Home automation on BlocNotes

Fix Home-Assistant statistics and energy erroneous values

Tue, 30 Sep 2025 00:00:00 +0000

Introduction

I have been experimenting with solar panels, Home Assistant and my electricity meter lately. This French meter from the utility is called Linky. It provides a small power supply (130 mW) and a serial link hidden via amplitude modulation.
To avoid using batteries, I built a prototype that harvests this power, demodulates the serial link, and feeds the data into an nRF52 development kit. It works well enough that I never bothered turning this messy setup into a proper PCB.

However, every now and then—despite CRC checks—I get an erroneous reading that messes up the Home Assistant Energy panel. Here’s how to fix incorrect statistics in the Home Assistant database.

Preparation

Before editing the database, it is strongly recommended to stop Home Assistant to avoid concurrent access that could corrupt the database.

Connect to the system running Home Assistant
Install SQLite3 if needed. For example, on Debian-based systems: apt install sqlite3
Locate the home-assistant_v2.db database file in Home-Assistant data folder and create a backup

Open the database for editing

Open the database using SQLite and set a more readable output mode:

$ sqlite3 home-assistant_v2.db
# Then enter the commands
sqlite> .header on
sqlite> .mode column

Find the data id

Identify Home Assitant’s internal identifier, it is called metadata_id. Replace easf02 with Home-Assistant’s Entity ID you are looking for.
Using LIKE '%PARTIAL_ENTITY_STRING% allows for easier search by matching any string containing PARTIAL_ENTITY_STRING

sqlite> SELECT id, statistic_id FROM statistics_meta WHERE statistic_id LIKE '%easf01%';
id  statistic_id            
--  ------------------------
6   sensor.linky_easf01     
15  sensor.linky_easf01_cost

In this case id = 6 is the one we want to fix, and id = 15 is the associated cost.

Search for erroneous data

We let’s ask the database for any sudden jump in the data, e.g change greater than 10,000 Wh. Adjust metadata_id = 6 and the threshold (> 10000) as needed.

WITH ordered_stats AS (
  SELECT
    id,
    created_ts,
    sum,
    metadata_id,
    LAG(sum) OVER (PARTITION BY metadata_id ORDER BY created_ts) AS previous_sum,
    LAG(created_ts) OVER (PARTITION BY metadata_id ORDER BY created_ts) AS previous_created_ts
  FROM statistics
  WHERE metadata_id = 6
)
SELECT
  id,
  datetime(created_ts, 'unixepoch', 'localtime') as date,
  ABS(sum - previous_sum) AS difference
FROM ordered_stats
WHERE ABS(sum - previous_sum) > 10000 ORDER BY id DESC LIMIT 5;

In a nutshell, this query:

Selects statistics data for the entity and associates each value with the previous one.
Displays the five most recent entries where the difference between consecutive values exceeds the threshold.

Example result:

id      date                 difference
------  -------------------  ----------
229351  2025-08-18 14:00:10  20676793.0
38711   2025-01-28 12:00:10  20020401.0

Looks like we found the 20.6 MWh spike reported in the Energy panel :-)

Update the value

Since Home Assistant records cumulative sums, we need to subtract the erroneous difference from all entries starting from the anomaly. Use the following query:

UPDATE statistics SET sum = sum - 20676793.0 WHERE metadata_id = 6 and id >= 229351;

Replace 20676793.0 and 229351 with your actual values.

The statistics_short_term sum should also be adjusted by the same value:

UPDATE statistics_short_term SET sum = sum - 20676793.0 WHERE metadata_id = 6;

Once done, exit SQLite by pressing Ctrl + D. The database is now fixed, and the energy panel displays accurate values!

Fixed Home-Assistant energy panel

Update iTead Sonoff Zigbee 3.0 USB Dongle Plus

Sun, 12 Dec 2021 00:00:00 +0000

I recently bought a Sonoff Zigbee 3.0 USB Dongle Plus from iTead to use it with Home Assistant.
While it worked out of the box, the update itself was not complicated but you have to find the right tools…

Update the firmware

The nice Python script cc2538-bsl can be used, but the branch feature/ITead_Sonoff_Zigbee-delay needs to be selected until the pull request is merged.

