I'd like to build an autonomous remote wind station + camera that I can install near my favorite kitesurfing spot so that I can

  • Know in "real-time" both speed and direction of the wind
  • See what's going on there (it's mainly to know whether the spot is crowded)

As I have little knowledge of electrical engineering, I'm asking here for some help designing this remote station.

Here is what I've come up with


  • Learn - I'm doing this to learn new things
  • Cheap - I do not exactly have a budget, but price will be an important factor
  • Autonomous - It has to work 365/24 without human intervention

Main subjects

  • Microcontroller
  • Communication
  • Power
  • Sensors (camera, wind speed, wind direction)


It has to be programmable, so I can update/improve the behavior of the station if needed.

Any programming language will do - I'm a software engineer, that's what I do ;)

I was thinking something like the Arduino, but I have no idea which model is the best fit for this station.


As the spot is 50km away from my home, I've found that the only communication protocol/network usable is GPRS/GSM (please, correct me if I'm wrong).

Sending the values read from the sensors (wind speed + direction) can easily fit in a SMS. I've found a SMS gateway that let me receive SMS for free. It appears to be exactly what I need.

But sending an image (even a 320x240) every minutes (or less) requires a data connection. Indeed, transferring 320x240 = 76 800 pixels will take about 590 SMS (at 140 bytes per SMS minus 10 bytes for the overhead).

Again, there's way too much choice here. Do I need a dualband, a quadband or a 3G/GPRS shield?

Note: I like the 3G/GPRS one as I can easily attach a 7€ camera. But 140€ is kind of expensive.

Note II: I am aware that using GPRS/GSM/3G requires a SIM card with unlimited SMS and/or data. But in France, they are pretty cheap (~15€ per month)


Solar powered appears to be the best (only) choice.

This solar module looks like what I need but I have no idea if that is too much or not enough?

  • Is there any way it can be cheaper?
  • Have you got any advice regarding the installation of the solar panel in order to maximize its efficiency?
  • Is there any warnings regarding the rain?



Apart from the cheap camera (which goes with the 3G/GPRS shield), there are plenty of cameras available.

Again, way too much choice here. I don't even know where to begin... What do I have to consider?

Wind Speed + Direction

I've found these weather meters and even a tutorial (in French) with some help regarding how to connect the sensors to the board and read both values.

There is also some code for Arduino.

This appears to be the easiest part (at least for me) of the station.

Reading values from the Wind Speed sensor is as easy as incrementing a counter and reading values from the Wind Direction sensor is as easy as doing a modulo.

Some other questions

  • How do I prevent the station from being struck by a lightning?
  • What is the best material to use for the support? Wood? Steel?
  • What kind of enclosure can I use to prevent the station to heat too much?


Note: I plan to update this post with your answers and hopefully a full tutorial on how to build it

  • \$\begingroup\$ This question is much too broad for our Question & Answer format. Try breaking it down into smaller parts. \$\endgroup\$ – Toby Jaffey Aug 13 '12 at 9:06
  • \$\begingroup\$ Hi - It is true that the question is to broad. You need to ask a specific engineering question. I am very interested in your project - I am working on something like this already. Click on my profile and contact me if you want. \$\endgroup\$ – Piotr Kula Aug 13 '12 at 9:16
  • \$\begingroup\$ @ppumkin: Check you e-mails ;) \$\endgroup\$ – ZogStriP Aug 13 '12 at 12:01
  • \$\begingroup\$ Are your calculations for the camera data for color images? Maybe you using b/w images is enough? \$\endgroup\$ – 0x6d64 Aug 13 '12 at 12:03
  • 1
    \$\begingroup\$ @ZogStriP I would recommend breaking up your project into several separate questions. As it stands now, each one of your sub-questions would make good question for the site. For example, you could ask (as separate questions) "What features and bands are necessary in a GPRS module for SMS messages?" and/or "How do I calculate how much solar power I need for a system?". \$\endgroup\$ – W5VO Aug 13 '12 at 13:09

Data transport:

If there is an internet connection within a few km you may be able to access it without vast cost. "Cantenna" type aerials give very substantial gain at very modest cost. You may be able to enthuse some locals with your project to the extent that they will let you use their internet access via a WiFi connection. A range of say 1 km line of sight can usually be obtained quite easily, up to 10 km with some care and maybe 20km plus depending on aerial heights obstructions etc.

