Here’s a simple, reliable way to use a soil-moisture [sensor](https://www.ampheo.com/c/sensors) so each plant gets the amount of water it actually needs.

**1) Pick the right sensor**

* Capacitive soil-moisture sensor (analog or I²C) → best for hobby/DIY (less corrosion than resistive “forks”).
* Tensiometer / matric-potential sensor (kPa) → most agronomy-correct, but pricier.
* Tip: Whatever you choose, plan to calibrate per soil mix (potting soil vs cactus mix) because raw numbers aren’t universal.

**2) Calibrate your sensor (5–10 min per soil)**

1. Stick the sensor in air-dry soil → note reading = DRY.
2. Fully saturate the soil, let it drain to field capacity (30–60 min) → note reading = WET.
3. Convert later readings to a %:

`moisture_% = 100 * (reading - DRY) / (WET - DRY) // clamp 0…100`

For capacitive probes: wetter soil usually = higher reading; confirm on your unit (some are inverted).

**3) Choose setpoints per plant (as % of your calibrated scale)**

Add hysteresis (two thresholds) so pumps don’t chatter.

**4) Hardware sketch (per pot)**

* MCU: [Arduino](https://www.ampheo.com/c/development-board-arduino)/ESP32.
* Sensor: Capacitive probe → analog pin (use 3.3–5 V as specified).
* Pump/valve: 5–12 V peristaltic pump or [solenoid](https://www.onzuu.com/category/solenoids).
* Driver: N-[MOSFET](https://www.onzuu.com/category/fets-mosfets) or [relay](https://www.onzuu.com/category/relays) + flyback [diode](https://www.onzuu.com/category/diodes) ([1N5819](https://www.onzuu.com/search/1N5819)/[1N4007](https://www.onzuu.com/search/1N4007)) across the load.
* Power: Separate supply for pumps; common ground with MCU.
* Scaling up: Use an [ADS1115](https://www.onzuu.com/search/ADS1115) (I²C ADC) or analog multiplexer to read many pots.

**5) Minimal Arduino example (multi-plant, hysteresis)**

struct Zone {
  uint8_t ain, pumpPin;
  int DRY, WET;                  // from your calibration
  float lowPct, highPct;         // plant setpoints
  unsigned long maxMs;           // safety cutoff per cycle
  bool pumping;
  unsigned long tStart;
};

Zone z[] = {
  // ain, pump, DRY, WET,  low%, high%, max pump time
  {A0, 5, 1780, 2650, 25, 35, 20UL*1000},   // succulent
  {A1, 6, 1820, 2700, 45, 60, 25UL*1000},   // tomato
};

float pct(int raw, int DRY, int WET){
  float p = 100.0f * (raw - DRY) / float(WET - DRY);
  if (p < 0) p = 0; if (p > 100) p = 100;
  return p;
}

void setup(){
  Serial.begin(115200);
  for (auto &q : z){ pinMode(q.pumpPin, OUTPUT); digitalWrite(q.pumpPin, LOW); }
}

void loop(){
  static unsigned long last = 0;
  if (millis() - last < 1000) return;   // ~1 Hz control loop
  last = millis();

  for (auto &q : z){
    // simple median of 5 reads to reduce noise
    int rmin=4096,rmax=0,sum=0;
    for(int i=0;i<5;i++){ int v=analogRead(q.ain); rmin=min(rmin,v); rmax=max(rmax,v); sum+=v; delay(5); }
    int raw = (sum - rmin - rmax)/3;
    float m = pct(raw, q.DRY, q.WET);

    if (!q.pumping && m <= q.lowPct) {
      q.pumping = true; q.tStart = millis();
      digitalWrite(q.pumpPin, HIGH);
    }
    if (q.pumping && (m >= q.highPct || millis() - q.tStart > q.maxMs)) {
      q.pumping = false;
      digitalWrite(q.pumpPin, LOW);
    }

    Serial.print("zone:"); Serial.print(&q - z);
    Serial.print(",moist:"); Serial.print(m,1);
    Serial.print(",pumping:"); Serial.println(q.pumping);
  }
}

* Replace DRY/WET with your measured values.
* Tune lowPct/highPct to plant needs.
* maxMs prevents over-watering if a tube pops off.

**6) Practical tips**

* Depth matters: place the probe near the root zone, not just the top 2 cm.
* Avoid fertilizer spikes: EC/salinity affects capacitive reads; re-calibrate after heavy feeding or use distilled water for calibration.
* Temperature drift: readings shift with temp; take measurements at a similar time of day or add a temp sensor and compensate.
* Filter & average: small median/EMA filtering smooths noise; don’t over-filter—water decisions should lag by minutes, not hours.
* Watering window: restrict pumps to daytime to avoid fungus (unless urgent).
* Maintenance: wipe probes monthly; recalibrate each season or when repotting.

**7) Scaling to many plants**

* One sensor per pot = best.
* Or rotate one [sensor](https://www.ampheoelec.de/c/sensors) across zones to learn thresholds, then water by time/weight (load cells under pots work great) and re-check weekly.