OK, Sean asked for it, so here we go.

For what's gone on before see "the GT31 replacement" thread

Basically Sailquick and I have a couple of test units to try out.

Here's the Pros.
5hz means better accuracy.

When I first looked at the manual, I was seriously considering writing the basic instruction on my board with a texta!
But once I've become familiar with it there's no problem, the buttons are quite logical really, and easy to use inside the pacqua, although they do loose a bit of sensitivity through the plastic. Much better than using the GT31 toggle inside a pouch.
So it's possible to go to the 5 best of page and check your 5x10's on the water, no more wondering if you need one more 10s run to get the average you're after.

It's obviously smaller than the GT and almost half the weight, which means it doesn't slide around on your arm as much. Just feels more comfortable there.
It's simpler to use and setup than the GT, apart from the timing function it's all windsurfing orientated, there is a small display showing your bearings but that's all the navigation there is. So there's no great list of settings to scroll through, there's just 8 settings on one page.


Here's the Cons so far.
I'm not sure if we're just trialling an intermediate firmware, but there's a few things missing at the moment, most important in my mind is a distance read out, then a large font scroll of max speed for the session. There is a best of 5 readout but that's in small font, and if you need glasses, you won't be able to read it. To access the file you need to use a windows machine with their "utility" installed. The one they pointed us to had a "smart logger" option but if you select that, the unit doesn't log anything.

Smart logger sounds like a good idea, it logs at 1hz until the speed genie is triggered then it switches to 5hz for better accuracy, this saves memory, which at 5hz continuous last for just under 8hrs.

---------------------
OK, back to the on going saga.


Had another glitch today, processed my files on the windows laptop, so used realspeed. As seems to be the case the GW gave slightly better figures than the GT except for alphas.
I had two 21s a 20 and two 19s for my top five with the GT31, but only a 15, 13 and 12 for the GW, just those three no more.
So now I'm on the PC and using GPSarPRO, I get very similar results for both units in all divisions.

GT31 2s 31.10, 5x10 30.19, NM 22.46, hr 13.68, alpha 21.32
GW52 2s 31.16, 5x10 30.23, NM 22.49, hr 13.69 alpha 21.36.

The Gt track is much cleaner than the GW



Red is the GW52 yellow the GT31. Interesting there's a small time divergence in the two tracks going into the gybe, but there's a big spatial divergence of 12.59m. Even stranger is what happens coming out of the gybe.



The gap between the two tracks has decreased to 4.63m.
Today I had the GW on my upper right arm and the GT on my right shoulder, maybe the GT had a better satellite view so tomorrow I'll reverse them.

perhaps the extra processing for the 5Hz results in a slight time delay. I think on the Gt 31 you can alter the time of recording. (Tom Chalko got 2 gpses to act as one 2Hz gps by having the recording times 1/2 second out. Perhaps Mike your 31 and the 52 are not recording at the same time..... Perhaps Katy can offer some advice on what is going on.

Select to expand quote

yoyo said..
perhaps the extra processing for the 5Hz results in a slight time delay. I think on the Gt 31 you can alter the time of recording. (Tom Chalko got 2 gpses to act as one 2Hz gps by having the recording times 1/2 second out. Perhaps Mike your 31 and the 52 are not recording at the same time..... Perhaps Katy can offer some advice on what is going on.

I think that's a bit different David, I'm sure Tom's time differential between the 2 units didn't result in a positional offset if so it wouldn't have been much use.
If you record at a different time you should also record a different position. I didn't think it was possible, but if there a difference in the two units clocks, that could result in a positional off set, but why on earth would it vary so much? I'll try it at 1hz when I get time and see what the difference is.

It would be interesting to see if the alphas are still 12m(start) and 5m(exit) apart when run on another program such as gpsvisualizer. That is, maybe the software processes them differently. As you said earlier the Realspeed software only gave 15kn alpha on the GW52 compared to 21 GT31. Whereas GPSAR had them similar.

Perhaps a Canmore, GT31 and GW52 all on 1Hz together would be interesting.

Can someone send me a download link for the software for transferring data off the GW-52? I have a test unit, but no software.

Dylan.

Select to expand quote

Dylan72 said..
Can someone send me a download link for the software for transferring data off the GW-52? I have a test unit, but no software.

Dylan.