SMS: I suspect that even an "unlimited" SMS plan will not allow the data rate you want. They probably have a "fair use" clause in the fine print.

Data: You say "SMS or data" - if you have unlimited data you can probably achieve what you want. Your 76,800 pixel per frame will translate to 100 kiloBUTE or more. You have allowed one bit per pixel - OK only for sending monochome black and white on/off pixel images. Even a byte per pixel will be on the low side without encoding. Fortunately, coding and compression is your friend.
Using JPG compression on a randomly chosen image sized to 320 x 240 I got
JPG_Quality / Size kB
60 ... 61 kB
70 ... 63 kB
80 ... 67 kB
90 ... 73 kB

ie compression does not affect data size vastly in this example. In some it may but chances are a files size in the 50 kB - 100 kB range will be OK.

AND: For your application , if you are willing to put modest processing power at the transmit end, you can just send "differences" in the image when they occur.
This will vastly decrease you mean data rate as, most of the time, there will be little new to send if the beach is empty. You'd want to exclude waving tree branches and rolling waves from the areas considered dynamic :-).

Realtime wind speed and direction: That depends on how real you want the real time (1 second, 10 seconds ...?) and what resolution you want. If you were happy with 5 kph increments in wind speed you can handle 75 kph in 4 bits (16 steps of 5 kph starting at 0). By using logarithmic compression or an arbitrary scale you can get greater top end velocity in 4 bits. But 8 bits will handle all you sensibly need (eg 0-255 kph in 1 kph steps ! :-) ).
4 bits: ...0 1 2 4 6 8 10 15 20 25 30 40 60 80 100 BIG! or eg ... 0 1 2 5 10 15 20 30 40 50 60 80 100 125 150 Die
Or to suit.

For wind speed and direction SMS may suffice if you don't mind having the data with good resolution but poor "latency".
eg say you have 120 bytes/SMS available.
Assume 1 byte each for speed and direction . 120/2 = 60 data pairs/SMS.
That's an hours data in one SMS with 1 minute sampling once an hour. Or A minute's data with one scond smpling sent once per minute.

One SMS/minute = 1440/day - they may have something to say.
One SMS/hour = 24/day = likely to be acceptable.

Your example PV panel seems very expensive for what you get.
Slightly more DIY effort will give you substantially more energy for less $.

The offered panel appears to be amorphous silicon or may be CIGS and just may be monocrystalline silicon.
Don't buy amorphous silicon - efficiency is low and output decreases with time more rapidly than crystalline silicon. (Stabler Wronski also feature but it's bad without them :-) ).

If the offered panel is CIGS or CdTe ask re GUARANTEED lifetime with full replacement if it fails - early ones have poor lifetime.
Best = mono-crystalline silicon. Glass panel with Al frame and heat laminated in industry standard manner will give 20+ years of operation. I have a still working 30+ year one.

Available Watts from a PV panel

How to guesstimate the mean wattage available OUT of battery for a PV panel + battery system.

Wout ~~= Wpanel x SSH / 30 .... 1

  • Wout = continuious average available power.
    Wpanel = panel max watts properly loaed in full sunlight
    SSH = Hours of equivalent full sunshine per day.
    The factor of 30 is a guesstimated constant based on their being 24 hours in a day and allowing for PV to battery losses and battery storage losses.


Wpanel ~~= Wout x 30 / SSH .... 2

Mean daily Sunshine Hours in Toulouse can be found here and is as little as 1.2 hours per day on average in December and 5+ hours/day from May to August.

So in mid winter if you want 1 Watt continuous your PV panel will need to be

Wpanel = Wout x 30 / SSH .... applying 2 above.
= 1 x 30 / 1.2 = 25 Watts!

And your 3 Watt advertised panel will supply
Wout ~~= Wpanel x SSH / 30 .... applying 1 above. = 3 x 1.2 / 30 =~ 120 milliWatts continuous.

More anon ...

| improve this answer | |
  • \$\begingroup\$ @RusselMcMahon: Thanks for this great answer! I'll study that! \$\endgroup\$ – ZogStriP Aug 13 '12 at 19:22

Not the answer you're looking for? Browse other questions tagged or ask your own question.