for some time. Not sure exactly how long, something between 15 minutes and 4 hours, however long the capacitor holds enough voltage. I wrote a little Android program to log to a UBX file that sets the rate and messages. Send me a message if you want to use it.

Data analysis with GPS Action Replay and GPSResults is a bit complicated, since both have bugs. GPSAR does not read the accuracy correctly. Both programs have problems with the time. GPSResults is off by the number of leap seconds. GPSAR sometimes has the same problem, but at other times, it's off by a few days.

Finally got a few windsurf sessions with the dongle, and the results were disappointing. In today's session, it had my top speed at 35 knots, about 3 knots too high. 2 second speed was about 1.5 knots too high. In a previous session, 2 sec was too high by 0.5 knots. SDoP values in the problem areas were around 1.5 - high enough to indicate problems, but below the default filter settings in GPSResults (and probably also ka72.com). This indicates poor reception, quite possibly because the dongle was lying on the phone, and the bag sometimes slipped on my arm.

Considering these results, the fact that phones with the required OTG support appear to be expensive outside the USA, and the fact that phones get ever-larger and barely fit into the bags anymore, I see no point in any further evaluation. A bluetooth dongle that would allow the dongle to be separated from the phone would possibly give better results, but no cheap u-blox (or Locosys) dongle is currently available. Wearing the phone on top of a helmet might also work well enough, but I rarely wear a helmet.

GPSLogit with the Moto E4 had performed very well in most of my tests up until today, with numbers typically being within 0.1 knots of the GW-60. But in today's results, the phone was high by 0.25 knots for 2 seconds, and low by 0.41 knots for the nautical mile. The tracks show multiple sections with multiple artifacts in the GPSLogit data (drops by 1-3 knots for 1-2 seconds). In one region, the phone is almost 4 knots lower than the GW-60 over 4 seconds, with the phone tracks showing obvious artifacts. I may post pictures and data on my blog in the next couple of days. These results illustrate that the phone GPS is not accurate enough for the competition, even though it can give very accurate results in some tests.

Elmo was using gpslogit with his phone on Thursday, it gave him a 37knot Nautical mile, he was ecstatic. But his GT11 only had a very low 36Kt NM in there. Still a PB for him so not all that disappointed, but the discrepancy surprised me.

My reading indicated whist a patch antenna had better reception than a chip , the chip was not affected by oreintation so much. I guess that is what you found in practice.

There where some tests with a additional chip antenna in a GT-31, the results where also not oke.
I use quite a large patch antenna in my device as the smaller ones gave a poor result.
Tested sizes from 30x30x2 to 60x60x5 mm

probably more than you even wanted to now about GPS antenna here: avrproject.ru/EB-500/GPS_Antennas_ApplicationNote-GPS-X-08014-.pdf

Select to expand quote

decrepit said..
Elmo was using gpslogit with his phone on Thursday, it gave him a 37knot Nautical mile, he was ecstatic. But his GT11 only had a very low 36Kt NM in there. Still a PB for him so not all that disappointed, but the discrepancy surprised me.

Yes I can confirm the variances between GPSLogit screen(Samsung galaxy J1) results and GT11& Reelspeed results are quite a bit.

I don't faff around with the phone results on realspeed.

If anyone's got a recommendation on a better phone or a way to improve it, I would greatly appreciate it.

Select to expand quote

elmo said..
If anyone's got a recommendation on a better phone or a way to improve it, I would greatly appreciate it.

I have a recommendaton... wait a few years -> phones will eventually get more accurate.
Whether it ever produces error-data is a separate discussion.

Interesting find. It's also on ebay for less than $70 from China, or $100 from the US. With IP 54 or 65, it will still need a bag, but accidental exposure is less of an issue. It's much smaller than current phones, so using it with a helmet mount may work much better. The screen is too small for GPSLogit, but big enough to show a few numbers.

If the VK dongle does not work out, perhaps slightly larger "GPS mouse" units would work better. Ublox 8 devices are available for $7.40 (www.aliexpress.com/item/32800761853.html), and even Locosys has some listed on their web site.

Thanks for all the feedback after my last frustrated post. It made me realize that the issue was most likely from the bag slipping on my armband, which led to the GPS signal being partially obscured by my arm and body. I used a bit of velcro to keep the bag always facing upwards, and got much better results in my last two windsurfing sessions. GPSLogit can be quite accurate, and the USB dongle does again look like a viable option that provides accuracy estimates, as long as the phone stays facing upward. Details at boardsurfr.blogspot.com/2018/02/gps-dongle-test-results.html

Select to expand quote

elmo said..

decrepit said..
Elmo was using gpslogit with his phone on Thursday, it gave him a 37knot Nautical mile, he was ecstatic. But his GT11 only had a very low 36Kt NM in there. Still a PB for him so not all that disappointed, but the discrepancy surprised me.

