Sandaysoft

Great forum and a pleasure to be able to contribute. (Yes, Steve, I'll also donate! Cumulus is a great Product.)

I installed my Fine Offset WH1081 about 6 weeks ago -- unfortunately the anemometer supplied did not work and while waiting for a replacement I have had a temporary installation for testing which favours easy access rather than a well-performing WS .

While I have been cooling my heels waiting for a replacement, I have been lurking and learning on this forum and planning the installation and mods.

My primary concern has been my windvane which is very sensitive to low speed wind but appears to be too sensitive, with frequent windmilling, in gusty weather. From other entries in the forum this seems to be a common problem.

For some time I have had a bee in my bonnet about using eddy-current damping to solve the wind-vane instability problem as I consider its characteristics to be ideal for this application. I have now completed a proof of concept prototype that I can share.

How does eddy current damping work? The theory is quite simple, a changing magnetic field creates eddy currents in conductive materials, think of them as vortexes. These eddy currents create forces which oppose the change in the magnetic field. In application of the theory, I rotate a magnet over an aluminium plate. The movement of the magnet creates a changing magnetic field in the aluminium plate. The eddy currents induced in the plate create a reverse polarity magnetic dipole which attracts and slows the magnet. The faster the magnet moves the greater the retardation force. Thus when stationary there is absolutely no attraction so the sensitivity of the wind vane to low air movement is not affected. This is the ideal characteristic I refered to.

The Prototype
Mindful of my warranty, the prototype for the PoC had to be low-impact without permanent change.

Step 1. I attached a powerful magnet to the vane rudder. [I chose the rudder to keep the new magnet as far as possible from the existing one]. The magnet chosen was a neodymium button 9.5 mm diam x 6.4mm thick with a pull of 9.2lbs. This was purchased over the Internet [USD8 for 10 (AUD20 including p&h)]. I considered using a fridge magnet, but these are not very strong, necessitating very close separation of the magnet and the plate to get equivalent drag. The magnet is fixed to the rudder using the cap of a bic-click ballpoint pen (I have a lot of these for some reason) which has a thin cut vertically to slide onto the rudder. At the moment it is a push fit only but I will silicon seal it soon. In the tip of the cap there is a 6G self-tapping screw to which the magnet self-adheres. The screw gives a crude adjustment capability to control the separation of the magnet and plate.


For the plate I used a 20cm fryingpan, with handle removed, and drilled a 15mm hole through the centre of the base. To the north and south sides of this hole there are joined two 6 mm holes to accomodate the wires. The base is about 2mm thick and quite rigid. The plate is pushed up against the skirt of the vane and is held in place with a thick rubber washer (rescued from a Nylex Gardena hose fitting). Only slight compression of the washer was needed to insert the mounting bolt and a snug fit was achieved.


The magnet is tracking about 1mm above the plate. The gap seems to be quite uniform over the locus of the magnet. If it had been variable I would have levelled the plate using shims between the skirt and the plate.


I considered removing the curved sides of the fryingpan but elected not to as I don't have a lathe and thought I might warp the plate if I cut it by hand. The plate is not attached to the frame but if necessary could be supported using the handle mounting hardware.

Testing has been very rudimentary. I gave the rudder a tap and counted the revolutions -- without damping I counted four revolutions for a moderate tap; with damping I could only get a quarter revolution for even quite hard taps. Light breezes, which is all I have had in the past 24hrs are still moving the vane. That is sufficient for me to assume the mod is performing as designed. Here is a short movie of the tests. [It's pretty ragged as I took it wth my phone, but you will get the idea].
http://members.optusnet.com.au/~rstuart/Weather/ModDoc/Windvane_Eddy_Current_Mod.wmv


For ligth wids it's performing quit well. It remains to be seem whether it will significantly affect the recordings over the full range of wind speeds!
You can see the published results for my temporary installation at
http://members.optusnet.com.au/~rstuart/Weather/


Design Issues.

• I won't have a problem with snow but the plate would be a good place for it to lodge. perhaps a vertical plate would be better. Could keep it in line with the skirt. A coke can is about the right size
• Don't know if 1 mm clearance will be enough to avoid problems with lodged rain water

Additional ideas and comments would be welcome.

