You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: README.md
+2-2Lines changed: 2 additions & 2 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -2,9 +2,9 @@
2
2
3
3
An ESPHome reader for the **INFWIN MT22A SDI-12** substrate probe, with checked water-content calibration, bulk EC, temperature and observed dryback trends. Runs locally on ESP32/M5Stack hardware with a web page, Home Assistant, optional MQTT and CSV logging.
4
4
5
-
**Open the [live substrate and calibration setup desk](https://jaketherabbit.github.io/TDR-Sensor/).**It covers cubes, cubes on shared slabs, coco containers, metric/custom sizes, weighed calibration records and an actual-size printable placement sheet. The [field guides](https://jaketherabbit.github.io/TDR-Sensor/guides/) are readable on the site too.
5
+
**Open the [live substrate calculator and calibration wizard](https://jaketherabbit.github.io/TDR-Sensor/).**Configure cubes, shared slabs or coco containers; switch all calculator measurements between metric and imperial; follow diagrams for each system; and export sensor settings or board-specific ESPHome YAML. The calibration wizard supports a wet reference or weighed A/B/C checks, with downloadable records and actual-size A4/Letter placement templates. The [field guides](https://jaketherabbit.github.io/TDR-Sensor/guides/) are readable on the site too.
6
6
7
-
For offline use, download this repository ZIP, extract it and open [tools/setup/index.html](tools/setup/index.html) in a browser. Calculator entries stay in your browser; the site does not connect to or control a sensor.
7
+
For offline use, download this repository ZIP, extract it and open [tools/setup/index.html](tools/setup/index.html) in a browser. Setup and calibration records are saved in this browser and can be exported/imported as a JSON project. The site does not connect to or control a sensor. Generated YAML provides explicit setup/import buttons; references are never written automatically at boot.
8
8
9
9
Copy file name to clipboardExpand all lines: docs/CALIBRATION.md
+1-1Lines changed: 1 addition & 1 deletion
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -1,4 +1,4 @@
1
-
# Calibrate without pretending saturation is 100% VWC
1
+
# Calibrate the MT22
2
2
3
3
There are two different tasks: saving a repeatable wet reference, and estimating actual volumetric water content from independent weights. Version 3 keeps them separate. All captures are available on the node's web page and in Home Assistant; calibration does not require reflashing.
Copy file name to clipboardExpand all lines: docs/CONFIG.md
+28Lines changed: 28 additions & 0 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -74,3 +74,31 @@ For MQTT, uncomment the `tdr_mqtt.yaml` package and fill in `mqtt_broker`, `mqtt
74
74
75
75
76
76
CSV logging: wide format now records each field's observation age, blanks readings after `--max-age` (default 120 seconds) or a disconnected stream, and refuses to append a mismatched header. Start a new CSV after upgrading. Use long format for entities that appear after the initial snapshot; wide mode warns rather than silently dropping new columns. Adjust maximum age to the actual reporting cadence, not the desired irrigation interval.
77
+
78
+
## Generate a configuration in the calculator
79
+
80
+
The [setup wizard](../tools/setup/index.html#volume) exports the selected board,
81
+
data pin, SDI-12 address, sampling cadence and substrate profile. It offers a
82
+
sensor-settings report, a complete device YAML and a `secrets.yaml.example` file.
83
+
Fill the secrets locally; the website never asks for them.
84
+
85
+
The YAML pins the v3 packages. It adds **Apply wizard setup**, which explicitly
86
+
sets the profile and capture/check limits and starts a new capture window. Saved
87
+
ESPHome preferences take precedence over initial defaults, so press this button
88
+
after installing the configuration when you want to apply those settings.
89
+
90
+
If the wizard contains a valid recorded wet reference or checked A/B/C set, the
91
+
export also adds **Import wizard references**. Use it only for the same physical
92
+
probe, medium and placement. It requires Calibration mode, the expected profile,
93
+
ten fresh readings and a stable RAW window. It replaces references only when
94
+
pressed; rebooting never reimports them. Verify all saved values afterward,
95
+
turn Calibration mode off, and wait ten seconds before disconnecting power.
96
+
97
+
If references were already captured on the device at their original moisture
98
+
levels, simply verify them against the wizard's report. Do not recapture old A/B/C
99
+
values at a different current moisture level. Substrate litres belong in the
100
+
volume record and steering app; they are not substituted for measured calibration.
101
+
102
+
The wizard accepts the probe's actual address character. The pinned driver uses
103
+
indices 0–9 for digits, 10–35 for a–z, and 36–61 for A–Z. The exporter performs that
104
+
conversion; its YAML comment preserves the address character for comparison.
Copy file name to clipboardExpand all lines: docs/PLACEMENT.md
+7-3Lines changed: 7 additions & 3 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -23,12 +23,16 @@ The factory does not publish an MT22/Prestige-specific minimum distance from eve
23
23
24
24
## Cubes alone
25
25
26
+
27
+
26
28
Use a repeatable side insertion into a sufficiently large block, with the rod row level. A midpoint height is an initial mapping position, subject to the same boundary and representativeness checks. A 150 mm-wide Hugo can accommodate the 88 mm contact face geometrically; that is not proof of whole-block accuracy.