Install cc2538-bsl

git clone https://github.com/JelmerT/cc2538-bsl.git
cd cc2538-bsl
git checkout feature/ITead_Sonoff_Zigbee-delay
python3 -m venv .venv
source .venv/bin/activate
pip install wheel
python setup.py install
pip install intelhex python-magic

Flash the dongle

Download the firmware from Koenkk’s Z-Stack:

Then use the command

python cc2538-bsl.py -e -v -w --bootloader-sonoff-usb ./CC1352P2_CC2652P_launchpad_coordinator_20211207.hex

Which should output:

sonoff
Opening port /dev/ttyUSB0, baud 500000
Reading data from ./CC1352P2_CC2652P_launchpad_coordinator_20211207.hex
Firmware file: Intel Hex
Connecting to target...
CC1350 PG2.0 (7x7mm): 352KB Flash, 20KB SRAM, CCFG.BL_CONFIG at 0x00057FD8
Primary IEEE Address: 00:12:4B:00:25:6C:59:51
    Performing mass erase
Erasing all main bank flash sectors
    Erase done
Writing 360448 bytes starting at address 0x0000000
Write 104 bytes at 0x00057F980
    Write done                                
Verifying by comparing CRC32 calculations.
    Verified (match: 0x6dca550e)

After that you can use the nice Zigbee Home Automation integration to easily setup the Zigbee dongle.

Automate shutters with an ESP and Home Assistant

Sun, 12 Jul 2020 18:50:41 +0000

I recently did a proof of concept on shutters automation by hacking the remote control: the idea is to wire the ESP in parallel with the remote control buttons.
My remotes are Bubendorff 41677 but it should work with a lot of other remote controls.

Wemos & the remote control

After poking with a multi-meter it appears that two test points are connected to the right signals:

TP103: Up
TP102: Down

The signal is active low: there is a pull-up resistor and it must be grounded to simulate the activation. A 680 Ohm resistor is used in series with each button of the remote to prevent current overshoot and maybe reduce power consumption, so I also used them:

Schematic

You might not see the resistors on the top picture as I only had SMD 0603 on hand; look carefully :)

On the first attempt, I used the great ESPHome project to simulate pressing the pushbuttons. I just created 3 virtual switches as a proof of concept but for correct integration, a cover component can be used.
I split the configuration into two files: one generic and the other only specifying the shutter name (room) and timings.

Per device file, called shutters-office.yaml in this example:

substitutions:
  # Used for HASS identifier
  name: office
  # Used for friendly name
  switch_name: "bureau"
  # Time from fully closed to fully opened
  full_time: "16s"
  # Time from fully closed to slightly opened
  partially_open_time: "5s"
  # Time for the "button pressed" event
  on_time: "150ms"

<<: !include shutters-common.yaml

Then the generic file called shutters-common.yaml:

esphome:
  name: shutters_${name}
  platform: ESP8266
  board: d1_mini

wifi:
  ssid: "MY SSID"
  password: "MY PASSWORD"

  # Enable fallback hotspot (captive portal) in case wifi connection fails
  ap:
    ssid: "Shutter fallback hotspot"
    password: "12345678ABCDEF"

captive_portal:

# Enable logging
logger:

# Enable Home Assistant API
api:

ota:

switch:
  # Define with GPIOs are used
  - platform: gpio
    id: up
    pin:
      number: 12
      inverted: yes
    restore_mode: ALWAYS_OFF
  - platform: gpio
    id: down
    pin:
      number: 14
      inverted: yes
    restore_mode: ALWAYS_OFF

  # Define template simulating the push on the button
  - platform: template
    name: Shutter ${switch_name} open
    turn_on_action:
      - switch.turn_on: up
      - delay: ${on_time}
      - switch.turn_off: up

  - platform: template
    name: Shutter ${switch_name} close
    turn_on_action:
      - switch.turn_on: down
      - delay: ${on_time}
      - switch.turn_off: down

  # Define template for slightly opening the shutter
  - platform: template
    name: Shutter ${switch_name} slightly open
    turn_on_action:
      - switch.turn_on: down
      - delay: ${on_time}
      - switch.turn_off: down
      - delay: ${full_time}
      - switch.turn_on: up
      - delay: ${on_time}
      - switch.turn_off: up
      - delay: ${partially_open_time}
      - switch.turn_on: up
      - delay: ${on_time}
      - switch.turn_off: up

As there is no feedback information on the shutter state the “slightly opened” state is achieved by fully closing the shutter then opening it for a small amount of time.
Once the ESP8266 or ESP32 is flashed the only thing remaining is to add it to Home-Assistant, thanks to Configuration => Integrations => Set up a new integration => Search for ESPHome => enter the ESP IP address. Et voilà, you are good to go and create powerful automations in Home-Assistant.

