(Stupid ?) idea for wind speed & direction measurement

P

PaulRB

Senior Member
I've been thinking about a different and potentially simpler way to measure wind speed and direction versus the conventional anemometer and vane. I'm sure I'm not the first to think of this...

A "reed" of some kind, or a whip antenna mounted on a spring, is allowed to bend in the wind. The direction and degree of bending are measured using two detectors, at right angles to each other, a little way above the anchor point of the reed. Don't know that type of detector, could be optical, sonic, hall effect, movement sensor...

Any thoughts "spring" to mind?

Paul
 
hippy

hippy

Ex-Staff (retired)
It's an interesting idea. Perhaps a two-axis joystick pot with a large cone on it would be an omnidirectional reed in the wind.

More practical would be something hanging downwards rather than sticking upwards.
 
B

BCJKiwi

Senior Member
I had considered a similar approach but discarded it as impractical - why? (I'd like to be proved wrong!)
Well I considered the mounting which is typically on a pole which may not be fully rigid.
The antenna needs to have sufficient surface area to register low wind speeds which implies reasonable length/area = mass.
If the 'antenna' that has to be moved by the wind (or it's mounting depending on the sensing method) is flexible enough to be useful, even very minor movements in the pole will also move the 'antenna' generating false readings.
Comments?
 
Last edited:
techElder

techElder

Well-known member
It would have to be as friction free as a wind vane or anemometer so it could always return to the same zero position without springs.

I've worked with a laser-based sensor with an array of discrete detectors. The "vane" would move the laser so it pointed to a different detector.

You got a discreet measurement of both "wind" direction and "wind" speed depending on what part of the array of sensors was illuminated.

It needed to be very accurately calibrated; too accurately for the application I was working on.

Just an idea for you.
 
P

Puuhaaja

Senior Member
If I would do this kind of project I'd use linear hall effect sensor. I think you would need two of them. At the moment I'm plannin a project where linear sensor are needed but don't know yet much about them. I'm not native englis speaker so I don't understand everything perfectly. Anyways...if I understand everything right your idea on measuring is based on bending. The more flexible reed is the more it's going to bend. There are many things that have to be noticed. How long reed is, strength of material, surface area etc... The main point what I'm saying is that the amount of bending is not necessarily linear(Not sure about that) That means that bending in 4 m/s wind is not two times bigger than in 2 m/s. Accurate measuring would need math coprosessor. If there is possibility to calibrate and measure different kind of values there's always possibility to use readtable command but I'm not sure how much memory it would need, but it's possible to use eeprom memory chip...I think.

Hope this was helpful!
 
M

MFB

Senior Member
How about a simple pendulum with a 2-axis magnetometer mounted in a ping-pong ball at the end? Would give you two directions of wind induced tilt.
 
A

AllyCat

Senior Member
Hi,

You seem to be getting close to re-inventing the wind sock. :)

IMHO the main requirements of meteorologists (who may indeed be rather conservative) is for a stable (damped) and linear (speed) measurement (because calibration is so much easier). The pendulum/whip probably fails on both counts. But the wind sock is indeed a very useful indicator for its intended purpose because it has a "logarithmic" (i.e wide range) speed characteristic and gives a clear indication of direction at all speeds.

Cheers, Alan.
 
J

John West

Senior Member
BCJKiwi, I think a second of data averaging would suppress most mast vibration noise.
-
I like the idea of the sensor hanging down, too. A bit like the "weather rope."
 
S

SAborn

Senior Member
I like the idea of the sensor hanging down, too. A bit like the "weather rope."
Click to expand...
I have seen this method used for a wind limit switch, where a light ball is fitted to a metal rod/wire, with a wire loop around the vertical rod/wire, as the wind force increases on the ball it moves the rod towards the loop, if the wind is strong enough the rod touches the loop and effectively closing the circuit, this system is often used for things like solar panel trackers to monitor the wind force and park the panels in gail force conditions.

