I automated the rangehood above my stove (kitchen extractor fan and light), and learned a lot!

ndbroadbent · September 7, 2022, 2:28am

I’ve been dabbling with electronics for a long time, but I learned a lot from this project and wanted to share some of my experiences.

I figured out how to read the state of 4 buttons and light up 4 LEDs with only 5 wires
I discovered some powerful circuit simulation software
I used optocouplers for the first time (instead of relays)
I wrote some C++ code for a custom ESPHome component
I bought an oscilloscope

Introduction

I wanted to be able to control the light and extractor fan above my stovetop, so I decided to automate it using an ESP32. This would be a similar concept to the Raspberry Pi Microwave project that I built many years ago: I would put a “proxy” circuit between the buttons and the main board that controls the light and fan. The circuit would simulate button presses to control the main board, and it would read the state of the original buttons so that they would still work normally. The circuit would add WiFi connectivity so that the buttons could be controlled remotely and via automations in Home Assistant.

I still use Raspberry Pis, but now I prefer using ESP32 and ESP8266 chips. ESPHome makes it so much easier to build little devices like this, and the OTA (over-the-air) updates are really convenient. I love all the built-in components that you can easily add to your YAML configuration files.

Here’s a photo of the rangehood, with some of my tools. You can see the main controller board and the button board hanging out.

This is the button board, which has 4 buttons, and an LED for each button:

How to read 4 buttons and light 4 LEDs with only 5 wires

I wanted to preserve the original functionality of the rangehood buttons, so that you wouldn’t even know there’s an ESP32 chip inside. This button board was the first obstacle. If there were only 4 buttons, then that would be quite easy. You can just attach them directly to GPIO pins with an internal pull-up resistor. The LEDs were a bit confusing at first, but it was a fun puzzle to solve.

I took a photo of the PCB and traced it using different colors, so I could where each of the wires was going:

I found that button has a 1K ohm resistor, and each LED has a 148 ohm resistor.

Circuit simulation software

I’m a software developer, so I really like having fast feedback loops and the ability to try lots of ideas quickly to see if something works. I started wondering if it might be possible to do this with circuits. I wanted to see if I could figure this out with trial and error and use some kind of circuit simulation software.

I found TinkerCad, and my mind was blown! Now I had a virtual Arduino wired up to a virtual breadboard. I recreated this button/LED board inside TinkerCad. I could set up virtual multi-meters to see how much current was flowing through LEDs, and easily figure out values for resistors.

This is so awesome! It even tells you when too much current is going through an LED:

Here’s a link to my TinkerCad project where you can run the circuit simulation.

I started playing around with some code in an Arduino sketch. I figured out how to light up the LEDs. I figured out how to read the buttons. But I couldn’t figure out how to do them both at the same time. If I wanted to light up the LEDs, then I’d need to be using 3.3V on the shared wire. But if I wanted to read the buttons, then the shared wire needed to be connected to GND.

Then I finally had a breakthrough. Whenever I had worked with GPIO pins in the past, I had just assumed that “high” and “low” meant “on” and “off”. But I realized that “high” literally just means 5V (or 3.3V), and “low” literally just means GND. So you can just change the direction of a circuit if you set your output pins to normally “high”, and you swap the GND pin with a 3.3V pin.

So instead of needing to choose between the 3.3V pin and the GND pin, I could use a GPIO pin on the shared wire and toggle it between “high” and “low”. I set it to “low” whenever I needed to read the buttons, and set it to “high” whenever I needed to light up the LEDs.

I got something working in my virtual Arduino. The other side of the circuit was quite easy (to simulate the button presses.) I used some optocouplers as relays, where you send current through an IR LED and it turns on a transistor on the other side, but it keeps both of the circuits separated. Later on I realized that I should have been looking at “solid state relays”, which are optocouplers that are specifically designed for this purpose and can switch much higher voltages and currents.

I tried it out with a real Arduino, and it worked!

It was a pretty cool experience to simulate a whole Arduino circuit inside my browser, and then see it work in real life.

https://everycircuit.com is also a really cool option for circuit simulation:

It can handle some things that TinkerCad can’t do, such as astable multivibrator circuits where you need to inject a little noise to get it started. (TinkerCad just crashes with an “infinite loop” error.)

Custom C++ ESPHome Component

Alright, on to the next step. I’ve got something working with an Arduino, but how do I port this to ESPHome and get it working over WiFI?

