Shack Weather Beacon

This project began as a small ESP32 build acting as an APRS-IS node (VU35KB-13 on APRS.fi). The same board and shack install later grew into a weather beacon: live temperature, humidity, and pressure from inside the radio shack, with derived metrics calculated in the browser.

Live dashboard: Weather station

Scope: Indoor telemetry at Bangalore (grid MK82uu, about 920 m altitude). Not outdoor weather or a certified forecast. Three sensor fields feed all charts and derived values. No AI.

Hardware

Waveshare ESP32-S3 Geek with M5Stack ENV III. The onboard screen shows temperature, humidity, pressure, and APRS beacon status.

The ESP32 still participates in the APRS ecosystem as VU35KB-13. The web dashboard is a separate view built from the same sensor pipeline.

Data pipeline

The weather path is self-hosted on a private VPS. No third-party weather API or cloud IoT platform. The static site (ham.srinikb.in) and n8n workflow run on the same VPS. Formulas run in your browser. APRS-IS and NTP are the only external services in the sensor loop.

1. ESP32 samples ENV III every 15 minutes.
2. Readings POST to n8n on the VPS. The workflow writes each row to an n8n Data Table (full history kept for your own trend analysis). A JSON API returns the recent window the dashboard needs.
3. The dashboard fetches that API in the browser and caches results in memory and localStorage. Charts and cards show the last 24 hours; older rows remain in the Data Table on the VPS.
4. Auto-refresh when cache is older than 10 minutes, or when the latest reading is older than 20 minutes. Manual refresh has a 60 second cooldown.

End-to-end flow

                    ┌──────────────────┐
                    │ Boot: config LCD │
                    │ WiFi, NTP, web UI│
                    └────────┬─────────┘
                             │
         (background: config web UI, WiFi retry, LCD refresh 1 s)
                             │
              ┌──────────────┴──────────────┐
              │      Every 15 minutes       │
              └──────────────┬──────────────┘
                             │
                    ┌────────▼─────────┐
                    │ ESP32-S3 Geek    │
                    │ reads ENV III    │
                    │ SHT30 + QMP6988  │
                    └────────┬─────────┘
                             │
                    ┌────────▼─────────┐
                    │ Update LCD       │
                    └────────┬─────────┘
                             │
              ┌──────────────┴──────────────┐
              │      Beacon enabled?        │
              └─────┬──────────────────┬────┘
                 no │                  │ yes
                    │    ┌─────────────▼──────────────┐
                    │    │ APRS-IS (external, opt.)   │
                    │    │ login + WX packet TX       │
                    │    │ rotate.aprs2.net:14580     │
                    │    └─────────────┬──────────────┘
                    │                  │
                    └────────┬─────────┘
                             │
                    ┌────────▼─────────┐
                    │ n8n webhook      │
                    │ HTTPS POST JSON  │
                    └────────┬─────────┘
                             │
              ───── Self-hosted VPS ─────
                             │
                    ┌────────▼─────────┐
                    │ Data Table       │
                    │ full history     │
                    └────────┬─────────┘
                             │
                    ┌────────▼─────────┐
                    │ JSON API         │
                    │ last ~24h rows   │
                    └────────┬─────────┘
                             │
              ┌───────────────┴─────────┐
              │                         │
   ┌──────────▼──────────┐   ┌──────────▼──────────┐
   │ ham.srinikb.in      │   │ Browser fetch API   │
   │ Astro /weather page │   │ dashboard last 24h  │
   │ static site + JS    │   │ cache + formulas    │
   └─────────────────────┘   └─────────────────────┘

Self-hosted on a private VPS. No third-party weather API or cloud IoT. The dashboard shows 24 hours; the n8n Data Table retains full history for later analysis. The static site serves the /weather page; the browser fetches the JSON API directly.

Device detail (protocols and background loop):

If beacon is turned off in settings, APRS is skipped but sensor read and webhook POST still run every 15 minutes.