On the next version, I used a more clever strategy to command shutters from 0 to 100%, and a LoRaWAN board to allow battery-powered operation - to be published soon.

How to monitor pellet stove's tank

Tue, 26 Nov 2019 12:53:50 +0000

I have a great Rika pellet stove, but despite all the embedded electronics, it is unable to warn me just before the pellets tank is empty. As I already have my own low-power wireless devices and home automation tools, I just added a distance sensor and designed some enclosures.

Pellet stove

Sensor selection

IR sensor

Cheap usual Sharp sensor needs to be at least 10 cm away from the pellets. It is a no go for this use case as pellets are close to the lid when the tank is full

Ultrasonic sensors

As common as the Sharp IR sensor, the HC-SR04 was an option but I discarded it for two reasons:

Dust generated by pellets: sensor difficult to clean
Not working well with a power supply of 3.3 Vdc

STMicroelectronics time of flight sensor

So I went with ST VL53L0X: range adapted to my needs, low power, and it is easy to the clean sensor. Regarding dust, I used it for several months without having to clean it but summer was included, so let’s wait a full winter for feedback.

Hardware

I bought a cheap breakout board from eBay and plug the I2C lines into my NoteRF boards

Software

I just added the library from Pololu to my project then translated the data in Node-Red.

// VL53L0X
if ( sensorType == 7 ) {
    msg.topic = `data/sensor/noterf_distance_${sensorName}/state`;

	var distance = msg.payload.split( "|" )[2];
	distance_percent = 100 - distance/590*100;

	msg.payload = {
	    'rssi': rssi,
	    'battery': battery,
	};

	// Filter abnormal values
	if ( ( distance >= 0 ) && ( distance < 8000 ) ) {
	    msg.payload.distance = distance;
	    msg.payload.distance_percent = distance_percent;
	}
}

Home assistant

As for all other connected things in my house it reports to Home Assistant, which allows a nice interface, that I rarely use thanks to the automation: I am notified on my phone (through pushbullet) when tank level or device battery cross a threshold: the purpose of that device is achieved! Anyway, some fancy displays can be setup in a few seconds:

Tank status gauge - color changes according to the value

Tank status in millimeters - useless as percentage is more descriptive

Icons + values - I have view with it for all my devices

Device battery needs calibration :-)

Mechanics

I designed with OnShape two 3D printed boxes for the sensor and the electronics/battery, both are using magnets to stick in place The sensor is on the tank lid, while electronics and batteries are outside, hidden on the back of the stove.

Design files:

Time of flight sensor - Hot glued into 3D printed box

Enclosure for the electronics and the battery

ESP32 / MQTT - BLE beacon tracker powered by microPython

Thu, 19 Oct 2017 13:47:55 +0000

I am currently working with BLE beacons -only iBeacons for now- with Espressif’s esp-idf libraries. The aim is to use the ESP32 as a gateway to published detected beacons to an MQTT broker. It was surprisingly easy and I will write something about that when the code will be more polished.

Anyway, by curiosity and to speed up development, I wanted to try microPython port to ESP32, and I did. Unfortunately, BLE support is not yet reliable enough and a lot of advertisements were lost with 4 to 6 beacons. This post is to keep track of my work and, hopefully, use in the future when microPython port will be more reliable :-)

microPython compilation

While compiling microPython was really straightforward, I had one issue afterward: I was unable to use the MQTT library, neither by calling it mqtt or umqtt, nor by putting it manually in flash. I did not investigate why but I found that adding it to the “frozen” scripts worked.

The code used is from Pycom with Wipy 2.0 as target. I used this commit against this version of esp-idf.

