Water Tank Level & Valve — Farmers IoT Toolkit

Why you would want this

Know the tank is getting low before you run out
Open and shut the irrigation line from anywhere, or on a schedule
Keep a record of water use across days and weeks
Never irrigate from an empty tank — the level and the valve are on the same node

How it goes together

Wiring diagram: 12 V comes in on a terminal block and splits two ways — into an MT3608 boost set to 18 V for the QDY30A hydrostatic water level probe, and into a Mini360 buck set to 5 V for the ESP8266 NodeMCU and an RS485-to-TTL module. The probe's four wires go to the boost and the RS485 board; the RS485 board goes to the ESP8266. A 5 V relay driven from the ESP switches a normally-closed 12 V solenoid valve on the tank outlet. — One 12 V feed, two rails: 18 V for the probe and 5 V for the logic. The probe's red wire is the only thing that ever sees 18 V. The valve hangs off the relay's isolated switched contacts, so its current never touches the ESP8266.

What you will need

Bill of materials for Water Tank Level & Valve
Picture	Part	What it does	NZ$	Link
	ESP8266 NodeMCU v3 (ESP-12E)	The small computer that runs the node	$5.77	Buy ESP8266 NodeMCU v3 (ESP-12E) ↗
	QDY30A submersible pressure probe Buy the RS485 variant. The same listing sells 4–20 mA, 0–5 V and 0–10 V outputs, and only RS485 works with this guide. Check the options before you pay.	Sits on the tank floor and measures the depth of water above it	$93.12Variant priced: 5 m range, 8 m cable, RS485 output. A different range changes the price.	Buy QDY30A submersible pressure probe ↗
	MAX485 / HW-0519 RS485-to-TTL module The listing ships two different boards. The HW-0519 is auto-direction and needs no DE pin; the classic breakout has DE/RE and costs you a GPIO. Both work — but only the HW-0519 leaves you a spare pin for the valve relay.	Lets the ESP8266 speak RS485 to the probe	$0.86The blue TTL-to-RS485 board, single piece.	Buy MAX485 / HW-0519 RS485-to-TTL module ↗
	MT3608 boost converter Set it to 18 V with NOTHING on the output, before the probe is ever connected.	Steps 12 V up to the 18 V the water probe needs	$2.49	Buy MT3608 boost converter ↗
	Mini360 buck converter (MP2307) Fine for a sensor node at a couple of hundred milliamps. Do NOT use one to charge a phone — it misbehaves above ~1.5 A in a hot box. That job wants the 3 A synchronous buck in Module 1.	Steps the pack's 12 V down to 5 V for the ESP8266 and RS485 board	$3.65	Buy Mini360 buck converter (MP2307) ↗
	1-channel 5 V relay module	Lets the 3.3 V ESP8266 switch the 12 V valve	$4.56	Buy 1-channel 5 V relay module ↗
	12 V solenoid valve, normally closed Normally-closed is the safety property this module rests on: no power = shut. Also check the minimum pressure differential — a pilot/servo valve needs ~0.2 bar and will not open on a low gravity tank no matter how good your wiring is.	Opens to let water out of the tank and into the irrigation line	$9.01The 12 V variant. The same listing's 220 V mains variant is cheaper and is not what this module uses.	Buy 12 V solenoid valve, normally closed ↗
	1N5819 flyback diode Not optional. Without it the ESP8266 reboots — but only when the valve closes.	Absorbs the spike the valve coil throws when it shuts	$4.62A lot of 50, 1N5819. One node needs one.	Buy 1N5819 flyback diode ↗
	Jumper wires	Connects everything	$5.50New Zealand retail — the sheet has no AliExpress row for these.	—
	USB cable	Powers the ESP8266 and loads code onto it	$10Bought as a lot of ten for $100.	—
	Hose fittings for the valve	Joins the valve into your irrigation line	~$8.50	—
	Waterproof container	Protects the electronics	~$8.50	—
Approximate total			$156.58

Prices are NZD, read off the linked listing on 2026-08-08. A figure marked ~ is an estimate — we have no listing for that part yet, so treat it as a placeholder rather than a quote. Prices move, and shipping to a small island is its own line item. Links are not affiliate links and we get nothing if you use them.

