Skip to content

pcb

Directory actions

More options

Directory actions

More options

Latest commit

 

History

History

pcb

README.md

TankSync PCB Designs

Both boards were designed by Ravi Singh in EasyEDA — solo design work, no hardware team. The current production revision (REV 2.2, May 2026) has been bench-tested at 45 °C ambient through Delhi summer with no observable RYLR998 RSSI drift or boost-converter sag.

The TX is a custom circular PCB sized to drop straight into the circular-v1 enclosure. The RX is a rectangular carrier that sockets an ESP32 DevKit v1 (CP2102) module — no microcontroller redesign needed.

Files in this folder

File What it is When you need it
rx-schematic.png / .svg RX hub schematic Read before populating the board
tx-schematic.png / .svg TX transmitter schematic Read before populating the board
pin-assignments.md / .pdf GPIO → function table for both boards When firmware doesn't see a peripheral — check here first
system-block-diagram.pdf One-page TX↔RX↔cloud overview When explaining the system to someone new
rx-pcb-3d.step / tx-pcb-3d.step 3D CAD models of the populated boards If you're designing your own enclosure — import into Fusion / FreeCAD / OnShape
project-requirements.txt The original brief sent to PCBWay Reference only — supersedes by the schematics + pin-assignments above

TX board (transmitter) — at a glance

  • MCU: ESP32-C3 SuperMini (RISC-V, Wi-Fi + BLE, PCB antenna)
  • Radio: REYAX RYLR998 LoRa module (UART AT-command, 865 / 868 / 915 MHz selectable)
  • Sensor: JSN-SR04T / AJ-SR04M waterproof ultrasonic — connected via screw terminals so the sensor cable is field-installable
  • Power: 18650 Li-ion + CN3791 MPPT solar charger + MT3608 3.3 V boost
  • Battery telemetry: INA219 over I²C (current + voltage, signed) — replaced the older voltage-divider design in v1.0
  • Protection: inline Schottky on the 18650 (+) rail; 1 k + 2 k divider on the 5 V ECHO line so the ECHO signal is safe for the C3's 3.3 V GPIO

Power path

flowchart LR
    Panel[Solar panel<br/>~6V 1-2W] --> MPPT[CN3791 MPPT]
    MPPT --> Bat[18650 cell]
    Bat --> Boost[MT3608 boost]
    Boost --> C3[ESP32-C3 + RYLR998 + sensor]
Loading

All modules share a common ground. Vmpp tracking lands the panel near 4.5 V. The CN3791's BAT+ feeds the 18650 holder through an inline Schottky (SS14 / 1N5819) — destroyed two test boards before adding this.

Signal path

flowchart LR
    C3[ESP32-C3<br/>SuperMini] <-->|UART1<br/>GPIO20/21| LoRa[RYLR998]
    C3 -->|TRIG<br/>GPIO4| Sens[JSN-SR04T]
    Sens -->|ECHO via 1k/2k div<br/>GPIO5| C3
    C3 <-->|I2C<br/>GPIO1/2| INA[INA219]
    C3 -->|WS2812 data<br/>GPIO7| LED[Status LED]
Loading

The full pin map is in pin-assignments.md — that's the source of truth; if this diagram drifts, the table wins. Wire-by-wire detail also lives there.

RX board (hub) — at a glance

  • MCU: ESP32 DevKit v1 (Tensilica, Wi-Fi + BLE, USB-C onboard via CP2102)
  • Radio: RYLR998 (same module as TX, same frequency, addresses differ)
  • Display: 1.3" SH1106 OLED over I²C (status + Wi-Fi + IP + paired tanks)
  • LED strip: WS2812B — user-selectable 2 / 8 / 24 LEDs, configured via the local web UI
  • Buzzer (since rx-v2.8.0): active 3-pin piezo on GPIO 02 — audible nudges for low water, overflow, sensor errors
  • Power: USB-C 5 V, 1 A nominal (5 W adapter is fine)

Power path (RX)

flowchart LR
    USB[USB-C 5V] --> LDO[AMS1117-3.3]
    LDO --> ESP[ESP32 DevKit]
    LDO --> Lora[RYLR998 + OLED]
    USB -->|direct 5V<br/>+ 1000 µF cap| WS[WS2812B strip]
Loading

The WS2812B strip is on the raw 5 V rail, not the 3.3 V regulator output — a 24-LED ring at full white can draw >1 A, which would brown out the LDO. The 1000 µF inrush cap is critical.

Signal path (RX)

flowchart LR
    ESP[ESP32 DevKit] <-->|UART2<br/>GPIO16/17| Lora[RYLR998]
    ESP <-->|I2C<br/>GPIO21/22| OLED[SH1106]
    ESP -->|WS2812 data<br/>GPIO13| WS[LED strip]
    ESP -->|Buzzer<br/>GPIO02| Buz[Piezo]
Loading

GPIO 2 on the DevKit is taken by the on-board LED, which is why the WS2812 data line is on GPIO 13, not GPIO 2 — don't move it.

Schematic reading tips (from review)

  • The schematic PNG renders are correct but small; if you're hand-reviewing, open the SVG in a browser and zoom — the layer naming is preserved.
  • The ECHO divider (TX) is on the PCB; do not omit it even if your sensor is rated for 3.3 V. JSN-SR04T variants in the wild differ on ECHO logic level, and the divider is cheap insurance.
  • The I²C pull-ups on the TX board are populated as 4.7 kΩ. Some INA219 breakouts ship with their own pull-ups — if you use a breakout, remove or de-populate the on-PCB ones to avoid double pull-up.
  • All decoupling caps (100 nF) belong within 5 mm of the IC power pin. The schematic shows them grouped for readability, but the layout enforces the distance.

Critical things the firmware will refuse to boot without

  1. TX: RYLR998 must answer AT\r\n at 115200 baud within 1.5 s. If wiring is reversed (TX↔RX swap), nothing pairs.
  2. RX: SH1106 OLED must respond at I²C 0x3C. If absent or at 0x3D, the display stays blank but firmware otherwise runs — useful as a diagnostic.
  3. TX: INA219 should answer at I²C 0x40. If absent, firmware falls back to voltage-divider battery monitoring on GPIO 0 (Variant A).

If you're hitting a boot loop, scope the UART RX line on either board first — most "the radio is dead" reports are reversed TX/RX wiring.