Sent by email.

Yoyo, here's what GPSvisulizer makes of them, it's a bit easier to see the pairs of tracks when they are speed coded by colour.

Select to expand quote

yoyo said..
It would be interesting to see if the alphas are still 12m(start) and 5m(exit) apart when run on another program such as gpsvisualizer. That is, maybe the software processes them differently. As you said earlier the Realspeed software only gave 15kn alpha on the GW52 compared to 21 GT31. Whereas GPSAR had them similar.

Perhaps a Canmore, GT31 and GW52 all on 1Hz together would be interesting.

Here's some gybes, interesting all the tracks cross over half way through!


And here's some of the speed run going up and down wind.



How is the top left hand corner green going into red dot, one of them gets hoplessly lost, going right out of the picture then slowly works it's way back.
Seems as a general rule that the error gets less with an increase in speed, ie the dark purple are almost on top of each other.

OK I'll set the GW52 to 1hz for my next session and see if it makes any difference.

Great news, I've just installed a "trial, not for redistribution, firmware". That gives max history, distance and smart logger.
At the moment in large font speed genie now cycles between last run max speed and ns average, then history max speed/max ns average on the same window, then current max and ns average are repeated then distance is displayed. All this takes only 6 seconds, which is too fast for this old brain to work out what I'm seeing before it's gone, and that's not in a wet pouch.
But I'm sure there smart guys can slow this down so it takes about 20s to go through, and I think the repeat of current run can be left out.
If you use the up/down buttons to manually scrol, you get, Distance (in big font), best 5 max speeds, best 5 ns average, then coordinates.

Apart from battery life I think we're getting very close to what we want, maybe they can make the font a bit more visible that would help as well.

Hers's some more test tracks part bike part car, it's looking more and more like the faster you go the more accurate the posistion is.

And where it's red and stationary, I'm apparently climbing around on the rooves of several neighbouring houses.

Here's some pics of the speed genie displays.
It's under my desk lamp, unfortunately I've let it washout the top right hand side.

firstly distance.





then history max, note the "medal" alongside the max speed. This is displayed every time max speed increases, so you know you've improved the session score


Last run max speed.


Last run Ns max average

Good testing Mike.

Perhaps you are looking at Doppler Positioning rather than Trackpoint positioning? Can you switch off Doppler and map it?

Select to expand quote

sailquik said..
Good testing Mike.

Perhaps you are looking at Doppler Positioning rather than Trackpoint positioning? Can you switch off Doppler and map it?

I'm just loading the spd file into gpsvisulizer, I don't think you get a choice.

RealSpeed. Try checking which is being used to draw the tracks. Right click the window and see if there is a way to select the other. (There may not be, I am just brainstorming)

Select to expand quote

sailquik said..
RealSpeed. Try checking which is being used to draw the tracks. Right click the window and see if there is a way to select the other. (There may not be, I am just brainstorming)

OK next time I'm in windoze

Sorry Andrew, tried everything but I can't see any reference to how alpha positioning is drawn.

Overnight battery test gave me 10hrs 24min of data at 1hz continuous logging, the only interesting/worrying thing, it did 4km while sitting still in the window!

Select to expand quote

decrepit said..
Sorry Andrew, tried everything but I can't see any reference to how alpha positioning is drawn.

Overnight battery test gave me 10hrs 24min of data at 1hz continuous logging, the only interesting/worrying thing, it did 4km while sitting still in the window!

No worries, it was just a long shot. Probably not the reason for the observation. A more likely reason (after I thought about it properly ) is that the two GPS are using a different GPS datum model.

4KM in 10 hours is an average of 0.4kph, which is way higher than I think it should be. Did the GPS have a good view of the open sky? With a good sky view the error should be at least an order of magnitude smaller than that.

I will do a side by side geostationary test with a GT-31 for a few hours and report what I find.

Today I took the GT31, GW52 and Fangy's Canmore for a drive. All in the same pouch on my roof rack so they all had the same good sky view.

This has improved the look of the GW tracks, they're much smoother, so does this mean it's less sensetive to satelite signals than the GT31?





