While testing the DIY loggers, I've come across an interesting anomaly between 2s results from the units on my head and the ones on shoulder and wrist.
I was a bit concerned the first time I saw it, one of the loggers was below the it's error margin in a number of spots, by typically 0.1 to 0.2 knots. When I checked, all of them are a curved track during a bear off into a gybe.
To be sure it's not something peculiar to any particular logger, in the last 3 sessions I've changed which device is on my arm, and I still get the same results. Unit on arm gives a slower 2s going into a gybe, enough to be outside the error margin.
Here's the results from Yesterday.
So there's 3 units with time, knts and +/- columns.
Left hand unit is the GW52 on my head, (18 seconds behind) middle unit is V3 on my head, and right hand unit is V2 on my arm.
So all results are within the +/- value, except the ones with a red G alongside.
In each case the data looks great, that's what made me check the tracks, this is typical.
So it doesn't look like a very tight curve, but all I can conclude is that the unit on the arm, is on the inside of the curve so going slower than the units on my head further out on the curve.
Any mathematicians here?
I guess 0.2kts is around 1% of 25kts, so could the arm and head have a 1% difference in track length.
Is that also 1% radial difference , I think so, from memory circumference is pi x D.
So if it;s a 50m radius, (and I think it's bigger than that) then the two units need to be 50cm apart on the radius. Maybe between my head and wrist, but not head and arm.
So what's going on here????
Anybody else seen anything similar?
Trouble is my gybes on the other tack are normally crap, I probably don't accelerate into them as much, becvause it's normally more choppy going the other way.
But Thanks Dave, it gives me the next experiment.
Wear it on the other arm!!!!
And it should be faster, we'll see next sail.
if I remember!
I remember seeing this before, although I think it was in the alphas for some of your earlier test data. Your radius theory is a good one, but what puzzled me a bit was that the differences vary so much - for some runs, the arm unit has exactly the same speed, and in others, it's 0.3 knots slower. The variation seems too big to be random, given the low error estimates.
You may actually see real speed differences between your arm and head! A speed difference of 0.3 knots is 0.15 m/s. Over 2 seconds, that's 30 cm. If you do a nice oversheeted jibe entry, the back arm should move to the back of the board, while your head moves very little. The arm movement can easily be 20 or 30 cm.
If this is indeed the cause, you should see a proportionally larger effect for a wrist-mounted unit. If you compare the speed tracks directly, you may even be able to differentiate between "arm movement" and "radius" speed differences: the arm movement should be short, while the radius should affect a longer part of the track. May be difficult to see, though, since the error per point is as large as the differences you're looking for.
If you want to really geek out, you could film your jibes with a GoPro, and try to correlate your arm movements in the jibes to the observed speed differences.
Select to expand quote
I remember seeing this before, although I think it was in the alphas for some of your earlier test data. Your radius theory is a good one, but what puzzled me a bit was that the differences vary so much - for some runs, the arm unit has exactly the same speed, and in others, it's 0.3 knots slower. The variation seems too big to be random, given the low error estimates.
Are you referring to my alphas here? Because in my 2s results above, it's only the lines with the red "G" next to them are gybe entries, the rest that are much closer, are all straight line speed runs.
I think this is happening before I over sheet, as I'm accelerating.
Quite often my gybe entry is very close to my top speed downwind, that's why these are showing up in my top 10 results for the session.
And typical old timers, I've confused diameter and radius. I compared the GPS Results 50m proximity circle with the 2s curve and my track looks to be wider, so it may only be 30m radius not 50m. I'll have a go at measuring it.
Yes the watch also shows slower speeds, and only a bit slower than the arm. Which is probably the case if I'm still sheeted in. I'll see what happens wearing it on the front arm instead of the back arm. Thinking about it now, I don't expect it to be faster, but the difference may not be as great.
Here's an earlier one where I included the watch., 
I fed the gybe into GPSVisualizer, printed it out and used the compass to measure the radius.
If the 20m scale at the bottom is accurate, the radius is about 110m. if my maths are correct, from my head to shoulder is about a meter.
Is this an alpha stretch phenomenon?
Select to expand quotedecrepit said..
Are you referring to my alphas here? Because in my 2s results above, it's only the lines with the red "G" next to them are gybe entries, the rest that are much closer, are all straight line speed runs.
Thanks for clarifying that, I missed that.
What I meant to say was that the observed speed differences for the 2s results amount to just 15-20 cm, which can be caused by a combination of 2 things:
1. Radius differences in the turn
2. Actual movements of the two GPS units relative to each other (arm/wrist vs. head).
Both can contribute, but the second part may well be larger. Here are some screen shots from a mast cam that shows three GPS units on my head, arm, and wrist:
Before the jibe:
Mid jibe:
After the jibe on the new tack:
Relative to the GPS on the head, the one on the arm has moved back about 60 cm, and the one on the wrist about a meter.
Depending on how exactly you move your head, arm, and hands in a given jibe, it's quite easy to imagine that a GPS on the wrist or arm moves 20 cm relative to the one on your head, which would lead to the differences you observed.
