Wind sensor using car reversing kit

A wind speed and direction sensor from an Ebay car reversing kit and an Arduino

Public Chat
Similar projects worth following
Use an Atmega328 and a £8 Ebay car reversing kit to make a highly accurate ultrasonic wind speed and direction sensor.

Testing the prototype using a household fan to provide the wind. Wind sensor is mounted on a turntable. The Arduino output goes to ESP32 via serial port. ESP32 serves real time web page showing the data in graphical form.

The car reversing kit  is available on Ebay for under £10.

Video showing what is in the box and initial thoughts on how to hack the control box pcb for the wind sensor.

The PCB inside the control box

The Arduino Pro mini in place on the car reversing sensor pcb.

Video about how the Arduino was attached to the car reversing kit pcb

The prototype was designed to be  as simple as possible to make. A 210 mm diameter disc of ply with holes for the transducers at 150 mm diameter (cut out using the hole saw supplied with the kit) and 10mm hole for a central spine to support the 65 mm diameter "echo plate", at the top.  Spine was 10mm carbon rod, but steel or wooden dowel would work just fine. Vertical distance from sensors to echo plate around 250mm.

  • 1 × Car reversing kit A car reversing kit consisting of 4 ultrasonic transducers, a control box and a display.
  • 1 × Arduino Pro Mini Used to drive the transducers and time the pulses
  • 1 × Transducer mount A disc of plywood around 210 mm diameter with 4 holes at 150 mm diameter for the transducers and a 10 mm hole in centre for the spine
  • 1 × Spine 10 mm dia rod x 300mm.. Wood dowel, Carbon or stainless steel. The spine goes through the centre of the unit and keesp the echo plate at a set distance from the base
  • 1 × Echo plate The echo plate is what the signals bounce from. Just a ply disc around 50mm diameter with a 10mm hole in centre for spine

View all 11 components

  • Add ESP-01 to the hacked car reversing kit PCB.

    Andy10/12/2021 at 10:06 0 comments

    Update the add on board with ESP-01. Once added to the board Initially I found that ESP-01 had difficulty connecting to the network. I suspect that it is due to noise on the hacked PCB , since connection works fine, if the Arduino is held in reset.  That being the case, I solved the problem by connecting the ESP-01 GPI0 to the Arduino RST input ( via suitable level shift). Now When ESP-01 wishes to connect , it puts Arduino in reset and takes Arduino out of reset once connected. This works well as a solution on the demo board. Once connected the Wifi is smart enough to avoid the noise frequencies and the connection seems to be robust enough for testing and demonstrating.

    TODO: Provide circuit details for ESP-01 and a video of the updated board

  • Use custom polledSerial serial port class rather than Arduino.Serial

    Andy10/08/2021 at 12:10 0 comments

    The windsensor project is very critical on interrupt timing  to capture the pulse length. Looking at the Arduino.Serial code, it does (of course) use interrupts and even disables the global interrupts during a write. In this application a simple polled serial port will do the job without globally disabling interrupts and without serial write interrupt getting in the way of the envelope detect or zero crossing interrupts, so the Arduino code was modified to use the custom polledSerial class.

  • Added info and video on connecting up the Arduino

    Andy10/07/2021 at 16:06 0 comments

    Added video and info on how to hook up the Arduino to the extended car reversing kit pcb

  • Use atmega328p internal comparator instead of external comparator.

    Andy10/05/2021 at 19:35 0 comments

    To get the project working originally I used an external comparator. The comparator is used first to measure the magnitude of the received pulse stream and then to detect zero crossing, by means of positive feedback which moves the trip level between the two levels dependednt on output state

    I wanted to use the internal comparator and eleiminate theat extra component. One problem with the internal comparator is that the outout doesnt go to a pin, which means the traditional positive feedback cant be used. I originally decided to use a spare pin as an output and use it to set positive feedback to the input it in the comparator interrupt, but then I realised I could just use 2 different comparator channels and switch between them in the comparator interrupt.  After some headscratching  and layout issues this is now working well and I plan to stick with this option for the moment. This is actually a much nicer option since it doesnt put any load on the output and samples the filter ground directly rather than a voltage which is around the same level as I was doing with the external comparator. I also would hope that the settling time is much reduced since both of the negative outputs have constant values, one at filter ground and one at envelope detect voltage and the filter output is not loaded at all.

    I also got v2 of the circuit working using the car reversing pcb in the videos, which is much neater than V1, so now need to do a new video about it all.

  • Modify source code to work without requiring main to disable interrupts

    Andy09/30/2021 at 13:58 0 comments

    Modify source code so that we don't need to disable interrupts. Important since the comparator  is meant to have a lot of positive feedback, which has to be done in software in an interrupt, while only around 6 usec is available to flip the output.