The problem i see with using a vertical pendulum for wind speed measurements is the pendulum will swing either side of centre giving false measurements, dampening is not an option as this will then cause other errors in light wind conditions below 2m/s, if all you want is some rough ball park figure on a display for periodic viewing then it might be good enough, but should you want to data log the results then it will be far from efficient in my view.

A simple way to test the theory of a vertical pendulum method would be to use a old gaming joy stick with X & Y pots, fix the joy stick upside down with the return springs removed, and attach some form of an extention to the handle with a reasonable size light weight ball on the end of the extention.

Using a picaxe to read the X & Y Pots you should be able to get some indication how well the method will work.
 
D

Dippy

Moderator
I guess a weather cock or sock is a nice simple robust solution that can withstand a vast range of speeds and gusts.

Doubtless all these pokey-uppy , dangly and venturi things would work well in certain situations.
But, in extremes, they'd probably flap around like Jeremy Clarkson's face in an Ariel Atom.
Keep it simple and robust and then it won't fall off.
And, be honest, in the Real World of houses and back gardens, why would anyone want better-than 10 degree accuracy for wind direction?

How about an oboe reed in a tube with fluid manometer.
Then it might play a tune as the wind gusted.
 
S

SAborn

Senior Member
they'd probably flap around like Jeremy Clarkson's face in an Ariel Atom.
Click to expand...
Hahaha, well said. ;)

My point was for a simple test to prove this view.
As i fully expect the same result.

But some people need to find out the hard way, and do the tests to prove the rest of us wrong.
 
P

PaulRB

Senior Member
Here's an interesting idea:

http://www3.telus.net/chemelec/Projects/Anemometer/Anemometer.htm

I wonder if the circuit could be simplified a little by using 2 picaxe adc inputs? Maybe the two voltages being compared are too close for that.

Or this one (unfortunately the scematics are missing):

http://www.ednmag.com/archives/1996/031496/06di3.htm

Perhaps the picaxe's pwm could be used to maintain the heat in the transistor.

Then there's the matter of direction... perhaps 3 transistors at the corners of an equilateral triangle, with shields giiding the airflow in such a way that the wind direction cools the 3 by different amounts.

The maths would be difficult for picaxe, so I was thinking it would do the basic measuring and pass the results to a pc or rpi for further processing.
 
Last edited:
S

skatt

Member
Wind speed can be measured with an electret microphone as wind speed will be relative to the pressure created on the mike. Only issue is it needs to be pointing directly into the wind to get an accurate measurement.Should not be too difficult to design a amplifying circuit using low cost op amps then fed into a Picaxe.
 
D

Dippy

Moderator
Wouldn't a constant wind-speed create a DC pressure on the micropohone?
What signal would you get at a constant 10mph and 20mph?
Maybe better to base something on Bernoulli's work? (all us old physicists will know it).

The heating method has been used for many years for flow measurement in fluids. Might be worth a DIY attempt. Let us know.
 
P

PaulRB

Senior Member
The main problem for me will be, I suspect, my level of competance in analog electronics!

I also have a concern that all these heating techniques will require too much current for a remotely sited, battery/solar power supply.
 
S

skatt

Member
Maybe I should rephrase that.
The mike will pick up the noise generated by the wind, the higher the sound level, the higher the signal. I used to own a handheld anemometer that used 4 microphones in opposing positions which used the microphones for wind direction as well, but did not pick up adequate noise below 5mph wind speeds.It was fairly accurate with wind speeds up to 35mph compared to a £400 Raymarine unit.
 
J

John West

Senior Member
For wind speed, (just as an experiment,) I'd try using a fixed voltage through a vertically mounted piece of resistance wire, and measure the changing current through it as compared to the measured temperature and nominal barometric pressure. That seems like a PICAXE-able project.

BTW, I looked up the price of that sold-state weather station. It was well hidden. Ouch! I'd rather buy a good electric scooter with the money, and just look up the weather info online.
 
P

PaulRB

Senior Member
I suppose I should tighten up the requirements a little.

