Automated irrigation system manualAutomated irrigation system manual

What this is and what it does

This is an automated watering system for plants grown in pots of soil-less media. Sensors buried in each pot measure how wet the media is and how much fertiliser salt it holds, and the software decides when and how much to water. That replaces watering by hand or on a ‘dumb’ timer.

The whole setup runs inside Home Assistant, free open-source home-automation software that you run yourself, with no cloud account required for the core system. It controls 6 grow tables at once. The system is built in three layers you can adopt in order: a time-based irrigation package and a sensor-driven crop-steering integration that both run today, plus an optional advanced layer (AppDaemon) for fully autonomous decisions.

The three-layer design lets you start simple and add intelligence later without rebuilding. Run it on a plain timer for the first week while you learn to trust the sensors, then switch on crop steering once the readings look sane.

• Replaces hand-watering and plain timers with sensor-driven decisions across 6 grow tables
• Runs on Home Assistant; no cloud account or extra software needed for the core system
• Three layers: time-based package, crop-steering integration, and optional AppDaemon autonomy
• Start timer-based and add intelligence later without rebuilding

Accuracy, self-review, and grain-of-salt notes

Key terms, defined from scratch

These are the words this manual uses constantly. You don't need to memorise them, each one comes back in context.

The hardware: what gets wired up

The brain of the valve control is a KC868 E16S relay board connected over Ethernet and running ESPHome firmware. It physically switches every valve: the 6 tables plus the mains, mainline, and manifold valves.

Each table has a substrate probe that reads VWC, EC, and temperature over SDI-12, a simple digital protocol for soil sensors. The room is also covered by 4 environmental sensors, 3 CO2 sensors, 4 AC units, a humidifier, and 4 dehumidifier relays. There is no pump switch: the pump auto-starts on a pressure switch the moment the mainline valve opens, so water flows tank → mainline → table dripper → plant.

The phase logic: P0 to P3

Crop steering splits each lights-on day into four phases that control how the plant uses water and nutrients. A controlled dryback, letting the media dry by a set amount before watering again, is the signal that steers the plant toward vegetative root growth or generative flower production.^[2]

P0 (morning dryback) is no watering at all, letting the media dry from overnight. It moves to P1 once VWC drops by the target amount (50% in vegetative, 40% in generative) or after a 120-minute cap. P1 (ramp-up) gives progressively larger shots, starting at 2% of substrate volume and rising 0.5% each shot, 15 minutes apart, up to 10 shots, until VWC hits roughly 65%.

P2 (maintenance) holds the media steady with top-up shots when VWC drops below 60%, and nudges that threshold up or down based on EC: if the EC ratio exceeds 1.2 it waters more to flush salts, and below 0.8 it lets the media dry to concentrate nutrients. P3 (pre-lights-off) is emergency-only watering below 40% VWC before an overnight dryback resets the cycle.

Environment control and the safety net

The system manages room climate as well as watering. Dehumidifiers turn on when humidity is 5% above target (the 4 relays stagger on with 10-second delays to avoid an inrush spike) and off 2% below. The humidifier mirrors that logic. CO2 injects when 50 ppm below target and only during lights-on, always closing at lights-off.

Several independent safety layers run in parallel so that one failure never leaves valves open. The master gate only allows irrigation when the system is enabled, not in maintenance mode, no leak detected, tank OK, and the clock is inside the time window. A valve watchdog force-closes any table valve open longer than 3 minutes, a daily 3 AM audit closes everything as a reset, and maintenance mode shuts all valves instantly.

Commissioning: getting it running from scratch

Bring the system up in order so each layer is proven before the next one is switched on. Rushing straight to crop steering on unverified hardware is the fastest way to flood a table.

1. 1
   Verify hardware

   Confirm every probe and the relay board show Online in ESPHome. Check a table VWC sensor reads a sensible number (e.g. 47.4), not a dash or zero.
2. 2
   Test every valve

   Manually click each valve on, then immediately off, to confirm it responds. Toggle the mainline briefly to confirm the pump auto-starts on pressure.
3. 3
   Set the window

   Set the irrigation window (e.g. start 08:30, 30 min after lights-on; end 18:00, 2 hours before lights-off), interval 60 minutes, duration 60 seconds.
4. 4
   Enable the system

   From the Command Center, enable the system, switch maintenance mode off, and enable only the tables you actually want to water.
5. 5
   Run one test cycle

   Run a single cycle and watch the valves open and close one at a time. Confirm water reaches the drippers and runoff appears.
6. 6
   Set climate & steering

   Set day/night temp and humidity targets and a CO2 target. Optionally choose a steering mode and crop profile once watering is proven.

Day-to-day operation and the dashboard

Daily checks are quick. On the Command Center, all 6 tables should read VWC between 30 and 70%, EC in your target range (typically 2 to 6 mS/cm by stage), and substrate temperatures 20 to 26 °C, with every safety indicator green.

On the Trends tab the VWC graph should show a sawtooth: a gradual drop, then a sharp rise after each irrigation. A flat or only-falling line means watering is not actually happening. That is your first warning sign, before any error message appears.

• Healthy daily readings: VWC 30–70%, EC ~2–6 mS/cm, substrate temp 20–26 °C, safety all green
• A non-sawtooth VWC trace is your earliest signal that something is wrong
• Enable or skip tables via the Enabled toggle in Zone Control
• Emergency stop: toggle Maintenance Mode on (closes all valves) or call the emergency-stop script

Troubleshooting and tuning

Most problems are diagnosable from a short checklist. Work top to bottom, because the cheapest checks (is it on? is the window open?) catch the most cases.

If irrigation will not run, confirm the system is on, maintenance mode is off, the clock is inside the window, at least one table is enabled, and the irrigation-allowed binary sensor reads on. If VWC or EC shows Unavailable, check the ESPHome device is online and reload template entities. If the raw probe value is also missing, the probe has lost its connection. Reliable VWC depends on a working dielectric sensor, so a flatlined raw value points at the probe, not the software.^[3]

Realistic expectations and limits

Two more honest limits. Without a leaf-temperature sensor, VPD falls back to room air temperature and is less accurate. The fully autonomous P0–P3 phase transitions need the optional AppDaemon layer, which is not deployed by default. Budget about a week of supervised, timer-based running before you trust the automation.

Start by reading one normal day on the Trends tab, then change one thing at a time. For the biology behind the dryback and the P0–P3 rhythm, read the crop steering in coir paper; for how the probe actually measures the root zone, see the root-zone sensor guide.