Yes I can confirm the variances between GPSLogit screen(Samsung galaxy J1) results and GT11& Reelspeed results are quite a bit.

I don't faff around with the phone results on realspeed.

If anyone's got a recommendation on a better phone or a way to improve it, I would greatly appreciate it.

In my recent tests, I have often seen that strict filtering in GPSResults can lead to artificially low numbers. One example with 0.9 knot lower speeds is described in detail at boardsurfr.blogspot.com/2018/02/the-disappointment.html - in this case, the GW-60 watch had dropped a whole bunch of data points, which meant GPSResults ignored the region. When using GPS Action Replay, I got higher numbers that were virtually identical with the numbers from GPSLogit (and the USB GPS dongle I'm testing).

That said, I have seen regions in GPSLogit data where the speed was about 1 knot lower over 1 km simple because the phone slipped to the under side of my arm, so definitely make sure that the phone in close to your shoulder, and pointing up. If it is, I'd expect nautical mile results to be within 0.1 knots, and typically within 0.03 knots, with a decent phone compared to the GW-60. I have no experience with the GT11 or Realspeed, though. I have use the Galaxy J1 a bunch of times and never noticed any problems, but I have not looked closely for them, either.

Cheap Android phones with GPS-Logit can give some pretty good results. But we have got some shockers as well.
GPS-Logit with a ublox M8 BT dongle is extremely good, and I have seen no errors in the results, or indeed any variation from good GW** or GT31 results. (at 10Hz, dual GNSS)

My only complaint with the cheap phones is limited battery life, but it normally gets me through a typical sub 2hr session.

Select to expand quote

sailquik said..
My only complaint with the cheap phones is limited battery life, but it normally gets me through a typical sub 2hr session.

With the Moto E4, I get at least 4-5 hours, with both GPSLogit announcing speed over bluetooth and my little dongle logger app running. I need longer sessions to see what the limit really is. Regular price here in the US is around $100 ($130 unlocked), but it's often on sale for $40-$60. Battery is 2800 mA, compared to ~1800 for the Galaxy J1 that has a shorter battery life.

On battery life. My 3000 mAhr battery Chinese otg phone seemed to have less battery life than my old Alcatel pixi 3(4) with 1400mAhr which always had enough juice for my shortish sessions. Difference was the newer screen had twice the dpi (HD) and OS was Nougat 7.0 which seemed to use twice the resources of KitKat 4.2.2 especially after I took all the Google crap off the Alcatel.
Simpler the better for our needs.

I dont think that Moto4 phone is available in Australia, and if it was/is, it would probably be quadruple that price.

Select to expand quote

sailquik said..
I dont think that Moto4 phone is available in Australia, and if it was/is, it would probably be quadruple that price.

Maybe I'll take orders towards the end of the year and bring a couple of extra phones when we come over .

Select to expand quote

yoyo said..
On battery life. My 3000 mAhr battery Chinese otg phone seemed to have less battery life than my old Alcatel pixi 3(4) with 1400mAhr which always had enough juice for my shortish sessions. Difference was the newer screen had twice the dpi (HD) and OS was Nougat 7.0 which seemed to use twice the resources of KitKat 4.2.2 especially after I took all the Google crap off the Alcatel.
Simpler the better for our needs.

I was surprised to see the decent battery life on the Moto E4 - it has a larger, higher resolution screen than my other phones (Galaxy J1 & J3), and runs a newer OS (Android 7.1). I never activate the Android phones I get, and keep them in airplane mode, which may help a bit with the battery life, but that's true for all phones. To some extend, the battery life may depend on the junk ware that's installed on the phones (and often can't be removed without causing problems). Sometimes, new OS versions cause older software to use more resources, and the junk ware may not be updated. It's also possible that 7.1 fixed some of the issues introduced in 7.0, which had plenty of reports about battery life issues.

some phones kill the gps when put on airplane mode. not sure why as gps signal is passive.

I'd say most or all of them turn off GPS, but it usually can be turned back on separately, same as WiFi.

With the Alcatel OneTouch Pixi First 4024X (android 4.4) in flight mode, GPSlogit works using the Pixi's internal GPS.

Turning off google voice activation (not GPSlogit speech) improves battery life.
Having your phone connected to a network in a remote location with poor reception will force the phone to operate at its maximum RF power and use extra power.
Having apps running that you are not using will use extra battery power.
Having the screen on 100% for GPSlogit will use the most battery over anything else, so I turn this off and turn phone on to check results if/when required.