I needed to set an GPIO pin to output mode for a while, briefly switch it to input mode, read the state of a button, and then set it back to output mode again. I couldn’t figure out how to do this in a YAML configuration file for ESPHome, so I decided to write a custom component using C++.

The ESPHome documentation is pretty good, and I was able to get something working. Here’s my YAML configuration and my custom C++ module: Rangehood ESP32 Controller - ESPHome Configuration · GitHub

The most important difference between Arduino and ESPHome is that a component’s loop() function is only called once every 16 milliseconds. (Arduino restarts the loop instantly.) This was actually perfect for me, since it meant that I could read the buttons once every 16ms, and leave the LEDs on before I exit the loop.

I ran out of pins on my ESP8266 (Wemos D1 Mini)

I should have made sure that everything worked on a breadboard. Instead, I started to get a bit impatient, and I jumped straight to soldering a prototype PCB. I soldering some pin headers for my ESP8266 development board, and then started with one input, and one optocoupler for the output. I tested it and it was all working, so I kept going.

Then I checked the ESP8266 pinout and realized that I had run out of GPIO pins.

I tried to get away with using some of the yellow “OK” ones, but the boot started failing (I guess I was pulling some of the pins high or low.) I needed 4 pins to control the optocoupler outputs, and 5 for the button/LED board. That’s 9 GPIO pins, and the ESP8266 only had 7. I thought it had more!

This can be solved with shift registers. You only need 3 GPIO pins to control a 74HC595 8-bit shift register and get 8 digital outputs. You only need one additional GPIO pin if you want to add a CD4021BE shift register (parallel-in, serial-out) and get 8 digital inputs. This is because the input and output registers can share the same clock and latch pins. You can then daisy-chain both of these to get as many inputs and outputs as you need (or use bigger shift registers with more pins), and use only 4 GPIO pins.

Anyway, I switched to using an ESP32 development board which has 18 usable GPIO pins, and a few more with caveats. I used breadboard jumper wires to wire it up.

My resistors were too big

I had put 1k resistors on the optocoupler outputs, to mimic the original button/LED board. What I didn’t realize is that the optocoupler itself seemed to add about 500 ohms of resistance, so the resistance was slightly too high, and the rangehood controller board couldn’t reliably read the simulated button presses.

So I unsoldered them and switched to 560 ohm resistors. Should have tested on a breadboard.

I bought a USB oscilloscope

After doing a bunch of unsoldering and resoldering, putting random hookup wires all over the place, and swapping out ESP dev boards, I ended up with the ugliest prototype PCB you’ve ever seen. Of couse, stuff started breaking and one of my outputs wasn’t working properly or reliably. It seemed like there was a problem with one of my optocouplers, and it was a real pain to figure out what was broken since I had only had a multimeter.

I realized that a oscilloscope would be really useful, and I should have bought one a long time ago. I bought a USB BitScope 10. It’s pretty cool!

This helped me figure out where some things needed to be resoldered. One of the optocouplers was only working when I pushed on it with my finger. I originally thought it might have been a broken chip or something to do with capacitance, but it was just a broken PCB trace and some dodgy soldering.

I’m still learning how to use the BitScope software and have barely scratched the surface of what it can do. This is going to be extremely useful for future projects, especially for reverse engineering how stuff works. (P.S. You’ll need their pre-release version for the latest version of Mac OS.)

The WiFi sucked on my “Duinotech Wearable ESP32 Development Board”

I initially switched to a Duinotech Wearable ESP32 Development Board that I had bought a while ago. I got everything working, but then it really struggled to connect to WiFi and would cut out regularly, and my entities would become unavailable. Even though I had a Ubiquity access point only 10 meters away in the same room.

I found a few other people who seemed to have problems with WiFi as well, but they might be for different reasons:

Then I found the “WiFi Power Save Mode” section in the ESPHome documentation:

NONE (least power saving, Default for ESP8266)
LIGHT (Default for ESP32)

So I tried:

wifi:
  power_save_mode: none
  output_power: 20

I think this seemed to help a little bit, but it still wasn’t very reliable. I had ordered some more electronics stuff and threw in another ESP32 Development Board. I tried this out once it arrived and it was an instant improvement. I just threw away the board with poor WiFi. I also ordered a bunch of shift registers so I can get back to using ESP8266 boards (adafruit huzzuh feathers), and they seem to have much better WiFi as well.

