[Roxymoron]

The Fine Offset Wind Direction sensor tends to jump around at random (I have seen mine turn a full 360 degrees on more than one occasion). I have thought about various forms of mechanical dampening such as an aluminium plate and a strong magnet. I have also tried oil on the bearing and an enlarged fin but that has had little impact.

My set up is a standard Fine Offset model 3081 plus an Arduino receiver picking up the transmissions from the Fine Offset rooftop sensors (around once every 40-48 seconds or so).

I then tried an arithmetic dampening of the form
x = (xold*(d-1) + m)/d where
x is the smoothed value,
xold is the previously calculated smoothed value
m is the current measurement
d is the dampening facor (I initially used d=8)

The smoothing is applied to the North and East (i.e. x and y) component of each wind vector sample (strength and direction), then the smoothed N and E components are reconstituted as an average strength and direction. This has worked very well, but unfortunately, each measurement has only a one eighth impact every 40 seconds. Ideally this would be implemented in Weatherduino to use the sample every 5 seconds or even more frequently.

Snippets of my Arduino code to do this averaging are:

Code:

...
byte packet_size = 9, packet_ID = 10, state = 0, wind_dirn_index = 0, dampening = 8;
float sindir[] = {0.0, 0.256, 0.474, 0.619, 0.670, 0.619, 0.474, 0.256, 0.0, -0.256, -0.474, -0.619, -0.670, -0.619, -0.474, -0.256}; Sin of each of the 16 wind directions
float cosdir[] = {0.670, 0.619, 0.474, 0.256, 0.0, -0.256, -0.474, -0.619, -0.670, -0.619, -0.474, -0.256, 0.0, 0.256, 0.474, 0.619}; Cos of each of the 16 wind directions
float radiation, wind_E_vector = 0.0, wind_N_vector = 0.0, wind_avg, wind_dirn_avg;
...    

void loop()
  {
    ...    
                Serial.print(" ");                //Publish ascii data
                wind_dirn_index = packet[8] & 0x0f;        //Extract wind direction index
                wind_E_vector = (wind_E_vector * (dampening - 1) + get_avg_wind() * sindir[wind_dirn_index]) / dampening; //Smooth x-component
                wind_N_vector = (wind_N_vector * (dampening - 1) + get_avg_wind() * cosdir[wind_dirn_index]) / dampening; //Smooth y-component
                wind_avg = sqrt(wind_E_vector*wind_E_vector + wind_N_vector*wind_N_vector);                  //Wind average strength
                wind_dirn_avg = atan2(wind_E_vector, wind_N_vector)*180/3.1416;                        //Wind averade direction
                if(wind_dirn_avg < 0) wind_dirn_avg = wind_dirn_avg + 360;                        //Convert (-180 to 180) to (0 to 360)
                Serial.print(wind_avg, 1);                   //Average wind speed (knots)
                Serial.print(" ");
                Serial.print((0.67 * get_gust_wind()), 1);  //Wind gust (knots)
     ...

The output from this code is a smoothed wind speed (knots) and direction (degrees).

Would it be feasible to add this into Weatherduino Pro?

[uncle_bob]

On my old anemometer, I put on an old dishwasher drawer wheel. It worked wonders. Unfortunately my new anemometer can't have this mod fitted.

Here's a pick of the old one

[werk_ag]

The output from this code is a smoothed wind speed (knots) and direction (degrees).

Would it be feasible to add this into Weatherduino Pro?

It may be...
At least for the the wind direction it may help. In WeatherDuino Pro2, wind speed is already a running average of the last 2.5 minutes.

At some point I may try something like the code you posted, one problem is that I don't have a secondary wind speed / vane unit, installed on the same place of my main unit, to compare the results.

[Camino]

The output from this code is a smoothed wind speed (knots) and direction (degrees).

Would it be feasible to add this into Weatherduino Pro?

It may be...
At least for the the wind direction it may help. In WeatherDuino Pro2, wind speed is already a running average of the last 2.5 minutes.

At some point I may try something like the code you posted, one problem is that I don't have a secondary wind speed / vane unit, installed on the same place of my main unit, to compare the results.

Dear Werk,

Why don't you take a second Arduino unit, but use only one wind speed / vane unit. Use the signals for both duino's.
It makes comparing more easy.

[werk_ag]

I'm not sure if splitting the analog signal from the wind vane between two arduinos will work. Have you tested it?
And maybe there is another problem, both arduinos should read the same vane in sync, which is difficult to achieve, unless with a unit with a synchronized clock.

A careful adjust of the pot in the TX unit helps a lot in some types of bouncing characteristic of the FO wind vanes.

[AllanG]

A careful adjust of the pot in the TX unit helps a lot in some types of bouncing characteristic of the FO wind vanes.

My FO Transmitter failed completely around mid-October. I've just replaced it with the WeatherDuino TX, so using exactly the same anemometer and wind vane as before and mounted exactly the same. Unfortunately I don't have any "before and after" graph images to compare but I'm fairly certain that, having carefully adjusted the pot as Werk_AG suggests, the bouncing characteristics of the FO vane have been significantly reduced.