WH3081 - Wind Dir sensor possible fix

[aussiewmr]

Hi All,

I have had a WS3081 up and running for 2 days and like some here the first thing I noticed was the wind direction sensor was not that good at its job.

In my case one of the 8 cardinals was faulty (no reading at all, was quite obvious with Steves graphs) and all of the intermediate ones were missing (the ones where two reed contacts should close when the magnet is inbetween). So 7 of 8 or 7 of 16 depending on how detailed you want to get.

Thanks to the forum post showing the innards of this sensor I pulled mine apart. And after some testing I discovered one faulty reed switch ( i had spares for my old rain gauge) and then came to the realisation that in my case the intermediate directions were never going to happen because the magnet was too far away.

So rather that play with magnets etc I decided to raise the reeds off the circuit board by about 1 1/2 to 2mm by re-soldering them with a good blob of solder under the reed switch wire ends. This was possible as the distance between the board and the case could accommodate this.

I also noticed that most reed switches were orientated differently. Some had the reed's parallel to the board, some perpendicular, some on an angle like 45 degrees, some had the reeds at one end up, and some were down etc. From what I could research on the net it did not seem to matter for these type of reed switches, but non the less I made them all the same (reeds parallel to the board and the outer contact would pull up away from the board).

At the end of all of this I now have all of the cardinals working on 16 positions on the wind graph in Cumulus.

This is not an easy fix, requires reasonable soldering skills, a means of securing the board while you work, a magnifying glass and tweezers helps and a very steady hand using a fine tipped soldering iron (around 25 Watt max). The reeds would be very easy to break. and of course do this at your own peril - I aint responsible for broken WS's.
The good bit is you do not have to pull off the top rotating vane, just undo the three screws under the unit, and the three screws holding the board in place.

Cheers

Phil

[AllyCat]

Hi Phil,

Thanks for the information and congratulations on your soldering ability (and eyesight). One further word of warning for others, the captive cable should be glued onto the circuit board; if it is loose, the wires will probably soon break (mine did) and then there is further repair work to be done.

Obviously your sensor was "faulty", but AFAIK they all report much smaller intermediate directions (NNE, etc.), typically only 10% of the 8 primary points. Below I've attached a screenshot from Cumulus of a typical vane "out of the box". The vane was rotated uniformly at one revolution every 3 hours to give the "staircase" and Wind Rose (with constant simulated windspeed).

Vane3-uncorrx2.png

Theoretically (because the reeds are operated by the magnetic field along the axis of the switch) the orientation of the reeds shouldn't matter, but there may be some "second order" effect as the magnet seems so unsuitably chosen (and orientated) for the purpose. Did you actually measure the size/accuracy of the intermediate compass points after your modification?

Personally, I found that a single neodymium magnet 10mm (diameter) by 2mm (thickness) self-attached directly onto the existing magnet (inside the rotor) gave good results, but is rather too large to fit neatly inside the head. So I purchased some 8mm x 1mm from ebay and found that two side-by-side gave a similar result. However these magnets are so strong that I had to glue them onto a sliver of thin metal sheet to prevent one jumping onto the other. Now, I think the optimum arrangement might be three 6mm x 1mm magnets mounted similarly, but have yet to try. Below is my latest comparison of "before" and "after" wind roses.

Cheers, Alan.

WindVaneModification.png

[dpmiller]

I drilled a tiny hole in the top of the vane above the OE magnet; then pressed it down towards the sensors using a narrow probe in the hole. Having it *just* clear of the stator does improve the response.

[aussiewmr]

Hi Alan,

Some excellent info there.

Yes you are righ, my intermediate points are far less prominent than the primary points. But for me it was a case of I had no intermediates at all. I did notice my magnet was recessed into the head by about 1mm but was not wanting to drill holes as dpmiller suggested.

I have a spare so I will experiment with magnets as you suggest. I would certainly be happy with your second rose. Dont suppose you have a picture of how you ended up?

Phil

[AllyCat]

Hi Phil,

Yes, the wind rose is a very sensitive indicator of the vane accuracy (or lack of) and it took some tweaking to get my above result. Eventually I hope to write up the various measurement/adjustment methods that I devised, but there are so many "improvements" that can be made to the various FO sensors. Currently, I'm attempting to measure/derive a calibration factor for the anemometer, following Steve's comment in this thread.

Below is a photo of my magnet modification, with a second magnet assembly (inverted) beside. I drilled a 4mm hole in the thin metal sheet (before cutting/filing to final size) to fit over the magnet in the rotor head. It was easier to fit one magnet directly over the original, the second is then to one side, but the 10 degrees offset can be easily compensated, either when mounting the sensor or within Cumulus. As mentioned above, the two neodymium magnets are 8mm x 1mm glued onto a small metal segment.

Cheers, Alan.

FOWindVaneMagnetS.jpg

[aussiewmr]

That's great Alan,

Doing this obviously alleviates the need to raise the reed switches.

I presume you have used a hardening glue like resin or Araldite as opposed to a hot melt glue.

Thanks

Phil

[AllyCat]

Hi Phil,

Yes I did use an epoxy resin glue for longevity (probably far longer than the FO kit will last ) but even a heavy-duty double-sided tape was sufficient for initial tests and I suspect that hot-melt glue would be fine (even in Australia ).

Cheers, Alan.