I also ordered a few extra Wemos D1 Mini Pros that support external antennas. They haven’t arrived yet, but I might try these out for car presence detection. I’ve been struggling with WiFi range for this as well.

Putting it all together

I put everything into a little black box. Connected all the wires up and stuck it inside the rangehood. I chopped the power cable for the rangehood and added a screw terminal, and wired up a USB charger to power the ESP32 board.

So now I’ve got the fan and light in Home Assistant. (And the physical buttons still work as well.)

I’ve set up an entity controller to turn on the light.

Next steps

The rangehood controller board has a piezo buzzer that beeps every time a button is pressed. That’s pretty annoying. I might try to desolder or destroy the buzzer.

I want to put an air quality sensor in the kitchen and automatically turn on the extractor fan based on AQI. I’ve ordered some ZigBee air quality meters on AliExpress, and they should arrive in a few weeks.

I also bought a current clamp sensor that can measure AC current. I want to set this up for the stovetop so I can detect when it’s on and automatically turn on the fan (in advance, instead of waiting for the AQI to get bad.) It would be easier if I could use an energy monitoring wall plug, but it looks like the stovetop is wired directly into the circuit breaker, and it uses a lot of power so I don’t want to mess around with those wires. I’ll just separate them and put a clamp around one of them. I’ll follow this guide to set it all up and get it working on an ESP32.

I have this power meter that I’ll use to calibrate it.

I also want to learn how to make a proper PCB design in KiCad and order a cool purple PCB from OSH park.

I’m really enjoying this self-directed crash course in electronics. It’s really fun to learn so much while working on practical projects that we can use every day in our house. Thanks for reading if you got this far, it was fun to write up everything I learned!

SpaceCoast · September 7, 2022, 5:43am

This is such a great writeup. Thanks for putting all obstacles and setbacks too!

sachinss · September 7, 2022, 8:08am

Great write-up, and nice to see your positive attitude about how all the setbacks were learning opportunities for you!

Keep up the good work (and keep sharing!)

Mahko_Mahko · September 12, 2022, 8:37pm

Thanks for sharing the full journey of this! I find seeing other people stretch themselves with “learning projects” inspiring to do the same.

Mahko_Mahko · September 12, 2022, 8:54pm

Sounds like you’re sorted for AQI, but here’s some thoughts on AQI (other options). ESPHome based.

I’ve had good results with the PMS7003 for PM.
Ali Express PMS7003
https://a.aliexpress.com/_mt4wvBA

Senseair s8 has been good for me if you want CO2 (you could also use range hood for CO2 management!).
Ali Express AU $39.75

I check out price/quality options for PM and CO2 as they vary a lot!!

Alternatively, a lot of air purifier’s will also report on PM, so if you’re in the market for one of them, you could use that. My rangehood definitely isn’t enough to keep PM in range so my Air Purifier kicks in too to mop things up… Especially when frying.

ndbroadbent · September 13, 2022, 12:26am

Thanks for the tips! I received the ZigBee air quality meters that I ordered on AliExpress. They measure temp, humidity, CO2, PM25, Formaldehyde, and VOCs:

It looks like I have to set up some filters to remove outliers though: Filter - Home Assistant

Those sensors are great though, I’ve been thinking about making more of my own devices. Especially if I ever build or renovate a house. I would love to run wires for all the sensors behind the drywall, and power/connect everything via a LAN cable with POE.

Good tip about the air purifier as well, I have one in my office, but could probably use one in the kitchen/dining room as well.

Mahko_Mahko · September 13, 2022, 8:01am

Cool. I’m curious about that sensor now. Any chance you could open it up and take a photo of the chip? You’re looking for something similar to what’s at the bottom centre of this.

Cheap ones can be wildly inaccurate. But perhaps a relative value is enough to manage it .

I got one of these a while back mainly because I was curious about accuracy… It was sooooo bad.

