The Humidity Problem

2026-03-11

The Shelly fix worked. Temperature is a closed chapter. Humidity is the new front.

Zero crashes

Yesterday’s diagnosis was correct. At 2kW, the Shelly’s internal temperature climbed to 85°C and crashed every hour. At 1250W, it stabilises at 60°C — 25°C below the thermal protection threshold.

The last crash was at 00:24, before the power reduction was applied. The cooldown procedure ran at ~01:05: stop thermostat service, turn relay off, poll internal temperature every 30 seconds until it dropped below 50°C, restart thermostat at the new 1250W setting. From 01:28 onwards — zero crashes in 21 hours.

The tradeoff is visible in the cycle times. At 2kW, the heater cycled 15 minutes ON / 15 minutes OFF. At 1250W, it runs 35–45 minutes ON / 15–20 minutes OFF — longer heating periods to compensate for lower power. The thermostat adapts automatically; it doesn’t care how long each cycle takes, it just holds the bands.

Temperature: solved

Twenty-four hour stats:

• Min: 21.5°C (during the cooldown procedure at 01:28)
• Max: 24.0°C (morning solar warming around 11:00)
• Mean: 22.8°C
• 100% of readings within 20–25°C

Excluding the cooldown period, the thermostat held 22.5–23.1°C with clean cycling — ON at 22.5, OFF at 23.1, no excessive relay toggling. The morning solar gain pushed temperature above setpoint for several hours (09:00–14:00), during which the heater stayed off and the room cooled naturally.

Temperature compliance has been 100% for two consecutive days. This variable is controlled.

Humidity: 96.7% → 52.0%

This is the story of the day. Overnight, humidity was healthy at 43–46% — well within the 40–60% target range. Then it fell off a cliff:

The agent flagged WARNING status from 10:54 onwards — 30 consecutive entries for the rest of the day. It correctly identified the problem but had no way to fix it. There’s no humidifier to control.

The FUMMDUS standalone hygrometer confirmed the readings: 32% with a sad face and “DRY” icon.

Why it crashed: warmer air has lower relative humidity at the same absolute moisture level. The morning solar warming (temp rising from 22.8 to 24.0°C) drove relative humidity down. But the humidity didn’t recover when the temperature dropped back to setpoint in the afternoon — suggesting the room genuinely lost moisture, not just a relative humidity artifact. The heater compounds this: every heating cycle pushes warm, dry air through the room.

Yesterday, humidity only dipped below 40% for 95 readings (3.3%) — all at 39.3%, barely under the line. Today, humidity was below 40% for 48% of the day, reaching as low as 31.7%. The difference: yesterday was more humid outside (March weather is variable), and yesterday’s readings started from a higher baseline.

Combined suitability

The thermostat eliminated temperature as a variable. Humidity is now the entire gap. On humid days the system scores >95%. On dry days it drops to 50%. Without active humidity control, the score depends entirely on outdoor weather — which is not acceptable for tissue culture.

What humidity control looks like

Three options, in order of complexity:

1. Water tray — a shallow tray of water near the growing area. Passive evaporation. Free, zero maintenance, but uncontrolled and probably insufficient.

2. Enclosed propagation chamber — a clear plastic container or mini-greenhouse around the cultures. Traps transpired moisture, creating a humid microclimate independent of room conditions. This is how most home tissue culture setups work. Cheap, effective, but limits airflow.

3. Small humidifier — an active device on another smart plug, controlled by the thermostat or a similar bang-bang loop on humidity. Most control, but more hardware, more failure modes, more things to crash.

The propagation chamber is probably the right first step. It decouples the cultures from room humidity entirely — a 32% room can have 80%+ inside a sealed container. The sensor would need to move inside the chamber to measure what the plants actually experience.

Where things stand

The system is half-ready. Temperature, the easier variable, is locked in. Humidity, the harder one, needs either active control or a physical enclosure. No cultures go in until humidity is solved — a 32% environment would desiccate cuttings within hours.