Clone both repositories, update esp-idf’s submodules
Set PATH to xtensa compilation tools. Be careful to use only version 1.22.0-61 for the mentioned Pycom & esp-idf commits.
Set IDF_PATH to esp-idf directory
Change directory to pycom-micropython-sigfox
run
- cd mpy-cross && make clean && make
- cd ../ESP32/frozen
- wget https://raw.githubusercontent.com/micropython/micropython-lib/a09b3ec20a745cfd1bccb7bd9db513bfe36968db/umqtt.simple/umqtt/simple.py
- cd ..
- make BOARD=WIPY TARGET=boot
- make BOARD=WIPY TARGET=app
- make BOARD=WIPY flash

You can now connect to microPython REPL on your ESP32!

##Python BLE tracker Following is a quick and dirty code I used to test it. It does:

Connect to WiFi
Connect to MQTT server
Scan for BLE advertisements
Send one MQTT message by discovered device

from network import Bluetooth, WLAN
from mqtt import MQTTClient
from binascii import hexlify
import machine
import time
import ujson


def hexlifyNone(object):
    return None if object is None else hexlify(object)


def scan():
    devices = {}
    print("***   Starting   ***")
    bt = Bluetooth()
    print("Scan...", end="")
    bt.start_scan(10)
    time.sleep(10)
    print("OK")
    advs = bt.get_advertisements()
    print(advs)

    for adv in advs:
        devices[hexlify(adv.mac)] = {
            "Mac": hexlify(adv.mac),
            "Rssi": adv.rssi,
            "Data": hexlifyNone(adv.data),
            "MfrData": hexlifyNone(
                bt.resolve_adv_data(adv.data, Bluetooth.ADV_MANUFACTURER_DATA)
            ),
            "NameShort": hexlifyNone(
                bt.resolve_adv_data(adv.data, Bluetooth.ADV_NAME_SHORT)
            ),
            "NameComplete": hexlifyNone(
                bt.resolve_adv_data(adv.data, Bluetooth.ADV_NAME_CMPL)
            ),
            "Flag": bt.resolve_adv_data(adv.data, Bluetooth.ADV_FLAG),
            "Appearance": hexlifyNone(
                bt.resolve_adv_data(adv.data, Bluetooth.ADV_APPEARANCE)
            ),
        }
        """
        print( 'MAC: {}\n\tRSSI: {}\n\tData: {}\n\tName: {}\n\tMfrData: {}'.format(
                hexlify( adv.mac ),
                adv.rssi,
                hexlify( adv.data ),
                bt.resolve_adv_data( adv.data, Bluetooth.ADV_NAME_SHORT ),
                hexlify( bt.resolve_adv_data( adv.data, Bluetooth.ADV_MANUFACTURER_DATA ) )
                )
              )
        """
    print("\n***********\nDevices: ", devices)
    for key, data in devices.items():
        print("\n***********\n", data)
        msgJson = ujson.dumps(data)
        client.publish(topic="/test", msg=msgJson)

    print("***   End   ***")


def sub_cb(topic, msg):
    print(msg)


wlan = WLAN(mode=WLAN.STA)
wlan.connect("SSID name", auth=(WLAN.WPA2, "password"), timeout=5000)
while not wlan.isconnected():
    machine.idle()
print("Connected to Wifi\n")

client = MQTTClient(
    "device_id",
    "MQTT broker address",
    user="user if any",
    password="password if any",
    port=1883,
    ssl=True,
)
client.set_callback(sub_cb)
client.connect()
client.subscribe(topic="/test")
client.publish(topic="/test", msg="Started")
scan()

NoteRF - Sensors schematic and layout

Tue, 09 Feb 2016 09:38:04 +0000

I finaly published them: sources of my wireless sensors are available here.

For the record it has:

ATMega328
RFM69 wireless chip
All analog and digital ports
3.3V regulator
Battery tension can be sensed through A7
Footprint for DS7505 temperature sensor
FTDI and ISP connectors

Schematic

Layout

Dashing - Job to get data from InfluxDB v0.8 database

Tue, 14 Apr 2015 09:29:00 +0000

To use my dashboard, I wrote some Dashing jobs to be able to query some values from my InfluxDB database.