I just pulled it apart out of curiosity then threw it in the bin…

AU $12.10 42%OFF | Multifunctional 5in1/6 in 1 CO2 Meter Digital Temperature Humidity Tester Carbon Dioxide TVOC HCHO Detector Air Quality Monitor
https://a.aliexpress.com/_mqyXvyg

Here’s a AQI project I did a while back.

krzakx · October 11, 2022, 5:59pm

Can you share with us what exactly current clamp did you buy?
I also bought a current clamp sensor that can measure AC current.

ndbroadbent · October 11, 2022, 9:34pm

I bought some of these from AliExpress, rated for 75A: https://www.aliexpress.com/item/4001286887821.html?spm=a2g0o.order_list.0.0.21ef18020qmj6S

I haven’t started working on it though

ndbroadbent · October 11, 2022, 9:48pm

Sure, here are some photos of the chip and sensor inside:

I don’t know if it’s accurate or not, but I do see some spikes in pm25 when we’re cooking. I basically just use it as a binary sensor to turn the fan on and off. I should also use the CO2 readings to let me know if I need to open a window if my office. So I think it’s totally fine for that purpose.

I just made some breakfast (fried eggs and sausages, etc.), and here’s the pm25 readings for our kitchen:

I have to use a filter to get rid of some outliers, but it works great apart from that:

    - platform: filter
      name: "Filtered Kitchen Air Quality Sensor pm25"
      unique_id: filtered_kitchen_air_quality_sensor_pm25
      entity_id: sensor.kitchen_air_quality_sensor_pm25
      filters:
        - filter: outlier
          window_size: 4
          radius: 4.0
        - filter: lowpass
          time_constant: 10
          precision: 2

Mahko_Mahko · October 12, 2022, 2:15am

Ta. Looks like it has a ZT3L.

https://developer.tuya.com/en/docs/iot/zt3l-module-datasheet?id=Ka438n1j8nuvu

But I think I had a brain fart and thought it was a wifi chip and might be ESPHome flashable. Sorry/whoops.

You might find CO2 levels which can build up in poorly ventilated rooms quite surprising. I did…

aruffell · February 14, 2023, 2:54pm

@ndbroadbent Thank you for the writeup! You did an awesome job at the project and sharing it too! There is a lot of talk about CO2 in the thread however I don’t believe that to be accurate and it may lead you to pick the wrong sensor. CO2 is what you exhale, while CO is what is produced with combustion. CO is Carbon Monoxide and CO2 is Carbon Dioxide. All “smoke alarms” detect Carbon Monoxide and not Carbon Dioxide. Air Quality sensors that have multiple sensors in them will detect high levels of Particulate Matter and CO2 as high levels means you have been in a room with everything shut for too long and it is time to open your windows…

Anyway, I have PM sensors in my upstairs office and I see spikes when I cook so I am guessing that a PM 2.5 sensor will be more than enough if you don’t sense power and turn it on when there is power draw.

I did a similar project too… I replaced my terrible “builder grade” rangehood with a much more powerful and better quality rangehood a while back. I too wanted to control it automatically so I tried to find one that integrated some remote control feature. Unfortunately, or fortunately depending how you see it, the only one I could find with anything similar was one that had a super cheap IR remote similar to the one supplied with cheap Chinese LED strips. The cheapest and easiest solution was to control the rangehood fan via IR so I ended up using a Broadlink IR emitter placed on top of my kitchen cabinets above the rangehood. Somehow the IR bounces around the kitchen and manages to control the rangehood below it. The downside of this is that HA and Alexa are unaware of the state of the rangehood, but the remote control via HA or Alexa works great (most of the time… sometimes Alexa/HA decides to refuse to send the command as it thinks it knows that the rangehood is already in that state - preliminary theory, not sure).

One thing I found is that while there are many brands of rangehoods, many are Chinese products rebranded for a number of manufacturers. One I used does a final test or minor assembly in the USA so they claim something that makes you think it is a US product but it really isn’t. All these rangehoods use control systems from one company (I forget the name) so I am planning on buying one so I have hack it without touching my rangehood. I’d like to do what you did but it may be a bit harder as my buttons are of the touch type (not mechanical). In part, the need to do this is that the extractor fan is around 1200cfm which is above a threshold that requires me to add a fresh air inlet to balance out the negative pressure it creates in the home potentially sucking back in CO from a fire place or gas appliances if in the house. The inlet would have an automated shutter to block the fresh cold/hot air from entering when the fan is off… or I could just have it be controlled by a pressure sensor which I think is how it is usually done.

jmsblah · August 27, 2023, 1:03am

I have a Zephyr Titan hood and I would like to do something similar. For me for now I would be already very happy if I can at least control the lights. Would you be able to give me any direction on where to start?