Wiring

Water bus — QDY30A

Baud: 9600
Frame: 8N1
Slave: 1

Probe wire	From	Goes to	Note
Red	probe red	MT3608 OUT+ (18 V)	18 V — this wire and nothing else. 18 V into the MAX485 or the ESP destroys both.
Green	probe green	MT3608 OUT− + common ground	Prime suspect for an intermittent read. One ground, not two.
Blue	probe blue	HW-0519 A	A+ — note this is the OPPOSITE colour convention to the soil probe.
Yellow	probe yellow	HW-0519 B	B−

Module 2 as this guide builds it

wire this one firmware/water-level/water-level.ino

This is the assignment in the wiring picture above, and it leaves D2 free for the valve relay. On a classic DE/RE breakout, join DE and RE and drive them from D1; an auto-direction HW-0519 needs no DE pin at all.

RS485 pad	ESP pin	GPIO	Direction
RXD / RO	D5	GPIO14	in — data toward the ESP
TXD / DI	D6	GPIO12	out — data toward the probe
relay IN	D2	GPIO4	out — opens and shuts the valve
DE + RE	D1	GPIO5	classic breakout only — omit on HW-0519
VCC	3V3	—	power
GND	GND	—	common ground star

Ian's deployed farm-node (water + valve)

firmware/farm-node/farm-node.ino

Swapped relative to the build above — the pins follow how the board was actually soldered on 2026-07-19, and the firmware follows the iron. NEVER run water-level.ino on this board: it drives D1 as DE, and D1 here is the water probe's TXD, so it would clamp the driver onto the bus and the probe could never reply.

RS485 pad	ESP pin	GPIO	Direction
RXD	D7	GPIO13	in — data toward the ESP
TXD	D1	GPIO5	out — data toward the probe
relay IN	D2	GPIO4	out — opens and shuts the valve
VCC	3V3	—	power
GND	GND	—	common ground star

Register	Means	Scaling	Type
0x0004	Depth above the probe	1 count = 1 mm	signed int16

Both RS485 board variants work and the firmware handles both — but the auto-direction HW-0519 is what leaves D2 free for the valve relay. A classic DE/RE breakout spends an extra GPIO on direction, and the budget is only five.

The valve side

The valve never touches the logic side of the board. It hangs off the relay's switched contacts, which are electrically isolated from the ESP8266 — that isolation is exactly why 12 V cannot backfeed your laptop while you bench it. The only way to put valve current onto USB is to wire the coil to the ESP's 5 V pin. Don't.

Relay pin	Goes to	Note
VCC	5 V	From the Mini360's 5 V output — or `VBUS` if the board is on USB only. The ESP 5 V pin you aren't feeding is dead: `VIN` floats on USB, `VBUS` reads ~3 V on the pack.
GND	GND	Common ground with the ESP.
IN	D2 (GPIO4)	The one safe pin left once the bus has taken D5 and D6.
COM / NO	valve coil	NO, so relay unpowered = valve unpowered = valve shut.

Build it, step by step

Set both converters before anything else is connected

This is the one step where you can actually destroy something. Both modules' trimmers ship at a random setting, and the boost feeds an 18 V rail.
Connect 12 V to IN+ / IN− with nothing on the output. Put a multimeter across OUT+ / OUT−. Turn the MT3608's screw until it reads a steady 18.0 V, and the Mini360's until it reads 5.0 V. Power down. Now connect the probe and the ESP8266.

Watch out
The trimmer feels like it does nothing. It is a 25-turn part with no end stops, so it spins freely while the wiper sits parked. Pick a direction and turn ~30 full turns slowly, watching the meter, not the screw. It eventually climbs.

