Week 10: Output devices & power measurement

Individual assignment: XIAO I²C → WROOM ST7789

Per Output Devices I must add an output device to a microcontroller board I designed and program it. My designed board is still the Week 8 hub with the XIAO and sensors; the programmed output is the ST7789 on the WROOM carrier (Senlin ESP32‑S3 WROOM + touch TFT assembly). Week 10 is a two-MCU firmware link extracted from my download tree, not a PCB spin.

1. Task

Week 9 proved the hub can measure light and RH / temperature and log them on USB serial. Week 10 adds a machine-readable bus and a human-readable faceplate: XIAO packs RH, °C, and light into TelemetryPacked (cmd 0x01, slave 0x55); WROOM parses the frame in an I²C slave ISR ring buffer and redraws the Environment panel (Temperature / Humidity / Light + status pill).

• Hub (XIAO, Week 8 PCB): same sensor pins as Week 9 hub firmware: DHT11 on D2 (GPIO3), light ADC on D3 (GPIO4); D4/D5 = I²C SDA/SCL to the display board.
• Display MCU (WROOM): ST7789 over SPI; I²C slave Wire1 on GPIO37/36 (HostLink in board_pins.h).
• On-screen payload: the same three quantities as Week 9 serial: humidity %, temperature °C, light % of ADC full scale.
• Out of scope for this week’s repo slice: WiFi/Deepseek, ASR, five-page swipe UI, soil Modbus. Those stay in the full WROOM上传程序 / S3上传程序 trees.

Board context (Week 8) vs scope (Week 10)

The JLCPCB panel from Week 8 is laid out for two sites: a soldered XIAO ESP32‑S3 and a NanoStat landing zone for later plant impedance experiments. That geometry still fits both modules mechanically, but I am not changing the board or NanoStat bring-up in Week 10, same decision as Week 9. Evidence this week is: designed MCU board + new output device under firmware control.

The academy FAQ says the assessed output should not live on a breadboard forever. My sensors and MCU are already on the carrier; the TFT module is cabled from the XIAO header for now (like the small OLED I used at the end of Week 9). When I route a dedicated FPC or pin header for the plant enclosure I will update the bench photo, without respinning the Week 8 panel.

2. Learning

Week 9 was slow sensing; Week 10 is framed I²C plus pixel output. I copied the protocol header and the env-only paths from my upload folders rather than inventing a second format.

• Master framing: one Wire.endTransmission() writes 8 bytes: magic 0xA5, cmd 0x01, flags, RH %, temp×10, light×10.
• Slave parsing: bytes land in a ring buffer from onReceive; the main loop assembles TelemetryPacked and bumps a dataVersion so the TFT redraws only when values change.
• ST7789 / SPI: the WROOM firmware rotates MADCTL to match the Senlin module, then draws the same Environment layout as the full UI tree’s drawPageEnv().
• Power (group): TFT backlight current dominates. See group measurements.

3. Plan

1. Keep Week 9 sensor pins; add XIAO D4/D5 → WROOM GPIO37/36 with 4.7 kΩ pull-ups and common GND.
2. Flash WROOM week-10 slave build; confirm serial [HOST_I2C] slave 0x55.
3. Flash XIAO master; boot I²C scan must list 0x55.
4. Verify serial banner on XIAO and live Environment fields on the TFT.
5. Bench photo: Week 8 hub + linked display showing RH / °C / light.

4. Build (carried from Week 9 → ST7789)

Same hub, new output

Input chain (unchanged from Week 9)

Light and DHT11 stay soldered on the carrier with rainbow bring-outs. I will not move those nets onto a breadboard shim for Week 10.

Pin map: HostLink (XIAO master ↔ WROOM slave)

UI layout (WROOM Environment page)

The trimmed WROOM build draws the same three fields as the full UI’s drawPageEnv(): Temperature, Humidity, Light, plus a status pill ( WAIT S3 0x55 / AIR OK / STALE / DHT ERR). In the five-page firmware, swipe to page index 2; the week-10 teaching build stays on that page only.

Firmware in the repo (extracted from upload trees)

Documented in code/week10-i2c-env/README.md:

• XIAO master: week10-i2c-env/xiao/ : pushTelemetry() + serial banner (from S3上传程序/src/main.cpp).
• WROOM slave + TFT: week10-i2c-env/wroom/ : env_i2c_slave.cpp + Environment draw (from WROOM上传程序).
• Shared protocol: i2c_env_link.h (TelemetryPacked, 8 bytes).

// XIAO — one telemetry write (see pushTelemetry in week10-i2c-env/xiao)
I2cEnvLink::TelemetryPacked pkt{};
pkt.magic = 0xA5;
pkt.cmd   = 0x01;  // telemetry v1
pkt.flags = dhtOk ? 0x01 : 0;
// rh_percent, temp_c_x10, light_pct_x10 …
Wire.beginTransmission(0x55);
Wire.write((uint8_t*)&pkt, sizeof(pkt));
Wire.endTransmission();

Full bench trees (WiFi, LLM, touch swipe, ASR) remain in Downloads; the integrated copy for later weeks is also under code/week15-individual/final/.

Problems I watch for

• No 0x55 on scan: WROOM not flashed, missing GND, or weak pull-ups. Fix before blaming sensors.
• WAIT S3 0x55 on TFT: slave OK but master not running or SDA/SCL swapped.
• DHT ERR with good serial on XIAO: flags bit not set. Check kFlagDhtOk packing on the master.

5. Conclusion

Week 10 links sensing to showing: the Week 8 hub still reads the plant-facing transducers, and the programmed output is the WROOM’s ST7789, fed by the same 8-byte I²C frame I use in the full upload trees. Splitting MCU roles matches how I want the final companion to work: hub near sensors, display module at the human edge.

Design files: Week 6 electronics design · Week 8 production. Week 9 serial: week09-s3-serial-env.ino. Week 10 I²C + display: code/week10-i2c-env/. Group reflection: power measurement below.