There are thre versions:

In Ruby
In node.js (javascript) after my switch to Dashing-js
A way to update a ‘list’ widget

The third code also changes the background color of a list widget, to set it regarding data values. I will share the widget’s code soon, I just need some time to clean a bit the code

1. Ruby code

# Insert required components
require 'json'
require 'net/http'

# Parameter
serverAddr = 'http://localhost:8086'

# :first_in sets how long it takes before the job is first run. In this case, it is run immediately
SCHEDULER.every '1m', :first_in => 0 do |job|

  # Url to get data
  url = URI( "#{serverAddr}/db/db_name/series?u=xxxx&p=xxxx&q=SELECT%20*%20FROM%20%22/data/Conservatory/Humidity%22%20LIMIT%201" )
  # Query server
  res = Net::HTTP.get( url )
  # Transforms JSON in object
  j = JSON.parse(res)
  send_event('welcome', { text: j[0]['points'][0][2] })
  send_event('synergy', { value: j[0]['points'][0][2] })

end

2. In node.js, easier to modify

// Includes
var request = require( 'request' )

// Configuration
var dbAddress = 'localhost'
var dbPort = '8086'
var dbName = 'XXXX'
var dbUser = 'XXXX'
var dbPwd = 'XXXX'
var refreshTime = 60*1000    // in ms

function getData( parameterToRefresh, widgetId, unit ) {
  // URL creation
  var urlToGet = 'http://'+dbAddress+':'+dbPort+'/db/'+dbName+'/series?u=' + dbUser+'&p='+dbPwd+'&q=SELECT%20*%20FROM%20%22/data/'+parameterToRefresh+'%22%20LIMIT%201'
  // Request
  request( urlToGet, function (error, response, body) {
    if (!error && response.statusCode == 200) {
      // No error, send the value to the dashboard
   	  send_event( widgetId, { value: Math.round(JSON.parse(body)[0]['points'][0][2]*100)/100 })
    }
  })
}

setInterval( function() { getData( 'Outside/Pressure', 'outsideText', ' mbara' ) }, refreshTime )

<pre>// Includes
var http = require( 'http' )
var async = require( 'async' )
var request = require( 'request' )

// Configuration
var dbAddress = 'localhost'
var dbPort = '8086'
var dbName = 'XXXX'
var dbUser = 'XXXX'
var dbPwd = 'XXXX'
// itemsToQuery [ dataNameInInflux, NameToDisplayInWidget ]
var itemsToQuery = [ [ 'Conservatory',  'PLA' ],
                     [ 'SensorTest_1',  'Citrus' ],
                     [ 'LivingRoom',    'Salon' ],
                     [ 'Office',        'Bureau' ]
                    ]
var refreshTime = 60*1000    // in ms

// Function to fetch data
function getDataList(url, callback) {
  var data = ""

  var req = http.request(url, function(res) {
    res.on('data', function(chunk) {
        data += chunk
    })
    res.on('end', function(e) {
        callback(e, data)
    })
  }).end()
}

function refreshDataList( parameterToRefresh, widgetId, unit ) {
  // Variables declaration
  var urls = []
  var itemarray = []
  var average = 0

  // Set url regarding unit
  switch( unit ) {
    case ' %': urlPrefix = 'data/'; urlSuffix = '/'+parameterToRefresh; break
    case '°C': urlPrefix = 'data/'; urlSuffix = '/'+parameterToRefresh; break
    default:   urlPrefix = parameterToRefresh+'/'; urlSuffix = ''; break
  }
  // Table construction regarding the items ot query
  for( var i = 0; i < itemsToQuery.length-1; i++ ) {
    urls.push( 'http://'+dbAddress+':'+dbPort+'/db/'+dbName+'/series?u='+dbUser+'&p='+dbPwd+'&q=SELECT%20*%20FROM%20%22/'+urlPrefix+ encodeURIComponent(itemsToQuery[i][0])+urlSuffix+'%22%20LIMIT%201')
  }

  // Asynchronous call to get the data
  async.map( urls, getDataList, function( err, result ) {
    // Parse results
    for( var i = 0; i < result.length; i++ ) {
      itemarray[i] = {label: itemsToQuery[i][1], value: (JSON.parse(result[i])[0]['points'][0][2] + unit)}
      average = average + JSON.parse(result[i])[0]['points'][0][2]
    }
    average = Math.round( average/result.length*100 )/100

