Smart Irrigation Prototype Running Out Of Time
Three students plan a smart irrigation prototype for the school garden: a capacitive soil moisture sensor, a microcontroller, a relay, a 12 V pump and a small solar panel. Their timeline is four weeks at two hours per week, with a budget of 2,500 rupees. In week 2 the moisture readings drift: dry soil reads 62% and freshly watered soil reads 58% on the same afternoon. A multimeter shows the sensor output changing whenever the pump runs. The relay occasionally clicks on and off rapidly. They have spent 1,700 rupees and have not started the written report, which carries 30% of the marks and is due in week 4.
1. Diagnose the drifting readings and the relay chatter. Separate the causes and say how you would confirm each one.
Read the worked answer
Three causes, tested separately.
Calibration. A capacitive sensor's output maps to moisture through a curve that must be measured, not assumed. Take readings in air, in dry soil and in saturated soil, and fit thresholds. '62% dry, 58% wet' means the mapping is wrong or the sensor is saturated. Check supply voltage during the reading, because these sensors drift with it.
Electrical noise. Readings change when the pump runs, which points to the motor's inrush current pulling the shared supply down and to switching noise on the ground line. Confirm by running the pump from a separate supply: if the readings steady, the cause is confirmed. Then use a star ground, decoupling capacitors and keep sensor leads away from motor leads.
Placement. A sensor near the drip point reads the water, not the root zone. Fix the depth, the distance from the emitter and the insertion method.
Relay chatter is the same noise problem plus a missing threshold band: add hysteresis, a minimum on-time in code, and a flyback diode across the coil.
2. Re-plan the remaining two weeks and the remaining budget. State what you would cut and why.
Read the worked answer
Two weeks, two hours each, 800 rupees left, and 30% of the marks still unwritten. Freeze the feature list now: one sensor, one pump, a manual override switch. No app, no enclosure, no second bed.
Week 3. Hour 1: rewire with separate supplies, diode, capacitor and star ground. Hour 2: calibrate against three known soil states and log twenty readings. In parallel, each member writes one report section at home (aim, circuit, data table), so the report is not left to the final week.
Week 4. Hour 1: ten-cycle run on the garden bed, recording pump on-time and moisture before and after. Hour 2: assemble results, finalise the report and rehearse the demonstration.
Spend at most 150 rupees on the diode, capacitor and connectors; keep the rest for printing. Assign one owner for the report and set an internal deadline of day 12, not day 14, because the report is the largest single block of marks still at risk.
3. How should the team evaluate whether the prototype works, and what should they report honestly?
Read the worked answer
Define success before running: the bed stays within a target moisture band, the pump does not run when the soil is already wet, and the system survives a week unattended. Measure against a simple reference, either a bought moisture meter read at three depths or a weighed soil sample dried and reweighed. Log time-stamped readings so the on/off behaviour can be shown, and count false triggers after the rewiring.
Report honestly: one sensor, two afternoons of data, no plant-growth evidence, and a known noise problem that was diagnosed and mitigated rather than eliminated. State the observed failure modes and the fix attempted, then the next step, such as a second sensor and a longer trial.
A report showing a working prototype with measured limits scores better than an unsupported claim that 'smart irrigation works'. It is also the accurate engineering position, and it gives the examiner evidence of method rather than optimism.