This will bite you
Label the two converters the moment you set them. They look identical in a box, and putting 18 V where 5 V belongs destroys the ESP8266 and the RS485 board together.
Wire it up

Four wires out of the probe, four wires between the RS485 board and the ESP8266, and three to the relay. The tables below are generated from the same pin data as the wiring cheatsheet, so they cannot drift apart.

This will bite you
The 18 V goes to the probe's red wire and nowhere else. The MAX485 and the ESP8266 are 3.3 V parts — 18 V destroys both. Trace the wire with your finger before switching on.

Watch out
One ground, not two. The probe's green wire, the boost OUT−, the buck OUT−, the MAX485 GND and the ESP GND must all be one rail. The probe runs on 18 V and the ESP on 3.3 V — two supplies, one shared return. Miss this and you get failures that look exactly like a broken sensor.

Worth knowing
The TXD and RXD LEDs are a free logic analyser. TXD flashes when the ESP asks; RXD flashes when the probe answers. TXD only means the probe isn't replying — swap blue and yellow before suspecting anything else.
03
Load the code

The firmware is firmware/water-level/. Install the Arduino IDE, add esp8266 in Boards Manager, and select NodeMCU 1.0 (ESP-12E). There are no libraries to install.
Copy config.example.h to config.h and edit that — never the sketch itself. Start with BENCH_MODE 1 so it prints over serial and skips WiFi entirely: bring the sensor up before WiFi joins the list of things that can be broken.
```
# always compile --upload. A bare `upload` flashes a stale .bin
arduino-cli compile --upload \
  --fqbn esp8266:esp8266:nodemcuv2 -p /dev/ttyUSB0 firmware/water-level
```
Watch out
Note the baud rate. This probe is 9600. The soil probe in Module 3 is 4800. Same wiring, different speed — don't carry the number across.

Worth knowing
Reads are safe; blind writes are not. Reading a register cannot damage the probe. Writing one you haven't confirmed can land on its address or baud rate and cost you the ability to talk to it at all.
Zero it

Read the probe dry, sitting in air. It will not read zero — ours read about 26, which is 26 mm of water that does not exist. That is normal and within spec.
Write that number down. Your depth is reading − dry_reading. Do this once at install; if readings drift over months, do it again.
Check it actually moves

Lower the probe into a bucket. The number should climb clearly, hold steady when it stops moving, and fall back when you lift it out. If it does not respond to depth, stop — nothing after this will work.

Worth knowing
Don't try to calibrate better than ±25 mm. That is the manufacturer's spec (0.5% of 5 m) and it is far better than irrigation needs. We tried a bucket and a ruler and learned that a 90 mm bucket cannot measure an instrument whose own error bar is 25 mm.
06
Wire the relay and the valve

The valve goes on the relay's switched side, never the logic side. Use the NO (normally open) contact so that relay unpowered = valve unpowered = valve shut.
```
  12V +  ────────────────► valve terminal 1
  valve terminal 2 ──────► relay COM
  relay NO ──────────────► 12V −

  1N5819 across the valve's two terminals, stripe (cathode) to the + side
```
This will bite you
The diode is not optional. A solenoid is a coil of wire; cutting its power throws a large reverse spike back down the wires — enough to weld relay contacts over time and enough to reset the ESP8266 next to it. The symptom of leaving it out is maddening: the ESP randomly reboots, but only when the valve closes.

This will bite you
Power the relay from the 5 V rail that is actually live. In the field, take relay VCC from the Mini360's 5 V output — the same rail as the ESP8266's VIN. Only on a USB-only bench use VBUS. The NodeMCU's other 5 V pin is a ghost either way: on USB VIN floats at ~3.1 V, and on the pack VBUS sits at ~3 V. Both read plausible on a meter, and the relay simply won't pull in. This bit us once in each direction.

