search

mrjones-plip / YANPIWS

Yet Another Pi Weather Station (YANPIWS) - Use wireless sensors and Pirate Weather to show local temperatures and forecast
MIT License
5 stars 4 forks source link
bme280 forecast php rasberry-pi rtl rtl433 weather-station

readme

YANPIWS

Yet Another Pi Weather Station (YANPIWS) - My explorations in getting a Rasberry Pi showing local time and weather:

Overview

For a while I've had a wireless weather station. And for as long as I've had it, it's never quite worked right. Certainly the part about "Equipped with an atomic clock, time is set automatically via radio" has never worked. As well, maybe the sunset/sunrise times were accurate once or twice. The only thing that worked pretty well was the indoor and outdoor temperatures. I stumbled upon the rtl_433 project and I was super stoked to DIY a weather station.

Goals for this project are:

Hardware

Here's the parts I used and prices at time of publishing (March 2017):

Total for this is $127, so, erm, not that cheap. Ideally you'd have a lot of parts around you could re-use for this project. As well, you could reduce the price by going with a 3.5" screen (only $17) and Pi B+ (only $30) . For the B+ you'd have to use ethernet or bring your own USB WiFi adapter. I knew I'd have use for a 5" HDMI monitor, so I was happy to pay the premium.

Caveat Emptor - You'd probably be better off spending even more (ok, not so cheap here any more ;) on a less janky wireless setup like z-wave. I found out the hard way that the IDs of the $13 sensors change every time the sensor batteries die/are changed. ~~As well, the Pi WiFi seems to interfere with the USB SDR (I'm inspiring confidence, yeah?!)~~ Turns out I just needed to move the antenna further away from the Pi! Finally, I coded the HTML and CSS to work in an 800x480 screen or greater. If you use the cheaper, lower resolution screen (480x320), YANPIWS just works thanks to @media sensing.
As well, it works on mobile devices and desktop devices as well. All be it, mobile works best in landscape.

If you want to use the BME/BMP 280 I2C chip instead, this is now supported! So instead of getting the SDR USB dongle and Wireless Temperature Sensora above, instead get (prices as of Sep 2019):

While this reduces costs, it also changes how the set up works. You'll only be able to run one sensor on the I2C bus and you'll have to know how to solder. Finally, check out the alternate install steps below.

Install steps

These steps assume you already have downloaded Buster with Desktop to your Pi and it's installed and booting, online and accessible via SSH. Further be sure your timezone is correct! I recommend using a normal monitor for the install instead of the 5". It's easier this way.

Speaking of monitors - this install also assumes you have your 5" display (or 3.5" if you went that way (or what ever display you want!)) already working.

These steps also assume you're using the SDR dongle and wireless temp sensors. See below for BME280 sensors.

All steps are done as the Pi User - be sure you've changed this user's password from "raspberry" ;)

  1. Ensure your Pi is current; 
    sudo apt update&& sudo apt dist-upgrade

  2. Install rtl, git, apache, php, chrome and chrome utils for doing full screen (some of which may be installed already):

    sudo apt-get install -y curl git chromium-browser apache2 php php-curl php-xml unclutter sed rtl-433

    RTL Note: If your distro doesn't have the rtl_433 package available via apt, follow these compile/download steps: Download, compile and install rtl_433 or use the faster method cited on tech.borpin.co.uk which involves adding a testing apt repo. Note that the last apt call should be apt install rtl-433 with a dash not an underscore. Also on that page note that /etc/apt/preferences should be /etc/apt/preferences.d/.

  3. With your wireless temp sensor(s) powered up and the USB Dongle attached, make sure your sensors are read with rtl_433. Let it run for a while an note the IDs returned for the later step of creating your own config file. Here we see ID 231: 
    pi@raspberrypi:~ $ rtl_433
rtl_433 version unknown inputs file rtl_tcp RTL-SDR SoapySDR
[SNIP]
Sample rate set to 250000 S/s.
Tuner gain set to Auto.
Tuned to 433.920MHz.
Allocating 15 zero-copy buffers
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
time      : 2020-05-11 00:55:17
model     : Fineoffset-WH2                         ID        : 231
Temperature: 35.5 C      Humidity  : 1 %           Integrity : CRC
_

    If you get an error usb_open error -3 or usb_claim_interface error -6, unplug and replug the USB dongle, that should fix it.

