Experiments towards a differential air pressure based wind speed sensors

Den torsdag den 28. maj 2015 kl. 22.08.07 UTC+2 skrev Jan Panteltje:

you might be able to do it with three in a triangle

I like the acoustic windsensor, three ultrasonic transducers in a triangle measure speed of sound to and from each of them and calculate wind speed and direction from the differences

-Lasse

[snip]

Interesting! You might actually make up a ring oscillator of the ultrasonic transducers ;-) ...Jim Thompson

-- | James E.Thompson | mens | | Analog Innovations | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | San Tan Valley, AZ 85142 Skype: skypeanalog | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at

| 1962 | I love to cook with wine. Sometimes I even put it in the food.

Den torsdag den 28. maj 2015 kl. 22.45.37 UTC+2 skrev Jim Thompson:

I'm not sure I see how that would work, afaict you'll need to measure the speed in all directions to remove the variable speed of sound

-Lasse

Without running the numbers... I think there will be an effective vector with speed dropping out?? ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| San Tan Valley, AZ 85142     Skype: skypeanalog  |             | 
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  | 
| E-mail Icon at http://www.analog-innovations.com |    1962     | 
              
I love to cook with wine.     Sometimes I even put it in the food.

People have been making airspeed measurements that way for years, but it gets difficult at low airspeeds, like below around 20 Kts.

Have you thought about the effects of temp drift on the zero offset of the pressure sensor ?

--
Regards, 

Adrian Jansen           adrianjansen at internode dot on dot net 
Note reply address is invalid, convert address above to machine form.

Ultrasonics is the way it's done in most "no moving parts" type of weather stations. etc... They're especially useful in freezing environments, where ice on the turbine and wind vane are a problem.

Your pitot tubes will probably work in a storm, but might be problematic at low wind velocities. Even if you do get it working, you have a hole in the device which could easily fill with airborne dirt, and wreck the measurement accuracy.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

More... I think this is the OEM source for the ultrasonic wind speed/direction sensores used on some commercial wx stations: How it works:

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

On a sunny day (Thu, 28 May 2015 13:38:13 -0700 (PDT)) it happened Lasse Langwadt Christensen wrote in :

Yes

I already out-did the pitot in previous experiments with acoustic a 44kHz, and also looked for the travel time of puffs of ionized air created by sparks. and posted some results here, and code.

Although a pitot is very simple, I think it is extremely directional. You would need many to get wind direction, and it is big, AND it can get clogged up, for airplanes icing is a problem.

These little pressure sensors are sort of chemical proof, not very directional, and extremely small.

By putting front and rear in series (either in hardware or software) you should get a zero signal if no wind, and positive in one -, and negative in the other direction. Same for left right.

3 would make computation more complex. And these cost next to nothing.

Note I do not use N S and E W, an for the steering will subtract the magnetic compass angle. I dunno if you ever looked up the price of nautical wind meters, especially the acoustic ones, and compare that to the 12 $ I have spend so far, that by itself makes an interesting project,

Self cleaning, bit of rain...

And it is experimental of course, just curious. First idea was cool, then why waiting for the things from ebay to arrive I did some math, looked terrible (small signal), but in reality it is just about enough signal it seems.

So.. next will try the 2 in series, interesting by itself, as these are bridges:

Left Right sensor sensor ---------------------- + | | | | R R R R 5k -| |+ + | |-

- out ----| |-------| |---- out + | | | | R R R R | | | | --------------------- -

Connecting these low impedance points together, what is the effect?

Takes longer to write his than to try it :-) If that for some reason does not work I will use 2 x HX711, or 2 of the channels of one HX711.

Experimenting is fun. that is how the steam engine was discovered IIRC.

And Graham Bell's telephone started working because of some error, contacts burned together.

On a sunny day (Fri, 29 May 2015 09:16:58 +1000) it happened Adrian Jansen wrote in :

Yes, see my reply to Lasse, But it seems there is still a very usable signal at very low air speeds.

Yes, first I thought it was the temperature of my breath that 'did it', caused the signal. That was easily tested by applying some pressure, and seeing the same curves. Anyways any temperature effect can be compensated for in software using some LM335 or similar sensor.

Also I do not care too much about the difference between 3 and 4 meter / second. if i can get this to work on the standard wind-scales it is fine with me:

force km/h

1 < 1 2 1 - 5 3 6 - 11 4 12 - 19 5 20 - 28 6 39 - 49 7 50 - 61 8 62 - 74 9 75 - 88 10 89 - 102 your gone anyways... 11 102 - 117 12 > 117

So we are not taking parts per million here :-)

Errata: force km/h 0 < 1 1 1 - 5 2 6 - 11 3 12 - 19 4 20 - 28 5 29 - 38 6 39 - 49 7 50 - 61 8 62 - 74 9 75 - 88

10 89 - 102 your gone anyways... 11 103 - 117 12 > 117

So we are not taking parts per million here :-)

Do not remember who mentioned it here, but why not try it I though:

Got some of these absolute pressure sensors (barometer):

Wrote some PIC code to send sensor values over serial link.