    // Select color and 'moreinfo' message regarding data type and average value
    switch( unit ) {
      case ' %':
        backGroundColor = humidityColor( average )
        moreinfo = 'Moyenne : ' + average + unit; break
      case '°C':
        backGroundColor = insideTemperatureColor( average );
        moreinfo = 'Moyenne : ' + average + unit; break
      case ' V':
        mini = Math.min.apply(Math, result.map(function(table) { return JSON.parse(table)[0]['points'][0][2] }))
        backGroundColor = batteryLevelColor( mini )
        moreinfo = 'Minimum : ' + mini + unit; break
      case ' dB':
        mini = Math.min.apply(Math, result.map(function(table) { return JSON.parse(table)[0]['points'][0][2] }))
        backGroundColor = rssiColor( mini )
        moreinfo = 'Minimum : ' + mini + unit; break
      default:
        backGroundColor = '55555'
    }

    // Prepare structure to send
    var datastruct = {
      items: itemarray,
      backGroundColor: backGroundColor,
      moreinfo:  moreinfo
    }
    send_event( widgetId, datastruct )
  })
}

// Set job call every refreshTime
setInterval( function() { refreshDataList( 'Humidity', 'humidityList', ' %' ) }, refreshTime )

From Dashing.io to Dashing-js

Fri, 10 Apr 2015 09:15:00 +0000

I recently discovered Dashing.io, which is an easy to use dashboard, which seems to fit my needs :

Easy to install
Useful widgets available, easy to add new ones
Generation of templates for dashboards, jobs, widgets
Great design (subjective), some friends find it too close to Windows 8, but I don’t use it, so…
New data can be pushed via internal jobs or external API, by calling an URL
Tile based interface with drag&drop re-arranging (based on Gridster)
Auto-adaptation to smartphone/tablets (Responsive)

From my point of view, the major drawback is that it uses Ruby language. I have nothing about it, I just did not want to add another ‘technology’ to this project, which uses already lots of things. So I tried it, and then discovered Dashing-js: a Node.js port.

Dashing dashboard

The port is not as well documented as the Ruby version, so I will share my experience:

Run on a sub-directory (helpful for reverse proxy)

The easiest way to add it on /dashing is:

In file node_modules/dashing-js/lib/dashing.js
Add app.use('/dashing', app.router); (L75) inside app.configure(function() { } block
In file assets/javascripts/application.coffee
Add Batman.config.viewPrefix = 'dashing/views' after #= require_tree ../../widgets,

Stop a dashboard launch as daemon

If you start your dashboard in background with dashing-js start, dashing-js stop will not work, you must use kill PID command. I hope to have some time to explore the code and correct the bug/implement it soon.

Set default dashboard

Edit the file server.js, then line dashing.default_dashboard = 'subPathIfNeeded/sample';.

Save the current tile layout

Drag the tiles
Click on the ‘Save this layout’ button
Copy the code
Paste it in dashboard/yourDashboard.jade respecting the following example indentation, and the . (dot) after script(type="text/javascript"). Done !

extends layout

block title
  | Dashboard's title

block content
  div.gridster
    ul
      li(data-row='1', data-col='1', data-sizex='4', data-sizey='2')
        div(data-id='welcome', data-view='Text', data-title='Hello', data-text='This is your shiny new dashboard.', data-moreinfo='Protip: You can drag the widgets around!')
    center
      div(style='font-size: 12px') Try this: curl -d '{ "auth_token": "YOUR_AUTH_TOKEN", "text": "Hey, Look what I can do!" }' -H "Content-Type: application/json" http://#{request.header('host')}/widgets/we$


  script(type="text/javascript").
    $(function() { Dashing.gridsterLayout('[{"col":1,"row":1,"size_x":4,"size_y":2} [...] }]') })

NoteRF - Add a new sensor - Soil moisture sensor

Fri, 20 Mar 2015 08:29:00 +0000

As I like to have plants at home but always forgot to water them, I bought a cheap resistive soil moisture sensor. Actually, it is just two strips of copper wired as a voltage divider with a 10k resistor, the IC is used as a comparator to have a “digital” output, with a threshold set by the trimpot.