This turn doesn't sink below the speed genie setting of 10 knots so there's a nice smooth 5hz track.
The next pic shows what happens when you slow down below the speed genie setting


The blue highlighted line in the tracks table is where it's changed to 1hz, speed is down to 6knots, and it changes back to 5hz at 9.76kts.
You can see the kink in the speed graph at the change point.
And the sudden change to a faceted track.
So if you're using "smart logger" and doing alphas, be sure that speed genie isn't set too high.

And here's another worrying pic
Canmore Blue
GT31 Green
GW52 Red


So how important is this positional inaccuracy?
Any idea anybody?
Guess if you're looking for hidden treasure you'd go for the Canmore of GT31

Here's another strange one. Fangy has converted the .fit file into a GPX, so now I can run all 3 in GPSarpro

Here's the 2sec results.
GW52 34kts, GT31 33.75, Canmore 33.91

BUT, look at this!
Canmore 2sec



GW52 2sec


Is this due to the conversion to GPX or is there a huge time difference between the Canmore and the other two.
My poor old brain is starting to boggle, think I'll have a rest.
Or maybe I'll see how the alphas look first.

Yes same time offset
GW52 25.46, GT31 25.47, Canmore 25.51.
So despite the time difference the results are very similar.

I had another go with Realspeed and the GW52, asking for the alpha 500 result just crashed it, and this file is nice and smooth. I thought realspeed was having trouble with the jagged file from the 13th

Select to expand quote

sailquik said..
>>>> Did the GPS have a good view of the open sky? With a good sky view the error should be at least an order of magnitude smaller than that.

I will do a side by side geostationary test with a GT-31 for a few hours and report what I find.

No, it was inside, because it was stormy last night. I had it close to the window, but there a 700mm steel overhang plus a 2m timber solar pergola on the outside.
That's what I put the distance down to, but I thought it should have some sky view, to make it work, to get a reasonable take on battery life.

Select to expand quote

decrepit said..
Here's another strange one. Fangy has converted the .fit file into a GPX, so now I can run all 3 in GPSarpro

Here's the 2sec results.
GW52 34kts, GT31 33.75, Canmore 33.91

BUT, look at this!
Canmore 2sec



GW52 2sec


Is this due to the conversion to GPX or is there a huge time difference between the Canmore and the other two.
My poor old brain is starting to boggle, think I'll have a rest.
Or maybe I'll see how the alphas look first.

Yes same time offset
GW52 25.46, GT31 25.47, Canmore 25.51.
So despite the time difference the results are very similar.

I had another go with Realspeed and the GW52, asking for the alpha 500 result just crashed it, and this file is nice and smooth. I thought realspeed was having trouble with the jagged file from the 13th

Easy Decrepit, my GPS,just like me, is ahead of its time.

Decrepit asks a very good question: How important is the actual positional accuracy in our speed measurement?

Answer: (Bearing in mind that I am not a GPS technical expert but this is just my understanding from my reading and from conversations with those who know a lot more about it than me).

Not very important, and very important. In fact, it is almost irrelevant in one sense, but positional accuracy in another sense is critical. The Reason this sounds contradictory is simple. It is the relative positional accuracy that counts, not the actual geo location.

If the positional trackpoints are skewed north, south, east or west in a consistent manner, then it is not at all important. What is important is that when you return on your Alpha, the relative positional accuracy has not changed and we can be sure that we are within the 50m circle, or outside it as the case may be.

Critical observers of GPS tracks on maps (like Google Earth) will have noticed long ago that the tracks don't always match the roads. More often than not, they are offset slightly in one direction fairly consistently. If the offset is North for example, you find the track north of where the road goes if you are travelling east-west. You don't really notice any offset if you are travelling north-south.

The most obvious cause of this is the limited accuracy of the GPS system due mainly to atmospheric conditions affecting the GPS signals. Luckily for us, this affect tends to be consistent over the short term and may be likened to a relatively slow 'drift' in position. So in the time frame we do an Alpha, the 'drift' usually does not change enough to be a serious issue for us. (perhaps a few centimetres, but mostly less than a meter).

Another cause of this 'drift' may be the accuracy of the actual satellite position data. Satellites do 'drift' from their predicted orbits over time. This is constantly monitored by the system and updated in the system, but there may be a lag in these corrections that can affect the positional calculations.

