"Installed an auto-doser for pH — now I don't need to check." An auto-doser that you "set and forget" is a pump that can inject a litre of acid into the reservoir while you sleep — if the pH probe gets fouled and reads 8.0 when the actual value is 5.5. pH and EC automation delivers a real advantage — parameter stability that a human physically cannot maintain through manual measurement. But that advantage is only realised alongside a correctly tuned PID, reliable sensors, and fail-safe scenarios that stop the system on anomaly rather than continuing to dose.
Quick glossary: PID (Proportional-Integral-Derivative) — a control algorithm that calculates a corrective action based on three components: proportional (current error), integral (accumulated error over time), and derivative (rate of error change); enables smooth, precise parameter control around the target value without overshoot. Auto-dosing — automatic delivery of acid, base, or nutrient concentrate via a peristaltic pump triggered by a sensor signal. Fail-safe — a set of conditions and scenarios under which the system automatically stops dosing or enters a safe state to prevent catastrophic parameter deviation.
Why Simple "If pH > 6.2 → Dose Acid" Logic Is Dangerous
The simplest automation is threshold control: pH crosses a boundary → pump turns on → doses until pH returns within range. This looks logical but has a fundamental problem: lag and overshoot.
The pump has been dosing acid and pH is moving down — but the pH probe shows the change with a 30–120 second delay (mixing and response time). During that delay the pump kept dosing. pH drops below the target — pump stops. Now pH is too low — the automation starts dosing base. The system oscillates around the target, cycling "dose acid — dose base" — and each cycle adds unnecessary ions to the solution.
PID solves this through three components: the proportional term reduces the dose as the target is approached, the integral term compensates for systematic offset, and the derivative term responds to the rate of change and slows dosing before pH overshoots the target. A correctly tuned PID gives smooth pH maintenance with minimal oscillation and minimal reagent consumption.
Fail-Safe: What Must Stop the System
Fail-safe is not paranoia — it is a mandatory component of any automation controlling chemically active substances. Four scenarios that must automatically halt dosing:
Sensor reading outside physically possible values. pH < 2.0 or pH > 12.0 — the sensor is faulty or disconnected. Dosing under these conditions means controlling on a fictitious signal. Stop dosing and trigger an alarm — mandatory.
Accumulated dose exceeds the session maximum. Even with "normal" sensor readings — if the system has dosed more than X ml per hour without reaching the target, something is wrong: either the sensor is fouled, the reagent is empty and the pump is pushing air, or there is a leak. A per-session dose limit is the key protection.
Time to reach target is exceeded. If pH has not reached the target within 10–15 minutes of normal dosing — either the target is unachievable (e.g. due to very high alkalinity) or the sensor is not reflecting reality. Automatic pause and a diagnostic alert.
Sharp, rapid parameter deviation. pH shifts by 1.0+ within 2 minutes without a doser command — mechanical damage, contamination by a foreign solution, or sensor failure. Stop all dosing and trigger an alarm.
PID Tuning for pH and EC: The Differences
pH and EC require different PID approaches because of their different dynamics:
pH — a non-linear parameter (logarithmic scale), responds slowly and with significant lag during mixing. Typical settings: a wide deadband (±0.05–0.1 pH around the target where no dosing occurs), low proportional gain, moderate integral. Derivative — use with caution: with a noisy probe signal, the D-term can cause erratic pulses.
EC — linear and predictable, responds faster than pH. But dosing nutrient concentrate raises EC in steps and requires a pause for mixing between doses. Strategy: dose in small increments with 30–60 second pauses rather than continuous dosing to target.
Three Mistakes That Cost the Most
Trusting automation without regular sensor verification. A pH probe coated with mineral deposits reads 7.5 when the actual value is 5.8 — the automation continuously doses acid. After 6 hours, pH is 3.0 and the root system is damaged. Sensor checking and calibration with automation needs to happen more frequently than with manual control — not less.
Not setting per-session dose limits. "Let it dose until the target is reached" — without a limit, this is an open licence for catastrophe on sensor failure. A dose limit per session and per day is a mandatory parameter for any auto-doser.
Using default PID settings without verifying system response. Factory PID settings are calibrated for an "average system" that does not exist in reality. After connecting: make several manual test deviations to pH and observe how the system returns to target. If it oscillates — increase the deadband or reduce the proportional gain. If it responds too slowly — increase P or I.
How to Know the Automation Is Set Up Correctly
pH is maintained within ±0.1 of the target with no visible "acid-base" oscillations. EC matches the recipe within ±0.1 mS/cm. Reagent consumption is predictable and stable week over week — a sharp increase without changes to plants or recipe is a signal of a problem (leak, fouled sensor, changed water alkalinity). The fail-safe has been triggered at least once in test mode and confirmed to stop the system correctly.
For deeper understanding: Sensor and Monitoring System: What to Measure, With What, and How Often — the measurement and calibration baseline that must be in place before connecting automation — so it controls reality, not the readings of a faulty sensor.