Soil moisture sensor

The sensor can draw up to 35 mA, so I wired the power line to an output of the microcontroller. Be careful that on the ATMega328, it is 40 mA max by pin with a sum of 200 mA for all pins. It might be good to use a FET transistor if too much power is drawn.

Adding it to my home automation system is quite straightforward:

Add a new sensor type in config.h
Add some code to set the I/O in setup() function
Add some code to read the sensor in loop() function
Send it over the air :)

My citrus is complaining if it needs some water! For the record, I found that I need to water the plant if the value is more than 200 out of 255. This value must be changed regarding soil composition and its salt saturation.

Even if it works, I already see some improvements:

I read an analog value, so there is no need for the comparator: try a 100% DIY way with some copper + resistor
Standalone watering system: use the comparator + RC or a 555 timer to open a solenoid valve under a bottle of water
For vegetables in the garden, link it to the water pump of my water well

Copper strips might get some oxidation over time, which will change its resistance, it might be better to use capacitance sensor instead.

NoteRF - Why create a new set of things ?

Wed, 18 Mar 2015 08:20:00 +0000

Hardware

I had a discussion with a friend of mine on Why the hell are you designing a new PCB instead of buying one ? He is right: there is a lot of great cheap boars like Moteino, Anarduino, Jeenode… But on the other side:

I like to learn.
Some of them are expensive compared to a full DIY solution.
It is not so difficult: it is just an ATmega, an RF chip, an optional voltage regulator and some passives selected to optimize power consumption.
Adapted to my needs (RFM69hw, dedicated footprint for some sensors, form factor, regulator, connectors…).
I am not reinventing the wheel, so I will use libraries from these boards.
Regarding firmware they have done a great job, but I did not found it flexible enough or too complicated for my needs.

Software & IHM

A lot of big softwares, hhis time too, I see the pros and cons:

Jeedom
- Good GUI design, looks pretty
- Only use of MySQL, not really adapted for storing sensors data on a small embedded Linux card
- Open-source but commercial aspects are really intrusive
- Not so easy to add a module
Domoticz
- Seems more open and lightweight than Jeedom
- Bad GUI design
- Seems to be based on SQLite, lightweight but some old test on a Gumstix gave me issues after a few months
OpenHAB
- Created to be linked to any modules or database
- Not lightweight, might be related to the use of Java
- Bad GUI design for version 1, the v2 seems great.
- The v2 will be too late for me

Also, I discovered Node-Red which is really easy to use and created “for wiring together hardware devices, APIs and online services in new and interesting ways”, exactly what I am doing. The main drawback is that everything is asynchronous.

Database

On the database side I wanted to try time-series databases which are made to collects metrics, and NoSQL.

MongoDB with or without cube layer: I heard good things but never tried.
Graphite: I tried on a Raspberry Pi, it works but it is a complete solution: too big for me, a lot of dependencies and components
InfluxDB: I like the approach, the v0.9 add ‘tags’ which are good for my application, perform well on the rPi. Maybe a too unstable but I will give a try :)

On the GUI side, I did not found what I want yet, so I am testing Grafana to display data and might use some links inside to start actions until I find something great. I might also do it myself, but I am far away from being a designer so it can be really ugly…

Home automation on BlocNotes

Fix Home-Assistant statistics and energy erroneous values

Introduction

Preparation

Open the database for editing

Find the data id

Search for erroneous data

Update the value

Update iTead Sonoff Zigbee 3.0 USB Dongle Plus

Update the firmware

Install cc2538-bsl

Flash the dongle

Automate shutters with an ESP and Home Assistant

How to monitor pellet stove's tank

Sensor selection

IR sensor

Ultrasonic sensors

STMicroelectronics time of flight sensor

Hardware

Software

Home assistant

Mechanics

ESP32 / MQTT - BLE beacon tracker powered by microPython

microPython compilation

NoteRF - Sensors schematic and layout

Dashing - Job to get data from InfluxDB v0.8 database

1. Ruby code

2. In node.js, easier to modify

3. And a way to fill a ‘list’ widget

From Dashing.io to Dashing-js

Run on a sub-directory (helpful for reverse proxy)

Stop a dashboard launch as daemon

Set default dashboard

Save the current tile layout

NoteRF - Add a new sensor - Soil moisture sensor

NoteRF - Why create a new set of things ?

Hardware

Software & IHM

Database