Yet another Windvane Mod

[steve]

I think this is why wind direction graphs are normally done as point/scatter charts rather than line charts. You still get the dots spread between the top and bottom of the graph when it's around north, but you don't get the odd joining lines.

[Gina]

Yes, I agree, scatter plots avoid this problem to an extent but from my research they are not considered as readable as a line plot.

I have a couple of ideas I'm looking at.

[EvilV]

That 355 degrees / 05degrees thing is a tricky one. I have no idea how that could be solved. I'd be interested to know what Gina is thinking of as a solution.

Actually, come to think of it, that wind rose thing on the Cumulus screen does seem to be a way out. I'm assuming the length of the spikes represents the time the wind came from the particular direction.

[steve]

Actually, come to think of it, that wind rose thing on the Cumulus screen does seem to be a way out. I'm assuming the length of the spikes represents the time the wind came from the particular direction.

It takes speed into account, as well as length of time. And they're not supposed to be spikes; they become spikes because of the Fine Offset design problem where it mostly only reports 8 compass directions. You can make it look more like it's supposed to look by choosing 8 wind rose points in the display settings.

[Gina]

That 355 degrees / 05degrees thing is a tricky one. I have no idea how that could be solved. I'd be interested to know what Gina is thinking of as a solution.

I'll do a bit of experimenting before I make a total fool of myself It's certainly a tricky one!

[EvilV]

That 355 degrees / 05degrees thing is a tricky one. I have no idea how that could be solved. I'd be interested to know what Gina is thinking of as a solution.

I'll do a bit of experimenting before I make a total fool of myself It's certainly a tricky one!

I can't imagine anyone less likely to 'make a fool' of themselves. Now me - that's a different story.

[EvilV]

Actually, come to think of it, that wind rose thing on the Cumulus screen does seem to be a way out. I'm assuming the length of the spikes represents the time the wind came from the particular direction.

It takes speed into account, as well as length of time. And they're not supposed to be spikes; they become spikes because of the Fine Offset design problem where it mostly only reports 8 compass directions. You can make it look more like it's supposed to look by choosing 8 wind rose points in the display settings.

Thanks Steve.

I'm ashamed to admit that I only use the software to make the link between my gadget and the Wunderground, so I only look at the laptop that runs it as a server about once a week. I did find the software very versatile to set up the parameters that I wanted though. It also runs in an extremely stable and reliable way.

[Gina]

I can't imagine anyone less likely to 'make a fool' of themselves.

Thank you for your vote of confidence

One thing I was thinking of was to "look" at adjacent points and try to determine the relative movement by seeing which way gives the smallest movement. eg. 14 15 0 1 15 14 which is NW NNW N NNE NNW NW The change from 15 to 0 or NNW to N is actually only a one point change and not 15. Similarly. 1 to 15 is actually NNE to NNW, 2 points change rather than 14.

A possible snag with this is that the direction may be more than one rotation in the time scale covered. There is no reason why this should not occur in reality particularly in light and variable winds. Two ways of representing this 1) Let the y scale cover more than 360 degrees or 2) Show a step or break in the trace. Actually, Force 0 or 1 variable, calm or light, really could do without any points plotted - a break in the graph. I don't currently know how to do this in gnuplot (the graph plotting app I'm using).

[Gina]

I've got this working in principle - now need to change the y-axis labels to compass points. And I could add some averaging (smoothing) now too.

[Gina]

I decided to keep the 360 degrees range and to choose which compass point to centre it on and allow an end-to-end jump where the plot covers more than 360 degrees over it's time range. The first plot shows a 24 hour time scale.

24hrs-wind-dir-E.png

The plot below is for 7 days and shows a jump for Tuesday, 23 March.

7days-wind-dir.png

For the 28 day plot I decided to leave the data as the daily average but use steps instead of a straight line connection between points.

28days-wind-dir.png

[RCE]

Nice job Gina

[Gina]

Thank you

[scientistuk]

my god the things we do to mod our fine offsets . brill idea but sure your neighbours think u got a screw loose with part of a frying pan up there lol

[harrym1byt]

Also, might not damping prevent the vane moving in light very light winds?

It should make no difference at all. Eddy current damping only works to resist movement, when the movement is brisk. No attempt to move means the damping level is zero. It acts like a motor or generator, which has been shorted out - the harder you try to turn it, the more it will resist the movement. Turn it very slowly and there will be very little resistance.

I like the idea by the way.

[sanramonrover]

And yet another mod to the windvane!

Research for photos of professional units on the web usually show that the tail vane is a GREAT distance from the pivot point, while the counter weight at the pointy end is very close to the pivot.

Our units don't have room to make the tail much longer without interferring with the anemometer.

So I used a few bits of electrical conduit to raise the vane well above the anemometer. Then added a very crude long tail piece to the original.
Added weights taped to the pointy end to balance everything. It is not aesthetic yet, but seems to stop some 'windmilling'...

Now the only problem is my transmitter has lost contact with the receiver so often I can't tell if it really works or not.
New batteries, cleaned contacts, replugged everything but still no regular contact. Maybe another dead 1081