  • Remove the existing mcu with a hot air soldering iron

    Andy09/30/2021 at 07:10 0 comments

    The existing 8 pin soic mcu has to be removed. I have a hot air soldering iron for this task. (If you don't have one, then it is possible to remove smd ics by making a u shaped piece of thick copper wire and soldering it to the ic's pins with blobs of solder. Heat it up with a bigish soldering iron and then pull it off with tweezers)

    Also shown in the video is  the piece of proto board, attached to the existing pcb with double sided tape, which works much better than you would think..

  • Cut out and fit proto board to add area to reversing sensor control board

    Andy09/29/2021 at 10:13 0 comments

    Cut out and fitted the proto board to add some area to the original PCB.

    In the video I mention knitting the 2 boards together with copper wire. To line up the boards for drilling, I  joined the board with doubles sided tape. Once done, I discovered that the result was strong enough just using the tape, which makes life much easier.  Next todo is to remove the MCU with a hot air soldering iron, so will see if the proto board falls off with the heat!

  • Added comparator output to circuit description

    Andy09/19/2021 at 12:23 0 comments

    Added comparator circuit to circuit description.

  • TODO: Use Arduino onboard comparator

    Andy09/18/2021 at 09:43 0 comments

    The sensor is working as shown in

    (Need to redo video with sound turned on!) Originally the project was based on a stm32l432-nucleo board, which runs at 80 MHz. I decided to do the Arduino version because it makes the project more accessible.

    Next task is to breakout the output and ground of the raw filter output and connect to the Arduino Atmega328 mcu comparator, rather than using an external one. Since the Arduino is connected via a cable, I hope that interference won't be a problem. The good thing is that the pulse and address lines will be quiet while listening for the received pulse, so I hope that won't be a problem. First to look at the datasheet and get comparator source working. Unfortunately the comparator output isnt available on a mcu pin, since I need it for positive feedback to the comparator input. I will have to try to fake it by setting another mcu output in the comparator change interrupt.

View all 9 project logs

  • 1
    Get car reversing kit

    Firstly you will need to get the £10 car reversing kit off ebay. The type you want has 4 ultrasound transducers. You don't need the ones with video cameras etc. Beware that the control circuit will need some modifications which will stop it working as a car reversing sensor, unless you take care to make the modifications reversible.

    Please do understand that at this time, it requires removing some smd components from the board and soldering wires to small traces , so you will need some understanding and experience of electronics and testing and faultfinding of pcbs to be able to make this project.
    Note that your PCB may not be exactly the same as the one here, but I believe the layout of these boards is basically similar, at least on the 2 I have.

    Firstly you need to make an extra piece of circuit board and add it to the existing PCB. Just after doing this video I discovered that the simplest way to join the extra board was with double sided tape, which is what I then did.

    Next you will need to identify and remove the existing microntroller. On both boards this was an 8 pin SOIC chip (though on the second, in a footprint that could also take a 14 pin version), which I believe to be a PIC12. Beware that there is also an 8 pin soic dual op-amp ic on the pcb which is used for the band pass filter. The mcu is unmarked , while the op-amp is marked and surrounded by resistors and capacitors ( I have one marked as a 4558D  and the other as a N5532 ) so if you have a marked ic, hopefully, you can look up what it is exactly to verify it is the dual op-amp

    The Arduino can now be soldered in place on the PCB as shown in the photo. See also the later diagram.
    It is best to start with the ground connections, which should be short and using a decent thickness of wire. I would recommend at least 2 ground connections at opposite ends of the arduino, to the nearest ground on the pcb. You will have to scrape away an area of the car reversing kit pcb to solder to. The resistor voltage divider on the pcb is using a 33k and a 22k resistor with a 0.1uF capacitor to create a voltage of 3v which is the negative comparator input for the envelope detect on Arduino Pin 7. The positive comparator input Arduino pin number 6 ,  is connected to the filter output. The 2nd negative comparator input on Arduino pin A0 is the filter ground, which is available at both the non inverting inputs of the dual op amp. The other connections are from Arduino pins A2,A3 to the 4052 address lines on pins 10 and 9. (Do check the sense of these, so you get the corrcet addressing) . The next connection is the output that sends the transmit pulse. This is connected to the 4052 pin3 via a resistor.  It should be possible to find it on the footprint of the microcontroller we already removed. On both of my versions of the car reversing pcb, it was connected to pin 3 of the microcontroller we removed. Finally connect the "Raw" input to the Arduino, to the 12 V power input (after the reverse protection diode if possible).

    Below is a video describing how to connect the Arduino to the car reversing kit pcb.  Below the video is a diagram of the connections.

