Cumulus Support

Yes, the F.O. anemometer rotor is very light though the magnet makes it unbalanced so that places a slightly higher load on the bearing. And certainly good quality bearings should be good for billions of revolutions. Of course this AWS is built done to a (very low) price so would expect the cheapest bearings available.

This looks like a good replacement :-http://simplybearings.co.uk/shop/p2603/ ... _info.html Not that I've needed to replace one - my station has bee running fine and the bearings are still in perfect order.

Now! if I have this right. It works out to be 1.35/1.39 meters of wind run per revolution.

therefore my 2010 wind run was 62,744,000m / 1.3625m (Average) = 46,051,229 revs

Thanks for all the effort on this one.

Resurecting an old thread with a bit of new information.

Like Gina, I'm developing my own weather station (now that my Fine Offset is ready for the bin). I started with the FO anemometer, and tried to derive how it would work from first principles. The result closely matches Gina's empirical data, so I thought I'd share my thinking.

The FO anemometer produced 2 pulses per revolution. The distance from the center of the cup to the center of the anemometer is 6.8 cm. This is the radius of the circle a cup travels, so for a complete revolution, a cup travels 2*PI*R, or 0.42 m. (Gina worked this out to be 0.39, so we basically agree)
According to wikipedia, the most common anemometer factor is 3. It was originally thought to be 3 for all anemometers of this type, but later work showed the number had a variance based on a complicated formula. If we assume it's 3 for the FO, then 2 Pulses per second = 1 revolution per second = 0.42*3 = 1.26 meters per second or 4.54 kilometers per hour or 2.81 miles per hour.

A test against Gina's table:
20 PPS = 30 MPH (Gina) or 28 MPH (Charlie)

So, bottom line, if you use the FO anemometer, but build your own electronics to measure wind speed, the anemometer factor is 3. However, if this difference is too large for you, you can exactly match Gina's result using a factor of 3.2, which is in the range of possibility... of course, so is a minor error introduced by measurements and / or rounding.

Charlie wrote: 2 Pulses per second = ... [snip] ... 2.81 miles per hour.

Hi,

I've been doing some similar calculations but from the "other end", i.e. how the FO station processes the pulses and thus what data is put into any logs produced from a FO station.

It seems established that the anemometer (firmware) divides the 48 seconds between radio transmissions into 24 x 2-second intervals and counts the number of pulses in each. So the minimum is one pulse in two seconds, displayed as 0.7 mph which correlates rather well with Charlie's 4 pulses and 2.81 mph.

However, if one looks further up the scale it appears that FO are actually using 3.0 mph for 1 rev/second (e.g. 8 pulses displayed as 6.0 mph) but there is some rather dodgy "rounding" (probably because FO are calculating/rounding in another scale such as km/hr and then converting to mph) since the incremental steps are anything between 0.6 and 0.9 instead of the expected 0.7 - 0.8.

The reported peak (gust) value is obviously the highest of the 24 measured values, but it's less clear what FO are doing for the "average" value. One would expect them to add the 24 values and then divide by 24, but the value is still only reported to the same basic resolution (0.75 mph nominal steps) as the gust value. Furthermore, with an absolutely constant "wind speed" (driving the head directly from a synchronous motor, or simulating with an electronic pulse generator) the average always appears to be one "step" less than the peak. This is perhaps to be expected, because generally some of the 2 second intervals will have "n" pulses and others (n+1). The peak is obviously (n+1) but it does imply that FO are truncating rather than rounding the average value.

However, I suspect that FO might not actually add and divide the 24 values, but use some form of "rolling average". Various tests I've done suggest that the reported "average" value depends on when a gust occurs in the 48 second interval, and that on average the reported speed is significantly (maybe 20%) less than the true average.

More details later, perhaps in my "wind run" thread.

Cheers, Alan.

Sorry to resurrect an old thread, but if 1 revolution is 2 pulses per second is 2.81mph, then a wind speed of 100mph (I hope never to experience that, but I guess it's possible) is about 35.6 revolutions per second. or about 71 pulses per second. Or a pulse every 14ms or so.

What's a typical debounce time on these reed switches? I thought 15ms was considered safe, but that means no accurate windspeed measurements are possible above 90mph or so. Is that just the way it is?

Hi Scott,

AFAIK the bounce time of a reed switch is of the order of a millisecond. But anyway, I expect the microcontroller firmware uses either edge-triggering (hardware) or polling (software), both of which can be largely resistant to contact bounce.

I tried "emulating" the anemometer hardware with a transistor driven from a frequency/pulse generator and IIRC managed to record and display "gusts" up to about 200 mph! However, the average seemed to stall at slightly under 100 mph, which suggests that it's actually the "maths" in FO's software which is the limiting factor. The software does NOT appear to simply count the number of pulses/revolutions over 48 seconds to calculate the "average windspeed".

Cheers, Alan.

Did I get it? 50 km.h the wind is 20 pulses per second. I wanted to use to build a wind sensor anemometer. bicycle computer + wind sensor offset from the station. I need to know how many pulses per second is the wind 50 km.h? In my the sensor has 2 pulses per revolution
Do gust of wind is measured every 2 seconds or every 48 seconds?

Hi,

Yes, for the FO anemometer. But the value will depend on the physical properties of the hardware (dimensions, shape of cups, etc.). Note that the cups do NOT move at the wind speed, but at about 30% - 40% of the wind speed.

The transmitter's microntroller measures the number of pulses in each 2 second period for 48 seconds and then transmits (to the Console) the highest (gust) and the "average" of the 24 values.

Cheers, Alan.

I tray this .
http://www.youtube.com/watch?v=fUJNL7wW ... e=youtu.be

I pulse per second is ...m/s ?

1 pulse per second would be, give or take, 0,61 m/sec.

Hi,

On the previous page :

Gina wrote:4 pulses per second = 8 in 2s 8 * 0.34 m/s = 2.72 m/s

The Console appears to display exactly 1.5 mph/pulse/sec which corresponds to 0.67 m/s. But bear in mind that the manufacturer's specifications (which are generally rather optimistic) only give the accuracy as:

+/- 1 m/s (if wind speed < 10m/s)
+/- 10% (if wind speed > 10 m/s)

Cheers, Alan.

Thanks !
Do you know how reed sensor is used in the wind?
I mean the parameters of the maximum short circuit / s

If the station measured wind every 2 seconds and doing 24 measurements within 48 seconds, and then sends the highest gust of why the experiments which he conducted a simple anemometer showed któtkotrwały gust of wind and the station did not detect it (say, the 2-3 second gust and stop the blades ). ? Does anyone have proof that the electronics station measures wind every 2 seconds, 24 times for 28 seconds?
I'm so anxious to find out exactly how the station measured wind gusts, because in the last windy weather seems to me that the station is "lost" strong gusts that occur during those 48 seconds.

Hi,

I believe that the information came from the manufacturer.

For fast "gusts" you may need to consider the mechanical intertia of the rotor, and of course a "2-second gust" might fall across two separate 2-second "measurement windows". You probably will need to do your own experimental testing.

Similarly, I thought the calibration of my station might be poor, but have not (yet) managed to prove any major errors. However, it does seem that the "average" is NOT the sum of 24 individual values divided by 24, but is calculated using some type of "rolling average". Using a pulse generator, I emulated "gusts" of 200mph (which I believe were reported correctly on the Console), but the displayed average had errors (not the "--" claimed in the User Manual) above about 95 mph.

Cheers, Alan.