Probably the biggest cause of inaccuracy is low numbers of satellites tracked and used in the calculation. The more satellites you track and use, the better the accuracy will be. (The position of the satellites counts also, but that too becomes less critical when you have more tracked). That is only up to the point where you have lots (maybe 8 or more?) of satellites tracked and then the previously mentioned factors become more of a factor. But this is very important for us and explains why we go to such lengths to urge people to mount their GPS in a position and orientation that give it the best view of the most sky, like on the top of your helmet. It also explains why we have had some rubbish accuracy results from GPS's worn in bad orientations like in the back pocket of a harness, or in arm bags that have slipped around to face down and so get reflected signals. It is important to mention here that water is a very good blocker of radio signals, so anything covering the GPS antenna that holds a significant amount of water (like damp Neoprene wetsuit lining material) will significantly degrade the radio signal, and therefore accuracy.

Tom Chalko emphasised to me, that from his testing of the accuracy of the GW-52, giving the antenna a clear view of the sky was critical to maximise it's accuracy potential. ie. Mount it on your helmet!, or at the very least make sure it is in a stable position with the antenna facing the sky and no radio signal blockers over it! (of course, this also applies to all types of GPS).

Back to accuracy. The reason we went to using the Doppler based speed calculations is because they are inherently far more accurate than speed calculations derived from positional data.
The Alpha category is the only one that must use some positional data, and that is only for the calculation of the proximity circle. The speeds are done with Doppler.

It is theoretically possible to use the Doppler data to calculate position and Manfred Fuchs tried this in his GPS-Results program for the Alpha a few years ago. After testing, we had to revert to using the positional data for the reason that any error in the Doppler positional data (calculated from direction of travel) is retained and accumulated. To compound matters, we found the most likely place Doppler directional errors occurred was during the gybe! One small error here can have the effect of the plotted track diverging enough to miss the proximity circle by a significant amount. So we had to revert to using the positional data for the proximity circle, which may not always be as exact as we would like, but at least it does not accumulate errors.

Another possible cause of the difference between the plotted tracks from different GPS's is that they may be using a different datum set on the shape of the earths surface to calculate their results. To keep it simple, (the only way I understand it), the earth is not a perfect sphere. In fact it is far from it. there are various models (Map Datum) that are used to correct/calculate the positional data to compensate for the irregular shape of the Earth. The recommended one for us is called WGS84. If you look in your SETTINGS menu on your GT-31, that is what you should find it set to.

I don't know what Map Datum the Canmore GPS or the GW-52 uses. They may be different, and that could also explain the offset we are seeing between positional tracks. In any case, it should not matter much as the relative positional accuracy is what we need.

There you go. More than you ever thought you wanted to know. If there are any real GPS technical experts out there who know I have got something wrong here, please feel free to correct me.

Following the edges of the path with both GPS on head at walking pace. The GW52 is not even consistent with its error. In bottom corner it is 4-5m NE of true position whilst around the BBQ it is 5-6m ESE even though there is probable only a minutes time difference.




Andrew your confidence in Doppler is perhaps unfounded. Looking at the walking track of the GW52 on RealSpeed when selecting the Grid Field to UNFILTERED Doppler it was just as far out as TrackPoint and Speed. It was possible to see this was associated with a large change in Altitude and usually occurred just after positional change (direction or starting a walk from standing still)

It may be because of the GW52 using Doppler to calculate position and getting accumulated errors as you mentioned.

Select to expand quote

It is theoretically possible to use the Doppler data to calculate position and Manfred Fuchs tried this in his GPS-Results program for the Alpha a few years ago. After testing, we had to revert to using the positional data for the reason that any error in the Doppler positional data (calculated from direction of travel) is retained and accumulated. To compound matters, we found the most likely place Doppler directional errors occurred was during the gybe! One small error here can have the effect of the plotted track diverging enough to miss the proximity circle by a significant amount. So we had to revert to using the positional data for the proximity circle, which may not always be as exact as we would like, but at least it does not accumulate errors.

Tachograph with Speed selected



Tachograph with Unfilter Doppler selected


Doppler speed/Merged Doppler/Doppler 3D selected as well

Select to expand quote

yoyo said..

Following the edges of the path with both GPS on head at walking pace. The GW52 is not even consistent with its error. In bottom corner it is 4-5m NE of true position whilst around the BBQ it is 5-6m ESE even though there is probable only a minutes time difference.