  • Low-ish power, low enough to run on battery power with small solar panel to trickle-charge.
  • Reasonably inexpensive, common components.
  • Weatherproof, obviously!
  • Nothing difficult to build mechanically and with only average DIY skills, e.g. nothing machined out of metal. Wood and screws is about the limit!

Wind speed & direction, along with rainfall measurement, are the difficult parts of a weather station for a hobbyist to build from scratch I understand. Sunshine (light levels/UV), temperature, humidity and pressure should be easier, with off-the-shelf modules or components available that will directly interface to a picaxe. I was also thinking ERF/URF modules to communicate with a PC or Rasb Pi which would do the hard number-crunching, data logging and presentation.
 
R

rossko57

Senior Member
Hot wire devices are used in automotive engine systems to measure air input (and dose fuel accordingly). They need to be referenced to the ambient air temp, which allows measuring of air _mass_ flow ( which is whats needed there). You could then compensate for density (barometric pressure) to get air speed.
 
G

g6ejd

Senior Member
I think they are out of production.

If Ultrsonic distance sensors were more flexible, four of them could be grouped in a box format to enable both wind speed and direction to be sensed. It may be posible with the standard timed pulses, might be worth a trial. Result would be a set of vectors X / Y from which time difference of arrival (distance is fixed) was measured and which varies with wind speed and direction. This has already been done, so nothing new, just needs some coding...

Example: http://www.youtube.com/watch?v=ogpxszUfvGU

http://www.gill.co.uk/products/anemometer/windsonic.htm

BTW, by low cost, I think they mean £1000+

The process is not that difficult to do: - the sensor uses the time of flight of air (velocity) measurement.

There are four ultrasonic transducers arranged as two pairs at right angles to each other. Each pair is used to measure the component of the wind in the direction between the transducers.

For example if one pair is on a North South line and the other pair is on an East West line then the North South component VN and the East West component VE are measured. These are then combined to give the actual wind speed and direction.

Wind angle A = arctan ( VE / VN )

Wind speed S = sqroot ( VE 2 + VN 2 )

To look at how the component measurements are done we will take the example of the North South transducer pair:

A pulse of ultrasonic sound is transmitted by the North transducer and the time T1 it takes to travel to the South transducer is accurately measured. The time T2 is then measured for an ultrasonic pulse to travel in the opposite direction from the South transducer to the North transducer. These times are affected by the wind blowing along the line between the transducers. For example if there is some wind blowing from a Northerly direction, then the time taken for the ultrasonic pulse to travel from N to S will be shorter than from S to N.

The time for the pulse to travel from N to S is given by

T1 = L / ( C + VN )

and the time for the pulse to travel from N to S is given by

T2 = L / ( C - VN )

where
L is the distance between the transducers and
C the speed of sound

From this the northerly component of the wind can be found from

VN = L ( 1/T1 - 1/T2 ) / 2

and the speed of sound from

C = L ( 1/T1 + 1/T2 ) / 2
 
Last edited:
P

PaulRB

Senior Member
Hi g6ejd,

This is the kind of thing! Except the price, obviously!

Thanks for relaying all that detail. I'm confident I'll be able to manage that side of things (I literately used to do "rocket science", well, almost... There was a time when I wondered how much value a piece of s/w could possibly have if it didn't have an arctan2 function in it somewhere. Ah, the ignorance of youth). Its the bit of analogue between the sensor and the picaxe that would be my personal challenge (transistors & op-amps etc)

Paul
 
G

g6ejd

Senior Member
If you google around there are plently of simple circuits that interface to standard ultrasonic sensors, most use a CD4052 cmos logic chip, I suspect the PICAXE can do all that does. Using two SRC04 Ultrasonic sensors with a target of anything about 8" away will give the distance, then add being blown off course by wind and it takes longer, from the difference they are measuring wind speed. I have not worked out the resolution required to do this over a short distance, maybe something to think about.

The youtube link shows how the students built theirs using plastic water pipe.
 
P

PaulRB

Senior Member
Thanks eclectic. It was that thread, among others, that got me thinking about this whole problem.