It has been a very poor season (no wind) so I don't know how long I would get from my Pixi. My sessions this season have been less than 2 hours.

Useful tips, srtgumbee. Quite possible that different phone companies have different apps running by default, which would explain some variation in observed battery life. I had a 3-hour session yesterday, and the Moto E4 reported that the battery was still at 60% when I got home. The screen set to automatically adjust, which tends to be quite bright in the sun, but usually not 100%.

I'm not android developer and don't know what features apps have access to when the home screen is locked. But I always thought a useful feature (that could be turned on/off) for GPS-LogIt would be to use the phone's accelerometer to sense a movement not normally used in windsurfing and this movement would turn the screen on and/or toggle viewing modes in the app. OR just turn the screen on when going below (or above?) a speed threshold.

Assuming the phone is mounted on the arm, then arm up in the air 'fist pump' or arm down by your side for a period of time could be the trigger. I find the screen hard to see in the sun light when windsurfing, so I jump off to check results if needed. In my case, I would be happy if the screen would come on if I'm going less than 5knots.
For those that don't use the voice feature, I could see a benefit to having the screen come on when over a set speed threshold or session PB too.

the battery shouldn't be a problem.
I tested several devices and usually it's no problem to record 8h and more.
with some devices 24h are possible.

I would not recommend cheap china phones. they might work sure, but chance that they don't is high. As example a user reported that his phones kills all apps as far as the screen gets off. The company though that this would be a good way to save battery live, but they forgot about that some apps like logging /music apps need to run in background.

sure the movement detection is possible, but not easy to do. Acceleration sensors works in 3 dimensions. Even everyone wears the phone on the shoulder, it doesn't say that it might move to the right or bottom etc.
also when you do waterstarts you hold your arm up, if display gets on, in worse case, a water drop can unlock the screen and stop your recording.

so using the hardware button to turn on the screen is usually a similar move as move the arm up,

and for easier control and feedback I already have an idea, that will work, It won't be finished the next months but hopefully for next Australia season .

I have now tested the USB dongle in about 15 windsurf sessions. The observed accuracy for the six GPSTC disciplines was generally within about 0.1 knots of the GW-60 results (0.2 knots for 2 seconds). This is comparable to the differences observed between two GW-60s, one on each hand, or a GW-60 and a GW-52. Only one session was an outlier with a 0.4 knot deviation, caused by the arm band slipping (as described here before). More details with a graph and results from 10 sessions (including links to the data files) are at boardsurfr.blogspot.com/2018/03/dongles-beats-phone.html

One thing that came out of the testing is that the USB dongle is less sensitive than the phone GPS (with GPSLogit or Windsport Tracker) about the arm band slipping so that it is on the side of the arm. The phone GPS does not like that at all, and tends to give errors of 0.5 to 1 knot in one or more disciplines when that happens. The dongle still seems to be reasonably accurate even on the side of the arm. More about that (and the result summary for 2 more test sessions) at boardsurfr.blogspot.com/2018/03/the-difference-between-good-and-bad.html

Since the USB dongle has accuracy comparable to the GW-60 and provides accuracy estimates, it may be worth considering to have it as an "officially approved" device for GPSTC. That would require some polishing of the logger app, which I'd be happy to do. It may also be possible to convince Matthias to add the dongle logging to the Windsport Tracker app. Either of these approaches would requires some signs of interest, though. In the US, the dongle + phone solution is a low-cost alternative to the watch, with a total cost of $60 compared to ca. $250. But there are very few speed surfers in the US. In Australia and Europe, a lot of the cost advantage goes away, since the phones are quite a bit more expensive (and many current phones do not have the necessary OTG support).

It would be nice to get some feedback. If there's no general interest, I'll just pursue an ok to use it for GPSTC postings as a custom ublox-based device.

So I'm interested, but if I can't find a cheap suitable phone, what's the chance of you bringing a couple over when you come.

Select to expand quote

decrepit said..
So I'm interested, but if I can't find a cheap suitable phone, what's the chance of you bringing a couple over when you come.

^+1 on what Decrepit said.

Bringing a couple extras should be possible, but it won't happen until December.

That would theoretically leave time to fool around with a PiZero W + displays a phone alternative. Hardware cost for Pi + display + battery + cable would be around $50 in the US. Since that's not subsidized by phone companies, it should be comparable in Australia - could someone verify that? Highest cost would be for the display. The problem is that is would require a bit of assembly and geek knowledge to put together. Should be easier than making fins, though .

Yep, I bought a Pi Zero and including postage was about $15. AUD. The ublox 8 chip was the most expensive component but even then it was relatively cheap. The hardware seems straightforward and easy to assemble. I just fall over with the coding side of things. :-(