What's the expected effect on alphas? if the wrist moves back 1 m due to the tack change, that's just 0.2% of the total. For a 25 knot alpha, that would cause a 0.05 knot speed difference between head and wrist mounted GPS. The difference between mounting the GPS on the left and right arm would be about 0.1 knots. Both numbers are higher than typical error margins for ublox data (0.03 knots for alpha 500 in the session from which the pics above are).
Select to expand quotedecrepit said..
I've confused diameter and radius. I compared the GPS Results 50m proximity circle with the 2s curve and my track looks to be wider, so it may only be 30m radius not 50m. I'll have a go at measuring it.
Actually, the width of the jibe does not matter. We're looking at two half circles, one of size x and one of size x + d (where d is the difference in radius caused by different GPS positions). The length of half circle one is x * pi, the the length of the other one is (x+d) * pi. The difference is d*pi, regardless of x.
So if one GPS is on the helmet, and the other on the wrist, let's say that the difference is about 60 cm. That means the distance traveled by the GPS on the wrist would be about 2 meters less. For a 25-knot alpha 500, the GPS on the wrist would measure just 24.9 knots (looking at just the "radius effect").
But Peter, I'm not talking about what happens during alphas or gybes. I'm talking about the curved path before the gybe.
On you mast cam, it looks like the head unit is inside the arm unit. and the arm unit outside. But I wear my GPS on he other arm, so if my stance is similar, there would be a slight difference in radius.
Yes and I've been deleting the relevant column, the distance traveled!
If I get bored today, I'll attempt a statistical analysis of the average difference between a curved 2s and a straight 2s. That should give a clearer idea of the numbers involved.
From what your saying the position of where you mount the GPS unit when your turning can effect the measured speed, granted its only a little, and splitting hairs but...
If you were doing a 1hr GPS race and there were lots of turning points, lets say 20 gybes in the hour, those little bits x 20 could add up to a few seconds? Which may be 100m or so ? (I'm guestimating !:) you guys seem to have the math down to a fine art so hoping there is a calculation to confirm either way :) The longer the session and more turns the greater the difference?
Location of the GPS 'could' be the difference between 1st and 2nd, or last and 2nd last:) in a 1 Hr race with lots of turns?
Is it fair to say the fastest point to have the GPS unit located then is on the mast?
Hmm, depends on what happens on the opposite gybe, yet to be determined. If one way is slower and the other faster, there may not be much net difference.
And I don't think the mast will be faster, that's always going to be on the inside of a carve gybe, so will be slower. However if you're on a wally doing flair gybes it may be faster.
Although looking at Peter's gybe above, I could be completely wrong about that. Depends how you gybe I guess.
Select to expand quotedecrepit said..
Hmm, depends on what happens on the opposite gybe, yet to be determined. If one way is slower and the other faster, there may not be much net difference.
And I don't think the mast will be faster, that's always going to be on the inside of a carve gybe, so will be slower. However if you're on a wally doing flair gybes it may be faster.
Although looking at Peter's gybe above, I could be completely wrong about that. Depends how you gybe I guess.
Next you will analysing how far forward to put your back foot when initiating the gybe, how far to lean into the corner and if you tongue should be sticking out to the inside or outside of the corner... mind boggles. But keep at it fellas!
Select to expand quoteBSN101 said..
Next you will analysing how far forward to put your back foot when initiating the gybe, how far to lean into the corner and if you tongue should be sticking out to the inside or outside of the corner... mind boggles. But keep at it fellas!
Well, you obviously should do one-handed jibes, removing the hand with the watch from the boom and sticking it to the outside of the turn. You may be able to gain a few extra milli-knots by waving, as long as you wave perpendicular to the direction of travel. If you stick the tongue out, it should point in the direction you're waving 
.Select to expand quoteJonesySail said..
If you were doing a 1hr GPS race and there were lots of turning points, lets say 20 gybes in the hour, those little bits x 20 could add up to a few seconds?
Absolutely! You just have to make sure that all of your jibes are either starboard or port jibes - no mixing !
Mike can you check out what happens when you do tacks instead of gybes? Cheers Stretchy
Select to expand quoteBSN101 said..
Mike can you check out what happens when you do tacks instead of gybes? Cheers Stretchy
Nup, sorry Dave, can't tack. Boards are all too small
Select to expand quotedecrepit said..BSN101 said..
Mike can you check out what happens when you do tacks instead of gybes? Cheers Stretchy
Nup, sorry Dave, can't tack. Boards are all too small
No that was, um Stretchy askin.
im innocent
So yesterday I wore the GW52 on the left arm instead of the right, and made sure I went for plenty of alphas. Didn't work all that well as I either ran out of wind on the out bound leg, went to wide following weed channels or just bogged down in thick weed.
However most of my runs in were OK. so I have some valid results.
So now we have no red Gs only black, to indicte which of these results are gybe entries.
And it looks like the two biggest differences -0.092 and +0.082 each belong to a gybe, but they are different signs!
And one of the gybes at 0.003 is the equal second lowest difference.
So what is our conclusion????????????????
