.
My better half and favorite co-author very much likes her houseplants. It is not unusual for her to come home and unapologetically present a new acquisition which she will manage to find space for. Now, while I fully support this hobby, I am not particularly good at it, and so my involvement in it usually only rises to the level of “hold this so I i can put some soil around the roots”. That is, until we started talking about the important variables to monitor in order to keep a plant happy.
One of my hobbies is electronics. I love to tinker with Arduinos and the like. So once the important variables (moisture, temperature, light, and humidity) were identified, it was not long before we got the idea to build a thingy to keep track of these. It also doesn’t hurt that my better half is also an econometrician, and so is fully on board with having more data to work with.
So I set out to build this thingy with the following goals:
- Log soil moisture, temperature, light, and humidity
- Push this information to something easily accessible, like an online spreadsheet or web page, and
- Have a couple of LEDs to communicate some of its operations.
I did this in a few stages. Here they are!
Version 1: Arduino nano on a breadboad
The first attempt was a proof-of-concept with just one one component. For this, I used the Monk Makes Plant Monitor, which measures temperature, humidity, and soil moisture. Three out of four isn’t too bad for the first try! This was a relatively easy hook-up:
| Arduino | Plant monitor |
|---|---|
| 3.3V | 3V |
| GND | GND |
| TX | RX, through a 1k resistor |
| RX | TX |
And then I just ran the example code supplied in the above link. Everything worked, moisture went up when I dropped it in a cup of water, humidity and temperature went up when I breathed on it, all good!
The problem with this thingy was that it had to be plugged into my laptop for us to be able to read the data. As this was not a good long-term solution, I asked my better half if she would rather have the readings shown on a screen, or posted to somewhere on the web. She opted for the latter, so on the next iteration, I focused on connecting it to the internet.
Version 2: Connecting to the interwebs
As much as I love Arduinos, the cheap ones don’t connect to the internet. This meant for the second thingy, I would have to choose another microcontroller to use. Fortunately, I had a whole lot of WeMos D1 mini clones that would do this perfectly. I also had acquired some other capacitave soil sensors that didn’t require serial communication, and were substantially cheaper. This would mean I’d need to also have temperature and humidity sensors, so I went with the AHT10, which does both over I2C with 3.3V logic. This last part is important becuase the D1 mini pins work with 3.3V, not the 5V of Arduino.
On top of these, I added a momentary switch for recording events (e.g. so we could press it when we watered the plant), and two LEDs for communicating different things about the status of the thingy.
For the software, I found this tutorial on how to send data from an ESP8266 module (of which the D1 mini is one) to a Google sheet. This seemed like a good solution, as we could both have access to the data easily.
Version 3: Getting the most out of one analog pin
From here, the only thing left to do was add in a light sensor. It is fairly easy to set up a circuit that responds to light: use a light-dependent resistor in series with a (regular) resistor, and read off the voltage between them. However here the D1 mini poses a problem: it only has one analog pin, and I was already using it to measure moisture. One possible solution was to use a multiplexer integrated circuit, like a 4051. I had some of these lying around, but it seemed like overkill: it could handle eight analog inputs (I only needed two), and needs three digital inputs to manage, and that would require more wiring. Fortunately, this tutorial was exactly what I needed: basically, instead of powering the two sensors with the 3.3V pin so that they were always on, you use separate digital pins for power. This means that you can always power down a sensor, so its voltage goes to zero. Diodes then ensure that the signals only go to the analog pin, and don’t get drained into the other sensor.
Here is what I ended up with for the hardware:
.
That is, in order to take the two analog readings:
- Set pin D0 to HIGH and pin D5 to LOW. This fires up the moisture probe and turns off the light meter. Diode D1 blocks any current from draining into the light meter.
- Read the voltage at the analog pin (A0). This is the moisture reading.
- Set D0 to LOW and D5 to HIGH. This turns off the moisture meter and turns on the light meter. Diode D2 blocks any current from draining into the moisture reading.
- Read the voltage at the analog pin (A0). This is the light reading.
Some other features of this design that I have are:
- LED1 and LED2 light up when the light and moisture meters are being read, respectively. I put these in so that I could tell if the wiring and software were working properly.
- LED3 is there to communicate other things. At the moment I am just blinking it when the microcontroller is sending data to the Google sheet.
- I use the momentary switch (S1) for logging events. These are things like watering the plant, moving a humidifier in or out of a room, etc.
Finally, here is the Arduino code I used to program the microcontroller.
I set the Thingy to log data every ten minutes, which is probably overkill for this project, but I got impatient.
A look at the data
Here is what the data look like:
.
Some things to note about these:
- We first watered the plant being monitored by PlantMonitor0001 on Dec 18th. You can see this in the plot as a large drop in the moisture reading. The voltage the moisture sensor gives is negatively related to moisture, reading about 800 when dry and in the air, and below 400 when submersed in a glass of water.
- We watered PlantMonitor0002 on Dec 22nd. I think the slight increase in moisture reading before it dropped out after watering was due to shifting the plant around.
- The vertical dashed lines show when the button was pushed on the monitor. We use this to mark significant events for the plants. For example the red vertical lines corresponding to large increases in humidity mark when my better half misted some water into the room.
- PlantMonitor0001 does not have a light sensor, so we don’t get any light data from that
- PlantMonitor0002 is in the same room as our house’s thermostat, but PlantMonitor0001 is not. This really highlighted to us how much of a temperature swing we can get in this room.
- The drops in light measurements each day from about 800 to about 650 are when the grow lights turn off, and the drop down after then is when we turn our living room lights off.
Next steps
Probably the thing I’d like to fix the most out of these is a connection problem I am having with out WiFi. Roughly every 24 hours, the thingies get stuck in their transmit stage (I know this because of the status of the LEDs on the boards) and just stop sending data. Ironing this out would mean I would not need to go and reset them every day or so. Resetting is not hard to do, but it does mean that I need to be watching the data come in somewhat regularly so that I notice it quickly. Ideally these things would be set-and-forget, and we would only have to look at the data when we wanted to.
Once we have collected enough data, I would like to re-program the boards to send notification reminders for watering.