Energy, learning and calibration curves

Nameplate vs learned airflow, the anemometer method, CFM ownership, Phase A/B climate learning, the space energy model and tariff cost, and what soft cal does and does not do.

A fan's CFM starts as nameplate: duty % × the rated flow. That is a guess, drawn dashed. Learning replaces it with measured points: run the fan at a known step, read an anemometer in the duct (m/s × duct area), accept the point. The brain interpolates a curve per fan and the air path and Sankey switch to allocated flow. CFM cal ownership says which source is in force. A per-fan airflow correction (Settings › Climate) scales a curve without a session.

Every fan is a row with a tier: driven (a hub channel sets its level), switched (on/off through a smart plug) or known (not controlled, assumed running at its rating). An added fan on a hub channel can be measured like a kit fan; switched and known fans have no levels to step, so they stay on nameplate. A fan matched to a DSC-KitLib product offers its CFM rating and duct size as maker presets.

• Phase A learns how each tent responds to a rung as an exponential moving average (α, minimum samples shown on the card).
• Phase B may propose ladder wait times from that response. It is opt-in, has a lock, and never touches failsafe, min-off or fans.

Learning is a live calibration: Start holds live actuators until Finish.

Space energy is per tent: fixture nameplate watts × hours × dim, plus appliance duty × rated power, priced by the tariff bands. Watts come from the matched DSC-KitLib product (read-only) or from the Appliance power, Lamp power and Fan power cards on Settings › Devices; a device left unset is excluded from the estimate rather than counted as zero. Zigbee energy meters replace the estimate for whatever they measure. Learning flags timing or energy outliers only when they become the norm.

A calibration of the SF1000's brightness response (PPFD or lux at several levels) so percent on the desk maps to real output. It refines the lamp, not the maker PPFD map.

Soft calibrate shifts a probe's Got offsets on the brain (tap water → after water, or a known pH) so the desk reads honest without flashing anything. Lab wet calibration writes a real calibration stamp to the probe's ESP with buffer solutions. The probes use one calibration plane; N, P and K are not independent measured channels and are not soft-calibrated.

Guide contents  ·  How it works  ·  Troubleshooting  ·  Digital Stealth Care

DSC-Brain user guide · Energy, learning and calibration curves · this page is edited on the site; the Notion original was retired after the mirror