Worth knowing
The coil is non-polar; the diode and the supply are not. Pick either coil lead as “+”, run 12 V+ to it, and put the diode band on that same leg. Reversed, the diode forward-biases the instant the valve powers and shorts the pack.
Build the safety cutoff before you leave it alone

This is the step people skip and then regret. If a command is lost, or the link drops while the valve is open, the valve stays open — and a failed poll deliberately holds the last state, so an open valve plus a dead hotspot is an open valve forever. Nothing upstream can help, because nothing upstream can reach it.
So the node enforces its own dead-man timer, independent of any command: VALVE_MAX_OPEN_S shuts an open valve once it has gone that long without hearing the base station say “open”. It counts from the last confirmation, not from when the valve opened, so a planned run is never cut while the link is up — the run length is set on the base station's page. It ships at 1800 s and defaults to on, so a farmer opts out of the guard rather than into it.

This will bite you
This is the only module in the toolkit that acts rather than just reads. Everything else fails by telling you nothing. This one fails by emptying a tank into a field.

Worth knowing
The timer that matters is the one running on the node, because it is the only one still running when the node is alone. A cloud-side expiry cannot help if the link is what broke.
Install on your tank

Lower the probe to the bottom and let it sit flat. Fit the valve into the outlet line with the flow arrow pointing the way the water goes. Protect the ESP8266, RS485 board and both converters inside the waterproof container. Measure your tank's water depth when full, once — that is what turns millimetres into “how full is my tank”.

Watch out
Keep the loose end of the probe cable dry and open to the air. There is a tiny breather tube running through it, and the probe compares against outside air pressure through it. Seal it, kink it, or let it sit in water and your readings wander with the weather.

Watch out
Check the valve's minimum pressure differential against the head you actually have. A pilot/servo valve typically needs ~0.2 bar — about two metres of head — and on a low gravity tank it will click and pass nothing. If your head is marginal, spec a direct-acting “0 bar” valve.
09
Point it at the base station

Set BENCH_MODE 0, put your hotspot's name and password in config.h, and leave POST_HOST empty. The phone running the hotspot is the node's gateway by definition, so the node works out the address itself.
Test the endpoint before you ever blame the firmware:
```
curl -X POST http://<phone-ip>:1880/water \
  -H 'Content-Type: application/json' \
  -d '{"node":"test","ok":true,"depth_mm":500}'
```
This will bite you
Never hardcode the phone's IP. Android randomises the hotspot subnet — ours came up on 10.215.63.55, not the 192.168.43.1 every guide quotes — and it can reshuffle on any restart. A hardcoded IP goes stale silently: the node keeps reading perfectly and publishes into the void, and the only symptom is absence.

Worth knowing
The valve command comes back the same way. The node polls GET /valve once a second rather than running a server, because the phone can change the node's IP but the node always knows its gateway.

Supporting files

Everything in the repository worth having on your own disk before you start — the sketch you flash, the flow you import, the datasheet you check a pinout against. Download saves the file; the path opens it on GitHub if you would rather read it first.