Yoyo you don't mention you did the two walks on seperate days, so there's not going to be an exact match, it just may confuse people if they assume you had both units on your head at the same time.

Select to expand quote

decrepit said..

yoyo said..

Following the edges of the path with both GPS on head at walking pace. The GW52 is not even consistent with its error. In bottom corner it is 4-5m NE of true position whilst around the BBQ it is 5-6m ESE even though there is probable only a minutes time difference.

Yoyo you don't mention you did the two walks on seperate days, so there's not going to be an exact match, it just may confuse people if they assume you had both units on your head at the same time.

i was confused waaaayyy before then Decrep :-)

I like yoyo's house !!!

If only....

We just rig there because it is only 5m to the water....

Select to expand quote

fangman said..

decrepit said..

yoyo said..

Following the edges of the path with both GPS on head at walking pace. The GW52 is not even consistent with its error. In bottom corner it is 4-5m NE of true position whilst around the BBQ it is 5-6m ESE even though there is probable only a minutes time difference.

Yoyo you don't mention you did the two walks on seperate days, so there's not going to be an exact match, it just may confuse people if they assume you had both units on your head at the same time.

i was confused waaaayyy before then Decrep :-)

Here is a video of the speed Genie in action with the updated firmware that Decrepit describes above. Genie was set at 3 knots and I am pleased to see that I can still run at 10 knots!



In case you can't see it, the two low figures are the last run. Then there is the distance. Then a repeat of the last run and then the dual screen with the History max and History max Ave.

Good one Andrew, I've been meaning to do that for the last few days, haven't got around to signing up to a video site.
You see what I mean about the scrolling time.
I've emailed Kate to see if they can increase that to 3s to 5s per view.

I also did a geostationary test today with the GW-52 next to (30cm from) a GT-31. GW-52 set to 1 second logging.

I left them for just over 1hr.

Comparing the Doppler results in RealSpeed:

Apologies for the approximations rather than precise figures. It is more effort than I can muster to actually count the anomalous speed and SDOP readings out of thousands data points.

GW-52:

This GPS was using 9 sats for over 99% of the test.

The Speed display was mostly zero with a display of 0.01 knots every now and then. At the very start there were a few higher readings but after less than 1 minute I did not see any higher than 0.01 knots for the next few minutes.

RealSpeed data shows the Doppler speed varying from 0.000 to 0.019 knots with only a very few readings above that. The maximum was 0.039 knots.

The total accumulated Doppler distance over about 64 minutes was 6.9km. Note that every speed reading will have a small distance added to the total distance but those distances are very random in direction and don't cover much more than a total of a Metre in any direction.

The Trackpoint distance was over 27 KM!

The SDOP error was correspondingly small. It was always under 0.019 after the first couple of minutes.

GT-31:

This GPS was using 8 satellites for most of the test but dropped to 7 a couple of times for a minute or so.

The speed display varied a lot more and was between 0.01 and approx 0.20, with figures between 0.06 and 0.12 common for the time I watched it.

RealSpeed shows Doppler speed varying between 0.019 and 0.058 for a large % of the time, but with significant data up to 0.078 at times. This surprised me because I expected to see it agree closer with the screen display.

The SDOP error was consistently at 0.233 with little variation at all over the hour.

The big surprise was that the accumulated Doppler distance was just over 82 KM! This was actually larger than the accumulated trackpoint distance of just over 56 km. The positional graph showed about double the spread in area of the GX-52 wanderings.

It is difficult for me to quantify the 'wandering' area of the two GPS as RealSpeed only gives coordinates in Degrees and fractions thereof. But I can open both superimposed and see the scale of difference. I will see if I can do a better measurement in GPS-Results.

Thats about all the useful info I can assimilate at the moment. I may well see some more significant things when my head stops hurting from this analysis!

But, on first look it seems to support the GX-52 being significantly more accurate than the GT-31.More tests and time will tell.

Select to expand quote

decrepit said..
Good one Andrew, I've been meaning to do that for the last few days, haven't got around to signing up to a video site.
You see what I mean about the scrolling time.
I've emailed Kate to see if they can increase that to 3s to 5s per view.

Now worries.
The rotation appears to be 1 second. I think even slowing to 2 seconds a screen would make a big difference. I suspect any more than 3 seconds would be a bit too long and tedious.