  4. Clone this repo, symlink html into www and create your own config.csv: 
    cd
git clone https://github.com/Ths2-9Y-LqJt6/YANPIWS.git 
cd YANPIWS
cp config.dist.csv config.csv
sudo rm -rf /var/www/html
sudo ln -s /home/pi/YANPIWS/html /var/www/html
sudo chown -R pi:www-data data
sudo chmod -R 775 data

  5. Edit your newly created config.csv to have the correct values. Specifically, your latitude (lat), longitude (lon) and labels which you got in the step above running rtl_433. As well, you'll need to sign up for an API key on Dark Sky or Pirate Weather and put that in for the forecast_api value below. If you want static icons instead of animated ones, set 'animate' to false instead of true like below. Here's a sample:

    # edit these two to be your location.  Use one of these if
# you need help! https://duckduckgo.com/?q=what's+my+lat+long+finder
lat,31.775554
lon,-81.822436

# check https://www.php.net/manual/en/timezones.php for available timezones
timezone,America/Los_Angeles

# set your forecast API token from either Dark Sky or Pirate Weather
forecast_api_token,aabbccddeeffgghhiijj112233445566

# set your provider for forecasts (must end in an:
#   pirate weather: https://api.pirateweather.net
#   dark sky: https://api.darksky.net
forecast_api_url,https://api.pirateweather.net

# should we show animated icons for forecast? should be "true" or "false"
animate,true

# label ID map to human name. Format is "labels_" + ID + "," + NAME
labels_211,In
labels_109,Out

# How many temperatures to show on the screen (tested with 1, 2 or 3 - watch out if you do more!)
temp_count,2

# CSS font sizes
font_time_date_wind,35
font_temp,50
font_temp_label,25

# set dark or light theme. if unset or unrecognized, defaults to dark
theme,dark

# likely this won't need to change if you're following default install instructions.
dataPath,/home/pi/YANPIWS/data/

# there's only one of these, so likely this won't change
moondata_api_URL,https://aa.usno.navy.mil/api/rstt/oneday

# unless you're deploying a lot of nodes reporting back to a central server, don't touch these ;)
# api_password very likely should match servers_0_password unless, again, you know what you're doing ;)
api_password,boxcar-spinning-problem-rockslide-scored
servers_0_url,http://127.0.0.1
servers_0_password,boxcar-spinning-problem-rockslide-scored

  6. Run these commands to copy the kiosk.sh script into systemd and enable the service. This will to auto start Chromium in kiosk mode on the Pi's web server every time you boot:

    cd ~pi/YANPIWS
sudo cp scripts/kiosk.service /lib/systemd/system/kiosk.service
sudo systemctl daemon-reload
sudo systemctl start kiosk.service
sudo systemctl enable kiosk.service

    Thanks to Py My Life Up pages for the howto

  7. Reboot your Pi so confirm the browser starts loading the configured YANPIWS app.

  8. Add a cronjob for the Pi user to run every 5 minutes to ensure temperature collection is happening:

    */5 * * * * /home/pi/YANPIWS/scripts/start.sh >> /home/pi/YANPIWS/data/cron.log

    This step will need some improvement as the rtl_433 process can die and your temps will stop being updated :( Stay tuned!

Whew that's it! Enjoy your new weather station. Let me know which awesome case you build for it and report any bugs here!

Alternate install steps for attached I2C sensor

Follow the exact same steps as above, but don't do the last step with the cron job. Instead, you'll need to:

  1. Make sure I2C is enabled by running sudo raspi-config -> "Interfacing Options" -> I2C -> "Yes" -> Reboot
  2. Ensure that your BME280 sensor is attached correctly. Raspberry Pi Spy provided this great schematic.
  3. Assuming you installed in /home/pi/YANPIWS/, run python /home/pi/YANPIWS/scripts/bme280.py and ensure you see good data. This looks like this for me: 
    {"time" : "2019-09-05 14:05:00", "model" : "BMP280", "id" : 96, "temperature_F" : 80.456, "humidity" : 40.54}
  4. If that all looks good, as the pi user, set up a cron job to run once a minute and generate the stats: 
    */1 * * * * cd /home/pi/YANPIWS/html/php;/usr/bin/python /home/pi/YANPIWS/scripts/bme280.py | /usr/bin/php -f read_and_post.php

Multiple Sensor Nodes

If you have a lot of places you want to run temperature sensors, as of version 0.9.2, YANPIWS now supports a server/client deployment. This means you have the ability to run an install that does nothing but send it's data to another YANPIWS instance. To set that up, go ahead and deploy the 2nd (Nth!!) instance based on the steps above. Make sure everything is working. Then, follow these steps:

  1. Get the IP address of where you want to send the data to. See ifconfig if you need help (this command may work, but is likely fragile, ifconfig|grep -i inet|grep broadc|cut -d ' ' -f10). We'll pretend you got the ip 192.168.4.199 back. But use the real IP!