_________________
Cheers,

Rollo.

What an amazing bit of lateral thinking! One comment/question: The wind vane works by a magnet operating reed switches. Is there any possibility that the magnetic fields you are introducing could affect the normal operation?

_________________
Steve
Sanday Weather
----------------------------------------------------------------------------------------------------------------------------------
Like Cumulus and want to support it? Please donate!

Yes, Steve. I'm sure it would be possible to affect the reed switches. Which is why i located the new magnet as far away as possible from the works.

I originally thought to mount an annular ring plate to the vane, with a fixed magnet on the frame. Apart from the problem of keeping the plate perpendicular to the axis if rotation, I found that all switches locked off due to the proximity of the powerful magnet. [Though at that time i was using the stack of 10 magnets and the vane only pointed at the stack].

Mounting the magnet so the N/S poles are horizontal rather rhan vertical would also reduce interference.

_________________
Cheers,

Rollo.

I too have been thinking about the possibility of magnetic damping but also concerned about the reed switches. I was thinking of a disc attached to the vane spindle and a fixed magnet. But I think the vane itself is probably good enough now that it's a lot higher up. Next step, software smoothing/averaging.

_________________
Gina

Sorry, no banner - weather station out of action. Hoping to be up and running with a new home-made one soon.

very interresting,
seems most promissing (maybe THE solution)

Does the width of the plate have a minimum size,or does that depend of the magnets created magnetic field/eddy currents?
Can the plate be too big and thus counteract/disturb the magnetic field or eddy currents?

i agree that a vertical solution might be better,since a plate can create it's own (minimal)turbulence/drag,and should be easier to build,maybe an old soup/catfood tin

the magnet could be much lower thus further away from the reeds,also one could use more magnets(the weaker fridge magnets)around it.

just food for thought...

_________________
http://home.xmsnet.nl/hanmir/
http://home.xmsnet.nl/hanmir/radarbox.htm

Hi Gina, thanks for your support of this forum - I'm always interested in your opinion.

I think the manufacturers have done a good job in producing a sensitive instrument and maybe the ultimate answer is for them to address the damping issue. Maybe an aluminium plate between their magnet and reeds? I'd like to look inside to see if it's possible.

IMHO I don't agree that software averaging is the answer. I has its place depending on our need, but it should not be used to filter error points. As an example, if my vane is pointing west and there is a gust from the north, the instrument should register the north wind. However, in its pristine state the vane will windmill and any direction may be recorded depending on the sampling interval. Tha's what I would call an error point since the wind never actually blew from that direction. [Statistically, I would guess the 95% confidence band would be approaching +/- 180 degrees so a large sample is needed to provide confidence in even the average. Research into gusts would be meaningless].

G'day Hans,
By width of plate do you mean the diameter? or the width of, say, an annular ring over whch the magnet could move? If the first --- I am currently tracking over a 130mm diameter circle. If I came down to a 60mm circle (diameter of skirt at base) the tangential speed of the magnet over the plate would be approximately halved and thus the rate of change of the magnetic field in the plate would be halved and the opposing magnetic dipole would be less and retardation less. [I dont' know if it's linear, I'd need to brush cobwebs out of my brain and dig into the math. It's more fun to suck it and see!]. Nett result, with the magnet I am using, it should not matter --- Width of an annular ring should not be too critical. The further from the pole the less the current induced and the less effect on the dipole. FWIW i'd be happy with 1-2 widths of magnet. For the same reason the plate can't be too big. I'b be more concerned whether it changed the characteristics of the wind.

Your suggestion of cat-food can is a good one (as long as it is non-ferrous) the rim will give good stability which my coke can does not have. I'll take my magnets to the supermarket -- my cats don't normally get those very expensive little single serves.

I tried a fridge magnet on my rig. This was a ferrite(?) disk 17mm diam and 3 mm thick. However there is something funny abot it. I could only get it to adher to the screw on one face as if both N and S poles were on the same face. Its not that the screw has been magnetised because the other face won't adher to steel either. Weird! I had to blu-tac it on and got no appreciable retardation. May need to try different fridge magnets. More may help.