  • 2
    Construct the wind sensor frame

    The test frame is very easy to make from ply and dowel as can be seen in the picture

    TODO drawings and details of frame.

    This is the proposed final long term version of the wind sensor. It is designed to go at the top of a pole.

    The top is intended to take a solar panel from a solar fountain

  • 3
    Programming Arduino

    Installing and building the project

    Firstly clone the Github Project

    >$ git clone

    Next in the resulting project directory, install the quan-trunk submodule using the following command:

    >$ git submodule init  --update  --recursive

    Start the Arduino IDE. Click on 'File > Preferences' and modify the sketchbook location so that it leads to the ultrasonic_wind_sensor directory.
    Close Arduino and reopen.

    When Arduino reopens , click on 'File > Open' , navigate to the ultrasonic_wind_sensor/ArduUltrasonicWindSensor subdirectory and select "ArduUltrasonicWindSensor.ino".

    Next click on 'Tools > Board' and select (AVR boards) > "Arduino Pro or Pro Mini".

    Next click on 'Tools > Processor' and select "ATmega328P(5v, 16 MHz)"

    Now select 'Sketch > Verify/Compile' to compile.

    The code can be uploaded by attaching a serial cable to the Arduino Pro Mini and clicking Sketch > Upload.

View all 5 instructions

Enjoy this project?



Bharbour wrote 09/20/2021 at 13:42 point

It seems like the sensitivity would be reduced by the sine of the angle between the transducers and the support rod.

  Are you sure? yes | no

Andy wrote 09/20/2021 at 21:29 point

Sure. I don't know, but if you wish to try the inline version, I am sure the same electronics hardware and software can be used with trivial modifications to the code.

In fact there is a hackaday article from around 2013 about such an inline ultrasonic wind sensor also using an atmega328

However I think the inline design has to be much bigger to get the equivalent flight time and is more complicated to build. In practise, it is much easier prototyping with something that can sit on a desk top.

  Are you sure? yes | no

Bharbour wrote 09/20/2021 at 23:05 point

Thanks. I probably will. I need an anemometer for part of another project, and when I finish up some of the stuff I am on now, I am planning on building one.

  Are you sure? yes | no

Andy wrote 10/12/2021 at 10:25 point

On thinking more about this. Yes in theory, however the effect of the angle is to increase the distance travelled for a given distance between the transducers, which increases resolution  as time_of_flight * clock_frequency of the timer.  In other words resolution can be increased either by increasing timer clock frequency ( limited to 16 Mhz in Arduino)  or by increasing time of flight, by increasing distance between sensors bouncing at an angle as I do etc.

  Are you sure? yes | no

Bharbour wrote 10/12/2021 at 14:08 point

You would get the maximum resolution if the sound path was parallel to the flow direction of the wind. If the sound path was exactly perpendicular to the wind direction, you would not see any change of propagation time from the wind.

  Are you sure? yes | no

Andy wrote 10/12/2021 at 16:01 point

The accuracy is dominated by the timer frequency.

Assume  timer clock frequency of 1 Hz and a time of flight of 10 s then the resolution is 1/10  or 10% of the flight time

Now assume same clock frequency but  a time of flight of 20 s. Now the resolution is 1/20 or 5% of the flight time , twice as accurate

For a given horizontal distance x between the transducers we can for example double the flight time and hence the resolution by placing the echo plate so that a pulse at 60 degrees to the horizontal hits it so the height of the echo plate y = tan 60 deg * (x/2).

You can carry on increasing the angle and hence the flight time, but after some time the device gets very tall and the pulse received gets very weak. Also for high wind speeds the difference in angles between forward and reverse flights becomes significant in affecting accuracy of the calculations

  Are you sure? yes | no

Bharbour wrote 09/19/2021 at 19:15 point

Nice project!

I have seen ultrasonic anemometers done with the transducers facing each other, and I understand how they work. With your implementation, it seems like the wind is perpendicular to the sound travel. How does it work? What am I missing?

  Are you sure? yes | no

Andy wrote 09/20/2021 at 09:04 point


The transducers are pointing straight up on the prototype but the sound beam does diverge quite a bit and so enough will head in the right direction to bounce off the smaller ply disc at the top of the rod (which I call the "echo plate") and be detected by the receiver at the opposite side.

An advantage of having the transducers pointing straight up is that the transducers aren't in the airflow, which can cause issues with turbulence on the traditional "in line" style of ultrasonic wind sensor.

It was done this way for a simpler reason ;  it is very easy to make using simple tools! As you can see its just a piece of ply with holes drilled (using hole saw supplied with car reversing kit!)

  Are you sure? yes | no

Similar Projects

Does this project spark your interest?

Become a member to follow this project and never miss any updates