  2. One the remote node, edit the config.csv file and in the bottom section after the $YANPIWS['servers'][] line, do one of the following:

    • change the URL line url to be: 'url' => 'http://192.168.4.199',. Remember, this should be the IP you got in the prior step. This will cause the data to not be stored locally on the node at all. It will only be sent to the remote server. The final result should look like this:

      
$YANPIWS['servers'][] = array(
   'url' => 'http://192.168.4.199',  // no trailing slash please ;)
   'password' => 'boxcar-spinning-problem-rockslide-scored', // should match password above
);

    OR

    • add 4 new lines with your new IP. The final result will look like this:

      $YANPIWS['servers'][] = array(
    'url' => 'http://127.0.0.1',  // no trailing slash please ;)
    'password' => 'boxcar-spinning-problem-rockslide-scored', // should match password above
);
$YANPIWS['servers'][] = array(
   'url' => 'http://192.168.4.199',  // no trailing slash please ;)
   'password' => 'boxcar-spinning-problem-rockslide-scored', // should match password above
);

      This will cause the client to write to BOTH the local and remote YANPIWS instances

  3. If your using the BME280 chip on both your server and your client(s), you'll need to edit the bme280.py. This is because all of the BME280 chips have the same ID and your server won't know which sensor is which. To fix this, change line 165 from this:

    json = '{"time" : "' + str(rightnow) + '", "model" : "BMP280", "id" : ' + str(chip_id) + ', "temperature_F"

    To this:

    json = '{"time" : "' + str(rightnow) + '", "model" : "BMP280", "id" : "44", "temperature_F"

    What this does is hard code this sensor to ID 44. For each new node you deploy, you'll need to change this to it's own unique value.

  4. On your server update the $YANPIWS['labels'] to have an entry for your new sensor. In the case of our 44 example above, along with default 96 that the BME280 uses, that would look like this:

    $YANPIWS['labels'] = array(
    '96' => 'In',
    '44' => 'Out',
);

API calls

YANPIWS, as of version 0.9.2, now has an HTTP API that you can use to send data. This means that anything that can do a POST to the IP of your YANPIWS server instance, can send it temp data! The following fields are required:

Optionally you may pass:

You should use the following URL for your POST, replace IP_ADDRESS with your real IP address of your server:

Development

Check out this repo, cd into the html directory and start a web server:

php -S  localhost:8000

The rtl_433 works great on Ubuntu for desktop/laptop development. Manually kick off the input script and leave it running while you code to gather live temps:

cd html/php
rtl_433 -f 433820000 -C customary -F json -q | php -f read_and_post.php

If you don't want to deal with running the rtl-433 script, copy the sample data example.data to today's date (YEAR-MONTH-DAY) into the data directory. It has IDs 211 and 109 which are the ones already in config.dist.csv.

As well, if you want to simulate individual inputs via the HTTP POSTs, you can use this curl command. Note that we're using the default password, you may need to change this if you've changed it in your deployment:

curl --data "password=boxcar-spinning-problem-rockslide-scored&temperature_F=44.08&id=2&time=2019-09-18 23:59:02" http://localhost:8000/php/parse_and_save.php

As the bme280.py script outputs JSON, if you want to more closely similate the cron job that's run, incluidng using the config vars and a real POST, you can use this:

cd html/php
echo '{"temperature_F":"44.08","id":"2","time":"2019-09-18 23:59:02"}' | php read_and_post.php

Conversely, if you want to use the now deprecated STDIN method, you can use echo to pipe in JSON: 

cd html/php
echo '{"temperature_F":"44.08","id":"2","time":"2019-09-18 23:59:02"}' | php parse_and_save.php

Use your IDE of choice to edit and point your browser at localhost:8000 (or the IP of your Pi) and away you go.

PRs and Issues welcome!

Version History