_________________
Cheers,

Rollo.

Thanks for that Rollo I haven't checked but it would make sense for fridge magnets to have both poles on one side as that would increase the adhesion force onto a steel plate. Assuming it's the side facing the fridge, of course.

_________________
Gina

Sorry, no banner - weather station out of action. Hoping to be up and running with a new home-made one soon.

Here's an update showing my wind data from the last 24 hours without any averaging other than in the station itself. Sampling interval is 5 mins.

_________________
Gina

Sorry, no banner - weather station out of action. Hoping to be up and running with a new home-made one soon.

If as has been suggested by Gina, the main issue of the flip flopping wind vane is turbulence caused by installing the kit too low and in the shadow of wind obstructions which create buffeting, how will damping help?

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

_________________
My weather data page on WU:
http://www.wunderground.com/weatherstation/WXDailyHistory.asp?ID=INEWCAST6

This is an excellent idea! Can you post screenshots of the resulting direction graph?

One thought - if you moved your magnet to the front, any stray flux would impact the same reed switch as the stock magnet. Since the stock magnet really needs to be a bit stronger, you might get 2 birds with one stone. Also, (O.K. two thoughts) unless your tower is perfectly vertical, you will need to counterbalance the additional weight.

I hope the can idea works. Being in a snow area, vertical would be better for me. However, the thin walls will have higher resistance, thereby reducing (somewhat) the eddy currents. But maybe not significantly... Cut and try...

Gina and EvilV, I would have to concede that my present installation (2.4m of the ground and surrounded by buildings and trees) is more susceptable to gusts than a rooftop, but I don't think that would explain windmilling of 3-4 revolutions over 5 seconds. I may be wrong as weather is a chaotic system.
I have a 4m mast ready to install on my roof and I'll check again for windmilling when it is up. [I do want to get the bugs out first and do it only once as my wife thinks I'm too old to be climbing on roofs]. Can you confirm that you don't see any windmilling in that situation?
As I said previously, I am a bit suspicious of averaging (whether done by the sensor or later in the process).A concern about GI/GO. E.g. To interpret Gina's direction graph: does it mean that at 1650 or so, the wind backed off from SW to NE and blew like that for 5 minutes? Is that possible? I don't know as I'm a tyro at this game.

EvilV, The issue to me is that the wind gust adds too much kinetic energy to the vane and we have a overrun past the direction of the gust (what I have called windmilling) as soon as the gust back off. Eddy current damping converts this excess kinetic energy to heat.
There is no damping when stationary and little for light winds. As static friction is greater than rolling friction, if the wind is strong enough to start the vane moving it will be enough to overcome the limited drag for light winds.

Charlie, Sorry, I can't show the graph at the moment.
I tried to add a 3m extension to my windvane cable and have lost all readings since 2pm. I cant see an easy way to get a historical plot (is there one?), so I'll post a sample as soon as I have corrected the issue.

You're quite right about the weight and counterbalancing. There are a numer of hurdles to get over before I dare call it anything other than prototype. Prototype 2 using a coke can is underway. the first problem is to drill the centre hole accurately enough so that the magnet rides at a constant gap; also the thin walls are buckling with the slightest provacation! As you say, the effectiveness of the thin walls to hold the eddy-currents remains to be seen!

_________________
Cheers,

Rollo.

Thanks for your answer Rollo. I'm not criticising your efforts - I thought it was ingenious. I've seen a few suggestions about the flapping wind vane and when looked into, they were pretty much irrelevant (even if they did seem plausible). My vane is a bit more than 20 feet up, but has a small tree nearby which is a similar height. At this time of year, teh tree amounts to just a few twigs since the leaves are not on it yet. There are houses about 25 metres away on two sides which are obviously taller than my mast.