API record shape

Raw metrics (sensor)

Derived metrics (calculated)

All formulas run client-side. Display values use 2 decimal places.

Dew point

Magnus formula over liquid water:

• α = (17.27 × T) / (237.7 + T) + ln(RH/100)
• T_dew = (237.7 × α) / (17.27 − α)

T is °C, RH is percent relative humidity.

Use: Moisture in the air. Compare to temperature for condensation risk.

Dew margin

margin = temperature − dewPoint

Use: Gap before saturation. Below about 2 °C margin, condensation on cold metal is more likely.

Saturation vapor pressure

SVP = 0.6108 × exp((17.27 × T) / (T + 237.3)) kPa

Vapor pressure deficit (VPD)

VPD = SVP × (1 − RH/100) kPa

Use: How dry the air is. Higher VPD means stronger evaporative drying.

Feels like

Hybrid apparent temperature:

• If T at or above 26.7 °C (80 °F): Rothfusz heat index (US NWS regression, converted to °C), with low and high humidity adjustments.

• Otherwise: Steadman apparent temperature with wind assumed 0 m/s (no wind sensor):

  AT = T + 0.33 × e − 4 where e is vapor pressure in hPa.

Use: Perceived heat and humidity combined.

Sea-level pressure

Barometric reduction to mean sea level using station altitude (920 m for Bangalore):

SLP = P × (1 − (0.0065 × h) / (T_K + 0.0065 × h))^(−5.257)

Use: Compare shack pressure to weather reports and other stations.

Comfort label

Rule-based label from temperature, humidity, and feels-like:

Trends and aggregates

Chart window

Sliding window ending at the latest reading: 24, 12, 6, or 3 hours. Preference stored in browser localStorage.

Min / avg / max

Simple min, mean, and max over the selected window for each raw metric.

Hourly rate

(value_end − value_start) / hours_elapsed across the window.

Pressure shift

Compares average of the last 3 readings vs the previous 3:

Rain outlook (barometric rules)

Based on 3-hour pressure change plus humidity. Not a numerical weather model.

3-hour pressure delta

At each timestamp, find the reading closest to 3 hours earlier (within 45 minutes tolerance). Delta = current pressure minus pressure 3h ago.

Tendency classes

METAR-style barometric tendency on the 3h delta (hPa):

Rain score

Rule table inspired by standard barometric tendency practice (WMO-style tendency / Lancaster-style short-range rules):

Adjustments:

• Delta below -0.5 hPa and RH at or above 80%: score +15 (cap 95); risk bumped to moderate if it was low.
• Pressure under 1000 hPa and delta below -1 hPa: score +5 (cap 95).

Use: Hint for outdoor rain risk from indoor barometer trend. Confirm with local forecasts before field work.

Solar and time context

Station coordinates: Bangalore (stations.json, grid MK82uu).

Sunrise / sunset: NOAA-style solar position (equation of time, declination, 90.833° zenith for refraction). Used for daylight strip and night bands on charts.

Night shading: Timestamps between sunset and sunrise at the station. Fallback: IST hour bands (night 19:00 to 05:00, dawn 05:00 to 07:00, day 07:00 to 17:00, dusk 17:00 to 19:00).

Display time: 12-hour IST (Asia/Kolkata).

Daily rhythm chart

Groups readings by IST hour (0-23) and averages temperature per hour within the selected window.

Dashboard sections

Humidity chart shows a 40 to 60% RH comfort band.

Limitations

• Indoor air lags and differs from outside weather.
• Rain outlook is a rule-based barometer hint, not a forecast model.
• No wind, rain, or UV sensors; feels-like assumes zero wind.
• 15 minute sampling smooths short spikes.
• Several hours of history needed for 3h pressure delta charts.

Related links

• Live weather dashboard
• VU35KB-13 on APRS.fi
• Station page for solar countdowns and operating context