Anemometer coding woes

Turns out all my anemometer code so far is wrong, equations is not my strong suit and I seemed to be always trying to overwork the equation.

The above is mostly the code I am using now. I have tweaked it since to allow float seconds so I can use the same procedure to work out the gusting speed.

The above code is borrowed from Sparkfuns Weather Station Tutorial.

Arduino Uno to the rescue

Basically I was having all types of trouble trying to get the sensors working on the Pi, different versions of python (mainly due to library limitations) and probably just poor coding. The Lux sensor was the biggest headache, the returning values were all over the place.

So I decided to try the Arduino Uno, since most of the Adafruit tutorials use a Uno to get them working and it uses C programming language, which is a lot less annoying than python. Basically in the space of the weekend I have all four sensors working and even have the interrupt programming done for the rain sensor.

2014-05-04 12.43.15I also order the UV sensor from Pimoroni, which simply returns values for the UV index, which is something I also wanted to capture data for, to see if the index is slightly raising, which might be a sign of global warming/ozone depletion.

SI1145The SI1145 can also read visible light and IR but those returning values are not calibrated, so I am not going to bother with those.

Although I have found out that my RGB Leds that I have are around the wrong way, the common anode needs to be positive and not neutral, so ebay to the rescue again.

Got a little more work to do with code, then I can figure out how to get the Pi talking to the Uno and get the values posted to the web, so I can start getting that sorted.

Moon Phases, Sunrise and Sunset

So I want the web front end to give some other data, that might be of use to people.

Since I am also into astronomy, I would like to know what phase the moon is in (not that we’ve seen all that much of it recently). Turns out some clever bloke has written a php class to tell you what phase the moon is in, source is available via github.

PHP 5.1+ contains functions for calculating sunrise and sunset. As far as I can see there’s 3 functions, date_sunrisedate_sunset and date_sun_info. The latter looks like the best bet as you don’t have to supply as many variables.

Now I just need to get the GPS coords for my flat.

Nwazet Case Tyler Issues

I did post a product review to modmypi.com for the Nwazet Tyler case, but for some unknown reason it’s not been posted. I definitely think people should be aware of a few issues with the case.

1. No Manual

There is only the online manual and its a bit vague about which screws to use in which part. You do have to be careful as the threads are different on some of the screws. Plus some of the critical shots on the website of the screws are blurred making it difficult to identify the correct screws.

2. Foot Piece Broke

The feet is made up of two pieces that fit into each other making a cross, but the machining was a touch tight and when pushing the pieces together, one side broke into two pieces. Luckily I was able to use a plastic bonding solution to repair the piece and then file away some of the plastic, so as to make it fit together without breaking again.

3. Product Shots are not accurate

The Tyler case is supplied with a clear red insert to sit behind the face and red LED’s. It is mentioned in the product description on modmypi, but the pictures are misleading. The finished article will look quite different to the product shots. Nwazet allows you to set the colour when buying, but modmypi doesn’t offer the colour combinations.

Since I will be using multi-colour LED’s, I will have to source another piece of perspex that is opaque white. It does seem a little weird that such a happy case is supplied with red face bits and not a more neutral/happy colour scheme.

DSCF0200

Conclusion

It’s not really a case a child could put together because of the lack of a manual. Other than the issues mentioned here, it’s a nice case and only having the base constructed makes for a nice bed to work around whilst I sort out my wiring and code. Scored: 3 out of 5.

I will update the article when I’ve finished my case and post some pictures.

Wind Sensor Testing

UPDATE2: yeah ignore the drivel in this post and check this post for correct equation for the anemometer.

UPDATE: So it turns out that I haven’t a clue about how you measure wind speed with an anemometer. I found the datasheet for the sensors, which gives 1.492mph per pulse on the anemometer over 1 second, so I have had to rework the equation and procedure below. So the screen shot below gives wrong values, but is an example of the scripts output.

Just started testing with the Anemometer N96FY. I have the anemometer connected to GPIO pin 18 and 3.3 volt supply. This wind sensor is like a switch and it’s default state is closed, meaning electricity with flow through the anemometer except for two opposing points during a full revolution where the voltage will briefly fall to zero.

I work out the equation to be

wind speed = half the circumference / time taken

Wind speed will be in metre per second. Half the circumference is 0.21994 metres.

wind-speed-testing

 

Database Testing

I am going to use a local and remote mysql database to host the weather information. So I had to think about which variables to store and at 1 minute resolution, how much space would I need to accomplish this, as I want to store the information year on year. So I will store the following information:-

  • Date and Time of the reading
  • Humidity Outside
  • Temperature on the Humidity Sensor
  • Temperature on the Barometric Sensor
  • Humidity Inside
  • Temperature Inside
  • Wind Speed Average over 1 Minute
  • Wind Gusting Speed
  • Wind Lowest Speed
  • Light Full
  • Light Visible
  • Light Infrared
  • Barometric (Pressure)

This works out to 1,440 readings a day (60 minutes x 24 hours). Which is 525,600 readings a year, so making the database as small as possible is critical. For this I will be using the DECIMAL data type, so you can specify exactly how much information to store. So the rough table looks something like this:-

So using a local database on a desktop PC via phpMyAdmin, I used a store procedure to generate the dummy information. First create and store the procedure:-

Note: The dummy database name is “pi”. Then you can just call:-

Which will take a few minutes (~10 mins) to finish populating the database. You’ll notice on the stored procedure that I counted for 366 days, rather than 365, just to simulate what a leap year would be.

Final size on disk of the database is ~61MB. Plus a bit more for the unique index on the time_stamp column.

Issues encountered: 1. Day light saving time. I was originally trying to store the timestamp, but on the 30th March 2014 at 2am I would get 60 rows with exactly the same date and time and since the time_stamp field needs to be unique, this was a massive issue, so to cure the problem, I changed the store value to DATETIME and inserted the utc_timestamp() value, so I will probably need to convert the utc timestamp to localised time allowing for day light saving time if I need to refer to a date and time specifically.

2. Log files. During my testing and repeated procedure running, I managed to generate a >600MB log file. So disabling the log files on the testing and production machine will need to be done. Find the following in your my.ini or my.cnf file:-

find the line “log-bin=mysql-bin” and place a # and space in front of it, this will disable the log file generation.

The next step will be see how fast the hour average calls are for the last 24 to 48 hours for the chart mapping.

Parts Ordered

I have finally ordered the parts I wanted. ModMyPi was cheapest for most of the parts, some of the parts were sourced from Pimoroni, who have already shipped my order.

The Micro SD card and adapter, I sourced from Amazon. Decided to go for 32gb card (MB-MPBGCAEU), so the database can expand as much as it likes. Incidently, I read somewhere and for the life of me, I can’t find the article about the NOOBS card from the RP Foundation. That Samsung cards offer the best random access figures, hence why I have opted for the Samsung card, hopefully it will work. It looks awfully like the card that ModMyPi sells.

I do want one of these keyrings.

Charting Libraries

I have been trying to find a decent charting library for displaying the information gathered, but I’ve not been totally successful. Rickshaw was a library I really liked the look of, but it looks like I will use Grafico and build that into the bootstrap front end.

Raspberry Pi did tweet the following

Which is nice, I do want to make the sensor data available, via either a sql dump or php query into a comma or tab delimited format, the idea of having to port it all into Plotly through the api is something I might look at eventually but at the moment, it’s outside the scope of the project.