So far, I have extended the tail on the wind vane, and it made no difference. Then, in response to assertions that the anemometer was creating turbulence and the cups were chucking gobbets of air at the vane as it revolved, I stopped the rotor with an elastic band. and put it all back up again. This made no difference either. Gina has reported a calming effect when she put her kit higher up and this makes sense. Air moving over ground which has obstructions on it obviously flows in eddies and buffets the wind vane. The same problem is reported by the sellers of those mini wind generator kits that were being sold to people to tack onto their houses. Apparently, the output is seriously lowered by what they call 'dirty air' - air that has travelled over obstructions and arrives from different angles in a turbulent way.

I am convinced that the effects we are seeing are simple turbulence brought on by the fact that most of us live in towns and can't put the kit up high enough. Of course the simple weather stations like you and I have, while excellent value, tend to encourage us to put the whole lot up in the air. If you do that with the temperature and rainfall sensors, you will get ridiculous temperature measurements when the sun shines on the rather inadequate Stephenson screen. Unless you have a very stiff pole, the wobbling rain gauge will count extra rain as the tipping bucket is shaken about.

I spilt my wind sensors from the temp and rainfall ones by using an RJ11 phone extension cable so I can have one lot up top and the other down in the shade and stable.

You ask about windmilling. No - I don't get that - just a sort of oscillation.

Also, I think your explanation about kinetic energy pulling the vane too far misses something. In non-turbulent air, the swinging vane is responding to pressure on one side of the vane. When the vane is pointing in line with the wind, both sides come under the same pressure, so the movement is arrested. If the vane had a lot of momentum, which it doesn't - it is extremely light, it would carry on a little, but would very soon come under a large force trying to return it in-line with the wind. Since it is light weight, it will stop very rapidly, as long as the air is not turbulent, which of course, in most of our installations it is. This is the problem Rollo, not momentum, or bad design. That's just my ten pennyworth, anyway mate.

Be careful up on that roof too. Remember Rod Hull the keeper of the notorious Emu. He fell off his roof and died when trying to fix his TV areal so he could get a better picture to watch a football match.

http://news.bbc.co.uk/1/hi/entertainment/298801.stm

Good luck and take care.

_________________
My weather data page on WU:
http://www.wunderground.com/weatherstation/WXDailyHistory.asp?ID=INEWCAST6

There is still a certain amount of oscillation (waggling) particularly after a change in wind direction. I've observed this but you can also see it in the wind direction graph at about 0420. I have no explanation for the 1650 event other than I do see occasional bigger deviations from the general direction. My feeling is that some data averaging will give a realistic result but I would prefer mechanical damping as the system does show an under-damped characteristic. I have some experience of damping and the look of under and over damping. Correct damping gives almost no oscillation while giving as fast as possible move to the new position. In this case the "resonant frequency" appears to be of the order of half to one Hertz ie. a cycle time of a second or two.

With a 5 minute sample rate there is no way we can see changes in direction caused by gusts so it seems reasonable to average the direction over something like 15 or 20 minutes (possibly longer) to show the general wind direction. One must still bear in mind a deviation value (error) of one or two points of the compass - say around 30-40 degrees. If we had a higher resolution gauge and frequent sampling (at least once a second) we might be able to do better but whether this would actually be any more use, I doubt.

_________________
Gina

Sorry, no banner - weather station out of action. Hoping to be up and running with a new home-made one soon.

I average my wind readings in Cumulus over ten minutes. This is the result for today.



I lost contact with my wireless router last night and only discovered it this morning. That's why the data isn't complete.

_________________
My weather data page on WU:
http://www.wunderground.com/weatherstation/WXDailyHistory.asp?ID=INEWCAST6

With or without extra averaging, there's a problem with the usual wind direction plot when the wind is around north. One way round this is to change the datum to south instead, if this is better. This screen shot illustrates the point. These are plots with a 5min logging interval. (without averaging).

The plot below shows the direction averaged over an hour with the plot covering 7 days. In this case the directions crossed both the north and south points so just changing the datum won't work.

I'm giving this more thought as a plot of wind direction versus time would be worth having IMO.

_________________
Gina

Sorry, no banner - weather station out of action. Hoping to be up and running with a new home-made one soon.