Sprig Dashboard

A multi-app information dashboard for the Hack Club Sprig (Raspberry Pi Pico WH on the Sprig carrier board), written in MicroPython with uasyncio for a non-blocking UI and networking.

Four apps, switched with the D-pad cluster:

Button	App	What it does
W	Weather	Current temp (large), conditions, 3-day min/max bar chart (Open-Meteo)
S	Finance	One scrolling ticker list (incl. S&P 500, Dow, QQQ, futures); drill-down with a price chart + selectable timeframe (1D/5D/1M/6M/1Y); single-press ribbon search. All list prices fetched in one Yahoo spark request; fetches retry across query1/query2 hosts
D	Sports	Two modes (double-tap I/L toggles LIVE ↔ STANDINGS). Tabs (J/L): ALL · F1 · NFL · NBA · MLB. ALL is the universal default in both modes — all in-progress games across every league. League tabs show that league's scoreboard (LIVE) or table (STANDINGS): F1 = drivers + constructors (two scrollable sections), others = W-L tables. ESPN public API; all fetches retried
A	Ghost Sniffer	Three J/L views: RF signal histogram + waterfall + open-network count; 2.4 GHz channel-congestion analyzer; and a Sprig system monitor (die temp, CPU clock, RAM/flash, uptime, IP/MAC, RSSI)

All data sources are $0-cost, key-free public endpoints.

Controls

The right-hand action cluster (I J K L) navigates inside the active app:

• I = up · K = down · J = left / prev tab · L = right / next tab
• double-tap I or L → select / drill-down
• double-tap J or K → back / cancel

The left D-pad (W A S D) switches apps at any time.

Per-app extras:

• Finance list: I/K scroll all tickers in one list; double-I/L opens detail; the last row is search.
• Finance search: instant single-press (no double-click) — J/L move the ribbon, I picks the highlighted key, K backspaces; scroll to OK to search or CANCEL to exit.
• Finance detail: J/L change the chart timeframe, I/K flip to the prev/next ticker without leaving the chart.
• Sports: a visible two-row header — top row is the [LIVE] [STANDINGS] section selector (double-tap I/L to switch), bottom row is the league tabs ALL/F1/NFL/NBA/MLB (J/L). ALL = all live games everywhere (the universal default in both sections). I/K scroll; standings show ▲/▼ when there's more off-screen.
• Ghost Sniffer: J/L cycle RF scan → channel analyzer → system monitor; I/K scroll APs in RF view.

Hardware pin map

Taken from the stock Sprig MicroPython firmware — do not change unless your board differs.

Function	GPIO
Display SPI0 SCK / MOSI / MISO	18 / 19 / 16
Display CS / DC / RST	20 / 22 / 26
Buttons W A S D	5 6 7 8
Buttons I J K L	12 13 14 15
Audio (PWM piezo, configurable)	config.AUDIO_PIN (default 27)

Audio note: the stock Sprig uses an I²S MAX98357A amplifier, not a PWM piezo. This project drives a simple PWM buzzer on a configurable GPIO (a rear-exposed pin by default). Set AUDIO_PIN = None in config.py to mute, or change it to match your carrier.

Setup

1. Flash the latest stable MicroPython for the Pico W / Pico 2 W onto the board.

2. Copy config.example.py → config.py and set your WIFI_SSID / WIFI_PASS (and optionally your LAT / LON, tickers, focus teams).

3. Upload every .py file to the device root (e.g. with mpremote):

   mpremote connect auto cp *.py :
   mpremote connect auto reset

   main.py runs automatically on boot.

Files

File	Role
main.py	App router + global input handler (double-click logic), async task loop
gfx_engine.py	ST7735 driver + framebuffer primitives (draw_text, draw_rect, sprites)
wifi_manager.py	Async Wi-Fi connect + non-blocking HTTP(S)/JSON client
app_base.py	Base App class and navigation contract
store.py	JSON flash cache (instant boot, survives power cycles)
weather_app.py	Open-Meteo integration
finance_app.py	Yahoo Finance tickers + drill-down + search
sports_app.py	ESPN scoreboard aggregator
network_app.py	"Ghost Sniffer" RSSI visualizer
config.py	Local secrets/config (git-ignored)

Performance & caching

The dashboard is built to be usable instantly and seamless after a short warm-up:

• Boot preload. On power-up every app starts fetching immediately (the active app first, then the rest in the background) — you don't have to open an app to make it load. Initial warm-up is roughly ~10 s on a good connection.
• Adaptive background scheduler. The active app always has priority; other apps load one at a time. "Heavy" data (sports scoreboards/standings) is only warmed while you're on a light screen (weather/finance/scanner), so spare capacity is used for heavy pulls and the device isn't piled with work while you're already on a data-heavy screen. Off-screen apps refresh on a relaxed cadence; a background prefetcher warms all standings (F1 drivers + constructors, NFL/NBA/MLB) one league at a time and then idles once warm, so it doesn't waste power.
• Flash cache. Each app writes its last-known data to /cdata/*.json (store.py) and reloads it on boot, so subsequent power-ups paint real values instantly and just refresh in the background. Writes are atomic and throttled (≥2 min/file) to spare flash.
• Placeholders. Apps ship with built-in placeholder values so the very first boot (before any cache exists) is never a blank screen.
• Fewer/cheaper requests. Finance pulls the whole list in one spark request; sports streams + parses on the fly (never buffering the 100-350 KB payloads) and only the visible/needed league is fetched. The RF scanner is the only app that doesn't run in the background (its wlan.scan() blocks).
• One connection at a time. All network I/O is serialized behind a single lock — a Pico W can only afford one TLS connection's buffers at once, so concurrent fetches were the main cause of [Errno 12] (out-of-memory). Sports also keeps only the currently viewed league's data in RAM (everything else on flash), freeing heap for the TLS buffer so refreshes stop failing.
• Always-on rotation. Whatever screen you're on refreshes continuously; the background loop rotates through stock prices and live scores so they stay current. The sports LIVE view refreshes one league per cycle (instead of stalling on all four at once) — fast, and always rolling forward.
• Standings are pre-downloaded. They barely change between games, so they're hard-stored on flash: fetched once, served instantly thereafter, and only re-downloaded every ~6 h in the background. Opening a table is immediate.
• Charts use the spark endpoint (~6× smaller than the chart endpoint), which fixes the out-of-memory error when opening a stock chart.
• No overflow. All text is truncated to the screen width (no overlap / no half-glyphs cut at the edge); long lists scroll with I/K.

Design notes & constraints

• Memory: a full 160×128 RGB565 framebuffer is 40 KB. The HTTP client caps responses at HTTP_MAX_BYTES (90 KB) and gc.collect()s aggressively around every fetch. Sports only loads the active league tab at a time, since ESPN scoreboard JSON is the largest payload. Very busy game days can still approach the limit — if a tab fails to parse, switch away and back to retry.
• Networking: all requests use Connection: close and handle chunked transfer-encoding. TLS uses uasyncio.open_connection(..., ssl=True), which requires a recent MicroPython build.
• Clock: the header clock is UTC via ntptime (no timezone math).
• Colour order: gfx_engine.color565() pre-swaps bytes so framebuffer colours render correctly on the ST7735. Flip the swap there if your panel shows red/blue inverted.

License

MIT — do whatever you like.