I have helped several of my students design and build a
variety of windspeed circuits.

The typical land-based wind anemometer consists of an
upright shaft with at LEAST three arms (usually four)
to which are attached cups. The cups we usually use are
derived from plastic egg shapes that are sold for the
purpose of putting small prizes inside. These are
easily found in many stores around Easter time, but they
can be purchased any time of the year at many novelty
stores. We do not use the perfectly round end of these
eggs, but rather the pointy end, as it works just a bit
better as it is more aerodynamically shaped.

The radial distance to the cups influences the speed
of the shaft and the minimum wind speed it will run at.
The larger the radial distance, the more sensitive
the system is to slow speed winds, but this also
reduces the speed of rotation of the shaft at any
given wind speed. To measure really high wind speeds you
want a shorter radial distance, but then the system is
not so good at low wind speeds. Some systems involve
two separate sets of shafts/arms, one for measuring
slow wind speeds, and the other for measuring high
wind speeds.

The upright shaft is usually supported by TWO ball bearing
assemblies. We tried using a single bearing, but the
sideways forces caused by the wind militate against this.
The two bearings are best separated by several inches
so the forces can be distributed properly. A small
collar on the shaft fixes the height of the shaft. A
cheaper method is to drill a small hole through the shaft
and drive a small brad or nail through the hole and
then trim the ends of the brad or nail. If you plan
on measuring hurricane force winds, then the shaft
must be fixed so that it will not "ride up" in heavy
winds.

Some of our anemometers used small DC motors as the sensors.
The more poles the motor has, the better the sensor it makes.
I have found that small tape cassette motors work well, as
they have a large number of poles, and generally low
friction. Note that some cassette motors use a governor
on the rotor. If the motor is spun faster than its rated
operating speed, then the governor will disconnect one
of the wires (usually the red or hot wire) from the coils.
If you find that this happens to you, then open the motor
carefully and solder the governor shut.

You can mount the motor to the shaft using a pulley
or directly. To mount directly we obtain 1" aluminum or
brass rod that is thicker than the shaft. We then bore
a small hole all the way through this piece of metal.
This will mate with the motor shaft. Then we bore a hole
to accept the anemometer shaft about halfway through the
piece of metal. We hold the shafts in the adapter using
heat and press-fit if the holes were slightly smaller than
the shafts, or we use shims or setscrews, or a dab of
epoxy glue to set the shafts in place, depending on
how much time and energy you want to put into this aspect
of the system.

Because it is often hard to get the two shafts to line
up exactly using the tools we have here in the electronics
shop, we often find it useful to connect the two using a
short flexible shaft. In this case coupling adapters are made
for both the main shaft and the motor. We often use a piece
of thin coax cable or tubing as the flexible part. Thick
multistrand wire is also useful at times.

****
Sometimes we do away with using a motor as a sensor, and
instead we use a pulse sensor. We make a circular design
with radiating and alternating clear/black bars. We make
this using AutoCad and print it on our laser printer.
Sometimes we print it on overhead transparency material,
and sometimes we have our graphic arts department make us
up a film negative of the design. We then glue the
design to a sheet of clear plastic (Lexan/Plexiglas) and
then machine the piece of plastic so it is round and
can be mounted on the main shaft.

A standard emitter/detector light sensor is mounted so
that it can see the alternating lines.

***
The units can be calibrated in several ways. We can take
them over to a place where they have a commercial anemometer
and gather calibration data.

We can strap the thing to the top of a car and travel at a
known velocity IN BOTH DIRECTIONS. The AVERAGE reading
will be equal to the speedometer reading, as any additional
wind speeds will have been canceled out by this process.
This assumes that the wind speed and direction was fairly
constant during the two runs, and that the two runs were
done along fairly straight lines.

*****
Other methods we have used for measuring wind speed include:

Building a cylindrical housing 5" in diameter from PVC pipe.
A 1/4" hole is then drilled through the side at the center
of a 12" section of this pipe. A short plastic tube is
attached to this hole. As the wind blows into the PVC pipe
a suction will be developed at the small tube. We then
measure the suction by measuring how much a fluid with dye
in it rises due to the suction. Barometric effects do not
come into play because both ends of the system are
essentially unsealed. Which is the weak point of this
system, because you have to ensure that evaporation is not
changing the amount of fluid that is available.

This method also requires that the wind be blowing
directly down the PVC pipe. We accomplished this by
making a large weather vane that keeps the pipe aimed
right into the wind (at least once the wind is at a
moderate level)

***
Hot wire anemometer. The resistance of a wire is affected by
its temperature. There is a "wind chill factor" involved,
so you actually have to maintain two sensor wires, one of
which should be shielded from the wind but allowed to
experience the same environment minus the wind. If the wind
is zero, then both would read the same resistance. The greater
the difference actually measured, the greater the wind must be.
It is called a HOT wire anemometer because the temperature of the
wire is elevated above normal so that the cooling effects
can be magnified.

In general a constant current is shoved through each wire, and
the voltage drops are proportional to the resistances.

***
Friction plate. This is a flat plate of metal that hangs down
due to gravity. When wind blows against it, the angle of deviation
is related to the wind speed. Highly non-linear. Must be aimed
right at the wind (needs to be mounted to a weather vane).
Tends to oscillate. You have to measure the angle. Non-linearity
tends to follow the tangent function. That is, it would take
an infinite wind speed to make the plate tilt 90 degrees. Has some
good points, as the system is most sensitive to slow wind speeds,
yet can easily tolerate hurricane winds.

****
Other methods we have thought about but not implemented:

Measuring the lift developed on an airfoil. This method
will most likely not be very useful at low wind speeds.