27
29
28
30
**Small propagation blocks need a different sensor or validation arrangement.** A 75 mm-wide cube cannot accommodate the MT22's 88 mm-wide face in this orientation. Do not cut a larger hole or leave an outer pin in air. Some 100 mm blocks also leave very little lateral clearance; check the sensing footprint rather than relying solely on rod length. A physically smaller substrate-specific probe may be a better fit.
29
31
30
32
## Coco / peat containers
31
33
34
+
35
+
32
36
Measure at a recorded depth in the actual packed, rooted medium, away from the emitter stream, stem, drainage layer and container boundary. Map more than one depth or compare representative pots before choosing a standard position. The middle of the filled height is a possible initial comparison point, not a universal coco rule.
33
37
34
38
Where practical, place the sensor during filling and pack hydrated medium around all rods consistently without creating cavities or compacting a special dense pocket. On an established container, make a minimal side opening if the container design allows it and avoid repeated insertion through roots. A curved pot wall does not provide a flat 88 mm contact surface; verify full substrate contact along every rod instead of forcing the housing against the wall. Keep the chosen position fixed as moisture changes; coco shrinkage can create air gaps.
@@ -37,9 +41,9 @@ Different pot heights and media mixes need their own calibration and placement r
37
41
38
42
## Print and use the template
39
43
40
-
-Print A4,**Actual size / 100%**, with Fit, Shrink and browser headers/footers disabled.
41
-
- Measure both perpendicular 100 mm scale bars. Aim for no more than 0.5 mm discrepancy; reject a scaled print.
42
-
- Select the correct slab-height page. Align the base datum with the bottom face of the rockwool.
44
+
-The fixed slab PDF uses A4. The calculator generates A4 or US Letter sheets for your selected dimensions and units. Print**Actual size / 100%**, with Fit, Shrink and browser headers/footers disabled.
45
+
- Measure both perpendicular check bars: 100 mm in metric or 4 in in imperial. Aim for no more than 0.5 mm (0.02 in) discrepancy; reject a scaled print.
46
+
- Select the correct slab-height page. Align the base datum with the bottom of the medium. For a custom sheet marked Tall container, first mark the chosen height with a ruler and align its centreline; that sheet has no base datum.
43
47
- Transfer your actual sensor's three pin positions onto the printed centreline using scrap backing. Remove the paper before final insertion; do not drill oversized holes in the substrate.
44
48
- The printed 88 × 26 mm outline locates the housing. It does not calibrate VWC or prove the chosen location is representative.
@@ -25,7 +26,7 @@ The setup tests run the same JavaScript used in the page. They cover real Hugo d
25
26
26
27
Open `tools/setup/index.html` from an extracted download and verify it without a network connection. Check cube-only, cube-on-slab, slab-only and coco calculations; invalid input must clear the previous answer. Add a weighed record and download the CSV. Check the layout at desktop and phone widths.
27
28
28
-
The supplied two-page PDF has A4 pages and a dimensioned 88 × 26 mm face. The custom print sheet uses an A4 SVG in physical millimetres, including perpendicular 100 mm scale bars. PDF geometry checks cannot compensate for a printer driver scaling the page: always measure both bars on the physical print.
29
+
The supplied two-page PDF has A4 pages and a dimensioned 88 × 26 mm face. The custom template supports A4 and US Letter, with physical SVG geometry in millimetres regardless of display units. Its perpendicular check bars are 100 mm in metric mode or 4 inches in imperial mode. Tall containers use a ruler-marked centreline instead of pretending the entire height fits on one page. PDF geometry checks cannot compensate for a printer driver scaling the page: always measure both bars on the physical print.
29
30
30
31
## What remains a field check
31
32
@@ -45,3 +46,24 @@ There is no automatic deployment or flashing step in these checks. A successful
45
46
- Browser interaction checks passed for all four systems, preset/custom edits, gallons, tapered pots, invalid inputs, weighed calculations, CSV download, desktop/phone layout and a custom A4 print sheet.
46
47
- The custom printed PDF had one A4 page; its contact-face rectangle measured 87.999 × 26.000 mm in PDF coordinates. Physical printer scaling still requires the two ruler checks.
47
48
- No firmware was installed on a physical node and no Home Assistant or irrigation settings were changed. Consult the PR checks for the final five-board CI build result.
49
+
50
+
## Calculator and calibration wizard
51
+
52
+
The wizard checks substrate geometry, unit conversion, manual wet-reference records,
53
+
and weighed A/B/C records before generating a configuration. It uses the same
54
+
soilless response fit as the firmware, including 0.1% sensor-input rounding and
55
+
independent C bounds. Changing sample geometry, placement or tare invalidates
56
+
recorded references; changing display units preserves their canonical values.
57
+
58
+
`test_generated_configs.py` exercises 15 real exports: five boards, each with no
59
+
calibration, a wet reference, or a checked A/B/C import. It validates them against
60
+
ESPHome and their pinned remote packages using temporary dummy secrets. `--compile`
61
+
also builds the Atom Lite weighed-reference export, including the generated C++
62
+
import action. The Pages deployment waits for this check and the site build.
63
+
64
+
Browser verification covers unit round trips, US and UK container presets, sample
65
+
volume updates, undersized cubes, wet and weighed workflows, rejected C points,
66
+
YAML/report/CSV downloads, reload and project import, reference invalidation,
67
+
responsive layouts and a one-page Letter template (612 × 792 PDF points).
68
+
The browser checks entered data. It cannot verify the physical sensor's capture
0 commit comments