8000 Record irrigation EC O3 measurement programme by rafatahmed · Pull Request #31 · rafatahmed/Flahax · GitHub
[go: up one dir, main page]

Skip to content

R 8000 ecord irrigation EC O3 measurement programme - #31

Merged
rafatahmed merged 1 commit into
mainfrom
docs/ec-o3-measurement
Oct 19, 2026
Merged

rafatahmed merged 1 commit into
mainfrom
docs/ec-o3-measurement

Conversation

@rafatahmed
Copy link
Copy Markdown
Owner

Summary

Docs only. This records decision O3, the measurement programme for irrigation EC milestone M3, in the design (docs/superpowers/specs/2026-10-18-irrigation-ec-mccleskey-design.md, §11 and §13).

  • Evidence. M3 uses two sources:
    • published 25 °C tables, with KCl at about 0.001–0.05 mol/kgw and the single-salt tables already used by the model;
    • owner bench readings on five solutions made up by mass in one recorded water.
  • The five solutions. KH₂PO₄ at 1.000 g/kg (predicted 0.738 mS/cm), K₂HPO₄ inside the Pethybridge range, Tomatoes till First Cluster, Generic Wet Season (the widest bound) and Chilli (the high-EC end, beyond I = 0.014). Each prediction is computed before any reading, and nothing is refit.
  • Meter. Milwaukee MW802 in EC mode, with probe SE600.
    • Range 0.00–6.00 mS/cm; accuracy ±2 % of full scale, which is ±0.12 mS/cm.
    • One-point calibration at 1.413 mS/cm.
    • Automatic temperature compensation stays on; solutions are held at 25.0 ± 0.5 °C and checked with a separate thermometer.
    • Three readings per solution, with a repeat preparation for two of the solutions.
  • Reporting. The result is a facts-only sentence for those five solutions. It places the model range beside the meter band and includes misses. All other recipes keep the "accuracy not yet established" qualifier.
  • Gate. M3 code waits until these are recorded: the standard's lot and expiry, the thermometer, the water analysis, the K₂HPO₄ hydration, the balance, the date and the probe.

Note: at low EC the meter band is wider than the model range (about 16 % of 0.738 mS/cm), so the single-salt readings cannot resolve model errors smaller than that. The design states this.

Test plan

  • No code changed; the docs link check (release hygiene) covers this file in CI
  • CI green

M3 stays blocked until the MW802 blanks are filled and the five bench readings exist.