Was trying for alphas the other day in lightish winds, couldn't get the watch to read over 22.5, I felt I had done better than that. I was pleasantly surprised to see a 23.415 from the GW52 once I crunched the numbers, so did a bit of investigating and found the GW52 alpha had a separation of 49m it disappears at 48m. Whereas the watch has the same alpha with a 52m separation. So there's a 3m difference between the two.

I had the GW52 on my head and the watch in the middle of my inner forearm with display inverted.

The GW52 is the left one at 50m the right one is the watch at 52m
There's not much difference between them, the SDoPs are very similar. the watch has one very brief sat drop out and the doppler and positional tracks aren't quite as well aligned. But 3m different in separation, this could cost you a PB!
It's very tempting to aim for 45m just to be sure the alpha is legal.

I checked the rest of my alphas, and it's not consistent, some are identical, so it's a random variation as well.
Any thoughts?

could it be, that because the measured speeds are little different, that the fastest alpha for the gw52 is with 48m, where the gw60 have it with a 50m gate ?

but usually if the gate is bigger then 50m it sounds like, the gate for calculations is higher then 50m

the picture is very funny...

Select to expand quote

WST said..
could it be, that because the measured speeds are little different, that the fastest alpha for the gw52 is with 48m, where the gw60 have it with a 50m gate ?

but usually if the gate is bigger then 50m it sounds like, the gate for calculations is higher then 50m

Nup, this alpha doesn't exist at 48m for the GW52 and 50m for the GW60.
Yes, I've adjusted the gate width, in the pics above, (and sorry for the suggestion of male genitalia, purely accidental I assure you), the gate is set at 50m for the GW52 and 52m for the GW60. This allows the alpha to be displayed for both devices

The proximity circle is calculated from GPS positional data which is known to be not completely accurate. This source of potential error has been well known for quite a while.

It was the topic of long discussions between Mal Wright, Tom Chalko, Manfred Fuchs, myself and others when we changed to using the Doppler speed data (2007?). Manfred suggested we use the Doppler derived heading data for the proximity circle as he reasoned it should be more accurate. We tried it in GPS-Results, but unfortunately we got too many erroneous results. These were the consequence of a small error in the heading, usually during the gybe (where it is unfortunately more likely). If the the heading plotted a course slightly straighter than the true course even for one point, the track contuned to reflect this error on all subsequent points. In other words, the error in heading was not corrected in any way and this could result in the course being well outside the proximity circle according to the Doppler derived heading, but inside the proximity circle when calculated using the positional data.

This is one of reasons why we have always said that the Apha is a result that can not be completely relied upon to be always accurate and why there are not WGPSSRC records for this category.

We rely on our observations that positional accuracy is usually reasonably reliable over short periods, such as the time it takes to do an Alpha 500 so the error in position is usually fairly small. Unfortunately, this is not 100% reliable as you have found in your GPS comparison.

With the current state of technology regarding positional accuracy for normal GPS's, there is not much the can be done about this. The greatest potential increase in positional accuracy seems to be the use of RTK technology, but this seems to be some way off general availabliliy or acceptance at the moment. In the meantime, I think we are just stuck with this issue. We need to realise that if the Alpha is within a meter or two of the circle limits, it is more likely to be less reliable. Not much help to those of us who actually aim to be 'just' inside the circle.

Another potential source of confidence it to use more than one GPS. That way if they are both saying it was inside the circle you may be just a little more confident.

By far the best is to use the GPS(s) on your head where it has been shown to produce less errors in both Doppler speed and Position, especially during Alpha gybes.

I must add that over the years we have had quite a few instances where different analysis software give different results with the same track. (where the position is extremely close to the circle). We have generally had the policy that if one software accepts the run, then we also accept it. (And no, there is not one particular software that is known to accept more close results than any other! )

Thanks Andrew, So if anybody is setting up an alpha gate, best put it at 47m, then stay a tad inside it.

besides rtk wich depends on the device, its also possible to use offline data.

in Germany and other European countrys there are some service providers for offline data.
Also there are some internert services available using a so called Ntrip-protocol.

the last one would be interessting for mobile phones.

it cost money, but with such data you can achive a accuracy in about 2-3 cm.

Select to expand quote

WST said..
besides rtk wich depends on the device, its also possible to use offline data.

in Germany and other European countrys there are some service providers for offline data.
Also there are some internert services available using a so called Ntrip-protocol.

the last one would be interessting for mobile phones.

it cost money, but with such data you can achive a accuracy in about 2-3 cm.

The use of Ntrip still requires the use of an RTK or high end differential correction capable GPS.

In other words, one still need to have an RTK device or survey quality GPS that is capable of differential correction. It just eliminates the need for the user to also have a their own differential base station.

I have read about various projects where researchers are trying to use mobile phones for RTK, but as far as I know, this is not available commercially for cm positioning yet.

There is a challenge for you.