If only we could buy the "bare" mechanical bits and devise our own sensors to read the positions/rates....

Paul
 
G

g6ejd

Senior Member
Well you can buy spare anenometers/wind direction units for most products on the market, then displaying the wind-speed and direction is easy on a PICAXE. Try spares for Peet Bros (about the best as no moving parts, unlike Davis that use a rotary resistance track that eventually wears out).

You can get spares in the UK:

http://www.peetbros.com/shop/item.aspx?itemid=138

The Peet Bros system is brilliant, you get two pulses out, from which you get wind speed and direction, the coding is almost trivial to get that information. Pulse one is the reference, then measure time from Pulse 1 to Pulse 2 and you have direction and measure the number of pulses per unit of time and you get wind speed, that's about 4 lines of code in a PICAXE.
 
Last edited:
P

PaulRB

Senior Member
g6ejd said:
Well you can buy spare anenometers/wind direction units for most products on the market, then displaying the wind-speed and direction is easy on a PICAXE. Try spares for Peet Bros (about the best as no moving parts, unlike Davis that use a rotary resistance track that eventually wears out).

You can get spares in the UK:

http://www.peetbros.com/shop/item.aspx?itemid=138

The Peet Bros system is brilliant, you get two pulses out, from which you get wind speed and direction, the coding is almost trivial to get that information. Pulse one is the reference, then measure time from Pulse 1 to Pulse 2 and you have direction and measure the number of pulses per unit of time and you get wind speed, that's about 4 lines of code in a PICAXE.
Click to expand...
That is a better design than the ones with 8 reed switches and 8 resistors.

Where in the uk could I buy one of those and not the full weather station? I can't find a supplier with google.
 
A

AllyCat

Senior Member
Hi Paul,

Oh dear, 20 posts in the last day - where to start?

Including all the logging, analysis and presentation, your #25 specification is a BIG project and can't be particularly cheap with a RPi, URF and ERF (just one?), plus all the sensors. Personally, I'd omit the "UV" (Index) as being irrelevant and unmeasurable for a "hobby" project (however IR is definitely worthy of consideration). But the whole topic of (useful) solar/illumination measurement is an absloute can of worms that could easily fill numerous threads in itself.

A reasonable, complete weather station to that specification can be bought off-the-shelf for less than £100, often £50. Chinese-made of course and with many limitations, but time might be better spent improving some of the known "features". For example, the vane can be considerably improved with a couple of "button" magnets and a spot of glue. ;) Incidentally, I was amused by the specification sheet linked via #28 quoting the "calibration data" to 4 significant figures. The manufacturer only claims +/-10% accuracy (and that I suspect is more in hope than reality).

But back to the original topic header. Nearly ALL the wind speed sensors mentioned here need to be calibrated against a known reference. Not easy, particularly when most don't even have a linear scale (measured value/windspeed). The (Ultrasonic) "time of flight" method is about the only directly-calibrated method and certainly worthy of consideration. Perhaps it's not even necessary to use an ultrasonic frequency as the sensor presumably will be stuffed up at the top of an outdoor mast.

The basic maths is easy: speed of sound =~ 1,000 feet/second so time of flight is 1ms for 1 foot. To measure windspeed to better than 1mph needs a resolution of ~ 0.1%, or 1us for a 1 foot spacing. That's probably pushing a standard PICaxe/rangefinder - equivalent to your robot ranging to much better than 1mm accuracy.

However, IMHO the solution is simply to compare the phase of the transmitted and received signals. That can be done with a flip-flop, a low-pass filter (one R+C) and the A/D input of a PICaxe. It certainly can be made to work (I did so 20+ years ago using the cheapest available micro) but it might be an interesting challenge to try to do it all using the PWM driver, internal comparator and SR flip-flop of a (M2) PICaxe.

Cheers, Alan.
 
premelec

premelec

Senior Member
an example of wind direction....

From a local commercial weather site.... something with integration of speed and direction would seem useful...
 

Attachments

You must log in or register to reply here.
Top