Cloud Sensor Project

[Mike]

I have also obtained some phototransistors as it would be nice to have graphs of daylight readings overlaid on the temperature graphs. Would be able to keep histories of cloudy/clear-ness along with light levels, i.e. moonlight, etc.

Also, speaking to Mike Rosseel from www.cloudwatch.net the outputs from the Boltwood Cloud Sensor are used with various kinds of observatory control software so if I make my sensor output the same format it will be compatible with that software also. Mike has a data book with the ASCII codes for teh Boltwood and is emailing it over to me.

[RobertM]

Sounds like you've made a good start and a bit of good luck with contacts will help a lot.

[Whitters]

Sounds a pretty interesting project, will you give us a sneak preview at the next imaging session?

[Mike]

Sounds a pretty interesting project, will you give us a sneak preview at the next imaging session?

Sure!! 


I have rigged up 3 x sensors now and have code to read data off each one. I am surprised how well close the readings are from 3 different sensors. Here is a data run with readings taken every 5 seconds. As you can see, each sensor is only a fw hundreds of a degree out from the others.

23.75  23.93  23.75   
23.75  23.87  23.75   
23.75  23.93  23.75   
23.81  23.93  23.75   
23.81  24.00  23.81   
23.75  24.00  23.75   
23.75  24.00  23.75   
23.75  24.00  23.81   
23.81  24.06  23.81   
23.81  24.00  23.81   
23.81  24.00  23.81 

So my hardware is working correctly, the Arduino code is correctly reading sensor data and reporting it. So next is :-

a) Fit the sensors into the respective weatherproof crucibles to be fitted externally.
b) Write the code to read the serial data from the Arduino to the PC and run it through an algorithm to determine if it is clear or cloudy.

[Ian]

cheeky edit there Mike. So what was wrong with your code then? 

[MarkS]

Fascinating project Mike.  I'm looking forward to seeing the results.

Mark

[Mike]

cheeky edit there Mike. So what was wrong with your code then? 

Nothing, I did something stupid on the breadboard that's all. Works fine now. 

[Mike]

I have bought some Osram SFH-3310 Phototransistors today so i can add measurements of ambient light into the mix.

[Mike]

Last night I only had about an hour free to do anything. I conencted the light sensor up to the Arduino and after much messing around finally managed to get a sensible reading from the device. It was only after putting a hefty amount of resistance (via two resistors in series) that the device started to give sensible readings. The Arduino analog input reads 0 - 5.5V dc as 0 - 1024 values. So that part is now working correctly.

Next I am going to write the PC code to take the serial output from the Arduino for the 4 sensors and to start with, simply log the data. Once that is working I can then mount the sensors inside their relevant containers and start comparing the data to actual sky conditions to work out how the algorithm will work.

I'll get some pictures of the gizsmo so far uploaded tonight.

[Mike]

I have decided to do the coding using Processing (www.processing.org) which is a Java based language as:-

a) it has some nice libraries for serial communication with the Arduino
b) it has some nice libraries for graphics, including 3d graphics to make it easy to produce pretty graphs, etc.
c) it can output source code in Java meaning a browser enabled interface would be easy.
d) it can also output in executable formats for Windows, OSX and Linux meaning a native cross-platform application will be easy to make. (A native app. would enable me to output the ASCII codes in the Boltwood Cloud Sensor format meaning it could 'talk' to Boltwood compatible applications.)
e) the IDE for the Arduino is Processing based and the Arduino language is Wiring which is an extension of Processing.

[Mike]

Ooohh apparently this is getting quite technical. 

[Mike]

Here is a picture of the test bed so far :-



I have now written some code on the PC to read the temperature data over the USB port so I am now working out how to graph that data to the screen.

[Mike]

OK I have now made a prototype cloud sensor which is outside in the garden screwed to a post at the moment. This is what it looks like :-



It is current reading 12.31 Degrees C on both sensors for a 100% overcast sky. That is correct according to the theory. I now just need a clear sky to take the readings and ensure the theory works then too.

Still working on the software. This is a screenshot of what I have done so far. It is simply taking the sensor readings and plotting them on a graph over a 3 hour period so I can test out the theory.

[Mike]

OK the sensor has been running for a few days now and I have not been getting the results I wanted. I have now found out why. Last night I scratched off the black paint at onyl the top part of the black bulb so there was a 'window' to teh sky and I managed to get a greater temperature difference.

However, talking to the weather guys the bottom sensor needs to be open to normal airflow. The weather guys use a 'Stevenson Screen' but for my purposes some kind of ventilated container would do.

I will come up with Mark II of the design over the next few days and let you know how it goes.

[Mike]

Mark II of the prototype is now up and running and under a clear sky it is measuring a temperature difference of anywhere between 0.5 and 0.75 degrees, which is good. Will keep it logging overnight to see what happens once the sun comes up and/or it clouds over.