Could I calibrate a homemade anemometer by simply mouting it to the roof of my car and controlling the speed or would I get false reading because of air currents over my hood, what about the side of the car or the front of the hood?
The airflow over the car can vary. (Remember that a number of muscle cars had the hood scoop open at the base of the windshield, because that’s where there was a high positive pressure zone. Most newer cars are a little more aerodynamic but overall you won’t get clear air for at least two or three feet away from the sides and maybe twice that from the roof.
I’d mount it on a stick at the front of the car, about four feet up, and do at least four runs down the same stretch, up and back, and average the readings to cancel any minor wind effects.
Be sure to wear your clown nose while you do it, so the kids pointing and laughing will enjoy it properly. 
If I see a linear climb in volts as the speed goes up could I consider this to be a constant if I only test to about 40 mph, we get gusts on the salt flats up to 100mph at times and it would be kind of cool to be able to register those winds. From a working standpoint accuracy up to about 35 mph is really all that is needed.
Cup shape is another issue I am a little concerned with, the models I have looked at are using something like 1/2 egg shaped cups. I have plastic bowls with flat bottoms I would like to use, also considering the bottoms of plastic water bottles.
WTF?! A honey badger that actually cares about something? And the accuracy of a tool for a notable imprecise science, no less.
Sorry, but somebody had to do it, and it’d be better if it were somebody able to laugh at themselves.
I take it you’re using a small DC motor as the sensor? Not sure how linear that would be. I’d use at least four magnets passing a hall-effect sensor and see if the response is reasonably linear up to about 40MPH. From there, I’d bet on something close to linearity up to 100+.
Using a 12v computer fan motor, 4 wires so it may have 4 magnets, all 4 wires seem to give me a response.
A bazillion years ago I did just this. At the time my parents had a small sports car with a sun-roof, and we poked the sensor out of the roof as far as we could, to get past the flow over the roof. I’m pretty confident we were in undisturbed air.
I was quite astounded at how linear such a pretty simple design turned out to be. I made the cups out of toilet cistern float balls cut in half (the attachment point for the float arm was perfect for mounting the cups onto a carefully bent aluminium rod.) The whole thing seemed to work very well, and I am pretty confident that the limiting factor in the calibration was the car’s speedo.
In order to be able to measure vehicle fuel economy on a chassis dynamometer, automotive engineers first perform a “coast-down test” in which a vehicle on a flat, straight outdoor track is run up to 70 MPH, put in neutral, and then timed as it decelerates a certain amount. An anemometer is used to account for ambient winds. If a stationary anemometer is used, then the test is only valid for certain measured wind conditions. If the anemometer is affixed to the test vehicle, then the test is considered valid over a wider range of wind conditions. But here’s the catch: regulations dictate that the anemometer must be positioned six feet in front of the front bumper on order to assure that it gets an accurate reading.
You can probably relax from these requirements a bit and still get reasonably accurate readings. There’s clean air to be found a few feet to the side of your door or above the hood/roof, but the EPA regs don’t allow something sticking far up above the hood/roof because that will significantly affect the overall drag on the vehicle. You don’t need to worry about that, since you’re not measuring vehicle drag, so you might find it easier to mount something above your hood or roof. I’m inclined to agree with Amateur Barbarian: you’ll want to put your anemometer at least a few feet away from the car in order to find undisturbed air.
I don’t know what modern non-pro anemometers use, but the magnetic coupling of a motor and voltage generated might not be the most linear signal. A mechanism that counts the actual number of revolutions, with as fine a granularity as possible (4 points per rev minimum, 20 or more better) would be more accurate and reliable.
I have a speedometer on my modified Mustang that works just that way. It uses an electronic Autometer speedo (set in the original cluster, BTW) and a hall-effect sensor that reads four powerful magnets epoxied to the drive shaft. It’s around 2500 pulses per mile. The speedo is adjustable to almost any range of pulses. You could build a very slick anemometer with one of these (which are admittedly a little pricey) and a sensor that had perhaps 10 magnets per revolution.
It’s also easy to use digital electronics or a computer input to read and average the pulses and give a speed reading.
Not as easy as a voltage-generator hooked to a voltmeter, but… 
GPS. That is all.
The suggestion to run forward and back across the same track and average the readings is a good one, but you really need to do it on a completely still day. If you have any crosswind it will be additive (vectorialy, but will increase the final answer) whichever direction you go.
My dad bought one of these not too long ago. It uses a Cateye Velo 8 bicycle computer. Not sure, but I think the sensor involves a magnet and a reed switch, though I suppose it might also be a hall-effect sensor.
[quote=“Amateur_Barbarian, post:9, topic:695461”]
I don’t know what modern non-pro anemometers use, but the magnetic coupling of a motor and voltage generated might not be the most linear signal. A mechanism that counts the actual number of revolutions, with as fine a granularity as possible (4 points per rev minimum, 20 or more better) would be more accurate and reliable.
I have a speedometer on my modified Mustang that works just that way. It uses an electronic Autometer speedo (set in the original cluster, BTW) and a hall-effect sensor that reads four powerful magnets epoxied to the drive shaft. It’s around 2500 pulses per mile. The speedo is adjustable to almost any range of pulses. You could build a very slick anemometer with one of these (which are admittedly a little pricey) and a sensor that had perhaps 10 magnets per revolution.
It’s also easy to use digital electronics or a computer input to read and average the pulses and give a speed reading.
Not as easy as a voltage-generator hooked to a voltmeter, but…
That does sound like a better way to go
A non-linear measurement device isn’t necessarily a problem, as long as the non-linearity is repeatable and can be calibrated out.