Subtracted front from left to get direction...

Testing showed ever changing values... Why?

Then I had a guess: these sensors have windows that are transparent.

Took a LED flashlight and can change the difference between 2 sensors by a factor 10... At least...

So paint it black....

measuring in three directions should be enough,

with six measurements the system would be over constrained which makes for harder math, and potentially higher precision.

--
umop apisdn

Have you considered this?

I've used a pair of these with a PIC18F14K22 for differential pressure measurement. In my application I needed a tube connector. I did it by cutting the end of a 10ml plastic syringe and gluing it down over the BMP180. Works a treat!

On a sunny day (Sun, 31 May 2015 12:58:42 +0100) it happened Pomegranate Bastard wrote in :

Yes I have seen those, I have the SPC01 module from ebay, something like this:

that is also used for altitude, and magnetic compass.

Problem what happens when water gets into the small hole?

At least it is not light sensitve like my pressure modules...

Not a problem if you cover it with the cut-off end of a 10 ml syringe, which fits snugly over the BMP180 module. Put a bead of hot-melt glue between the syringe and PCB to give a perfect seal.

You can use any length of tubing to keep out water (perhaps with an in-line water trap) and keep the electronics remote from the pitot or venturi head or whatever you intend to use.

These things are extremely sensitive. When used as an altimeter, I can easily detect a height change of a metre.

On a sunny day (Sun, 31 May 2015 14:40:56 +0100) it happened Pomegranate Bastard wrote in :

yes, i have some of that stuff, use it ot put distilled water in super sealed battery..

Yes, same with het SPC01 module I think it has the same chip, or close. But in the module the math (for altitude, and also weather report) is done in a (closed source) PIC, and it is slow (one measurement per second).

If things do not work out with the other sensors (now writing PIC software for those transparent sensors using a tangent lookup table duh...to get angle from pressure ratios without math library.. things one tries... then I may try those. The price is good :-).

atan table is interesting... infinities that are not.

I'll give you my software if you want it.

It's for a barometer/altimeter I made to see how good the BMP180 is and written in Knudsden CC8E. Probably a lot more than you need but it's yours if you want it. Let me know if you do and I'll email it to you.

On a sunny day (Sun, 31 May 2015 16:50:28 +0100) it happened Pomegranate Bastard wrote in :

OK, I did some tests.

First I spray-painted the little transparent sensors I have with black car paint. Those prove to be light sensitive even then, although less. I think when the paint has hardened (I used lot) this will affect the sensitivity.

I tested the BMP085 (see below) with a VERY strong light source shining into that little hole, and cannot see any significant effect (pressure or altitude).

So last night I ordered some of those BMP180 boards from ebay.

After reading that datasheet I found the BMP180 is an improved version of the BMP085 I already have on that PIC board. Bosch makes the complete C source driver available as zip file:

and, as it turns out, I already tried that (very similar code) to test the BMP085. I did that by removing that PIC from that SPC01 ebay board, and directly accessing the i2c lines from a PC par port, my usual way to test new i2c chips, so easy on the PC, well these days I use Raspi GPIO, more I/O.

Their code looks written in a way bad enough to make anybody scared of C language, I think K&R would cry... So I rewrote the whole thing one day. I write a lot of code, and had forgotten about that one.

I do not know 'Knudsden CC8E' but chances that it is better readable than the Bosch code are very big. I mean looks like they just reformatted it to 40 characters leaving out all spaces too.

So, yes, I am interested.

One question: Do those chips come with the calibration values set, or do you need to calibrate on some known pressure? Other question: What C compiler did you use to create the PIC 18F14 code? Or was that asm?

Sorry, I mis-spelt "Knudsen"

This links to the schematic, the C code and the MPLAB files.

There should be enough there to get you started. If not, let me know. I use MPLAB with an ICD3.

All chips are individually calibrated. The calibration data is contained in registers 0xaa to 0xbe inclusive.

If you use a pair of BMP180s, you may find a **very** small offset and gain difference between the two. All I did when I used a pair was to use one as a reference. I set the offset for the second BMP180 at atmospheric pressure, then applied a vacuum and set the gain for the second.

Here's the link to CC8E

I've been using it for years and I love it! And there's a free edition which will be ok for most purposes. I'm not sure, though, if the free version handles 32 bit floats. If it doesn't 24 bit floats are nearly as good.