I did a test today, 10hz DIY V3 on my head, 10hz Motion on my forearm. Results are very close, except for when I flip the sail. It's no where near as bad as the watch on the wrist though.
Red is DIY V3, Green is GW52, Blue is the motion, magenta the GW60


Red and green on my head, blue forearm, magenta wrist. The peak of the magenta is +/- 1 kt the blue peak is +/- 0.5. red and green stay steady on +/- 0.15.

Here's the head upper forearm comparison, from speed reader. Motion on the left V3 n the right
Motion on the left, V3 on the right.


So I think motion on the upper forearm, is very good. The display is easier to see than on the upper arm and more accurate than on the wrist.
It's light and the band has good grip, so there was no sign of it slipping. I'm now confident to leave the watch behind, the motion is easier to see, and gives all our divisions.

Its just a great devise to use , be nice when we can just post to ka72 but still its such a good thing , after using it for long time i get very similar speeds , the fastest devise is still the gw52 than the motion and slowest is usually the gw60 but not much in the results .

Hello, KA72 added .oao upload a month or so ago. It's still in trial I guess.

Yes I have been uploading OAO files directly to KA72 since mid April. It has worked perfectly until the last one which was an 8MB file that it refused to accept. Not sure if it was the size or something else.

I think I remember a KA72 load problem was because of file size. So that's probably it.

With any luck there's an alternative on the horizon. The software I'm using above has a "post to GPSTC" button on it.

Select to expand quote

tbwonder said..
Yes I have been uploading OAO files directly to KA72 since mid April. It has worked perfectly until the last one which was an 8MB file that it refused to accept. Not sure if it was the size or something else.

+1 ^. I have been using Ka72 with .oao and not had any dramas. However I think all my files have been a bit smaller than 8MB!

Sorry didnt realise , yes works a treat

So Decrepit, are you saying that wrist mounted GPS can be +/- 1 whole knot out of say 20kt board speed, so 5%!!!???
Doesn't this defeat the argument about GPS chipset differences...

Select to expand quote

berowne said..
So Decrepit, are you saying that wrist mounted GPS can be +/- 1 whole knot out of say 20kt board speed, so 5%!!!???
Doesn't this defeat the argument about GPS chipset differences...

Anything worn on your wrist has a tendency for inaccuracy due to it's access to the satellites when you sail underhanded.

A general purpose fitness watch is not all that accurate to start with no matter which way it facing

No, you misunderstand. It is not about 'chipset differences'. It is about accuracy and error values.

And the error values in the particular example shown are whole orders of magnitude less than +/-1knot. They are all under 0.1 knots. (one tenth of a knot)

There is now a very large body of data to compare the approved GPS's, collected by a number of people, including comparing wrist mounting and upper arm and head mounting. Much of it comparing the exact same 'GPS chipset'

What is quite clear, is that error values do regularly show higher on wrist mounted GPS, especially when 'under grip' is used for the wrist worn device, and especially during the rig flip in a Gybe (Alpha).

BUT: Because of very good basic design, emphasising our specific needs, the difference between the actual GPSTC results is usually very much below the error range reported. But we only know this and can rely on it because we have error data in our comparisons, and a very large body of comparison data at that.

The speed reporting reliability of a device is much more than just what GPS chipset is used. Much, much more.

And one of those things is simply stated: 'Wearing of a GPS on the wrist has a lower accuracy reliability than wearing on the upper arm or head'.

Thankfully our data very strongly shows that this is very seldom an issue, and when it is, it can be almost invariably picked up by good software analysis program filters. On the very rare occasions where there still may be doubt, all the data parameters recorded can be used in conjunction with various software, by those with the experience and expertese, to validate.

That +/- 1kt is only at rig flip, I guess it's out for about 2s, that's not going to make a huge difference in the total alpha. On a speed run, it's preferable to wear the watch on whatever wrist you have up. The error on a straight run with the front hand down isn't as bad as in a gybe. But it's definitely better with the hand up.