sketch firmware/water-level/water-level.ino The sketch this guide builds — water level only, no libraries to install Download firmware/water-level/water-level.ino ↓
config firmware/water-level/config.example.h Copy to config.h and edit that. WiFi, the dry-reading calibration, and BENCH_MODE Download firmware/water-level/config.example.h ↓
sketch firmware/farm-node/farm-node.ino Our deployed node: water + master valve on one board, with OTA reflash. NOTE its pins are swapped relative to the guide — read firmware/README.md before flashing Download firmware/farm-node/farm-node.ino ↓
config firmware/farm-node/config.example.h Config template for the deployed node, including VALVE_MAX_OPEN_S Download firmware/farm-node/config.example.h ↓
flow flows/water-level-flow.json Node-RED flow that receives the depth readings. Import into the Module 4 phone Download flows/water-level-flow.json ↓
tool tools/poll-water.ts Bench poller — proves the probe from a laptop before any ESP8266 exists. bun tools/poll-water.ts Download tools/poll-water.ts ↓
doc project/bench/02-bench-water-level-rs485.md Bring-up notes from the bench: what we measured and what went wrong Download project/bench/02-bench-water-level-rs485.md ↓
doc firmware/README.md Build and OTA procedure, and why each sketch carries its own Modbus code Download firmware/README.md ↓
image docs/wiring-diagram-module-2-water.jpg The wiring picture at full size Download docs/wiring-diagram-module-2-water.jpg ↓
datasheet hardware/Manual - Water Level Option A.pdf QDY30A probe manual — register map and wire colours Download hardware/Manual - Water Level Option A.pdf ↓
datasheet hardware/Manual - MT3608 Boost.pdf MT3608 boost converter (the 18 V rail) Download hardware/Manual - MT3608 Boost.pdf ↓
datasheet hardware/Manual - Mini360 Buck.pdf Mini360 buck converter (the 5 V logic rail) Download hardware/Manual - Mini360 Buck.pdf ↓
datasheet hardware/Manual - TTL to RS485.pdf RS485 transceiver module — both board variants Download hardware/Manual - TTL to RS485.pdf ↓
datasheet hardware/Manual - Plastic Solenoid.pdf Solenoid valve manual — check the minimum pressure differential Download hardware/Manual - Plastic Solenoid.pdf ↓

Or take the lot: download the whole repository as a zip, or git clone https://github.com/sunrise-labs/farmers-iot-toolkit.git. Each sketch needs its config.example.h copied to config.h and edited — the real config.h is never committed, because it holds your WiFi password.

When it doesn't work

Ordered roughly by how often it has actually been the answer.

Symptom	Almost always	What to do
No reading at all	A and B swapped	Swap the blue and yellow wires. This is common — try it before anything else.
No reading, probe cold	Not enough voltage	Meter red to green: it must read 18 V, not the raw 12 V.
Boost converter reads 0 V	Trimmer parked mid-range	It is 25-turn with no end stops. Keep turning, watch the meter.
Readings 10× off	Wrong scaling	1 count = 1 mm. Check you are not treating it as centimetres.
Reading is about 65000	Read as unsigned	It is signed int16. Readings slightly below zero are real and legitimate.
Reading never changes	Wrong register	It is 0x0004.
Readings fail randomly	No common ground	Probe green, boost OUT−, buck OUT−, MAX485 GND and ESP GND must be one rail.
Drifts with the weather	Breather tube blocked or wet	Keep the loose cable end dry and open to air.
Reads, then silent, then reads	A wire, not the firmware	Wiggle-test each one while the node polls, prime suspect green/ground. Do this before soldering anything into a box.
Relay clicks but the valve doesn't move	Not enough volts at the coil under load	Meter across the valve while the relay is on. A sagging pack meters fine at rest.
Relay doesn't click at all	Relay VCC on a dead 5 V pin	On the pack, take VCC from the Mini360's 5 V output (same rail as VIN) — VBUS only reads ~3 V there. On USB only, use VBUS — VIN floats.
ESP reboots when the valve closes	Missing flyback diode	Fit the 1N5819 across the valve terminals, stripe to +.
Valve energised, nothing flows	Below the valve's minimum pressure differential	A pilot valve needs ~0.2 bar. Check the head above it; if marginal, spec a direct-acting “0 bar” valve.
Valve opens on a full battery, not a flat one	Solenoid under-volted	Measure the coil current and check the pack voltage under that load, not at rest.

How this fits with the others

Why a pressure probe and not an ultrasonic one

Nearly every beginner tutorial puts an ultrasonic sensor on the tank lid and bounces sound off the water. It is cheaper, and for a real farm tank it is the wrong tool — this module used to describe that design and we changed it deliberately. A tank is an enclosed cylinder full of reflective surfaces with a moving surface on top: the worst possible case for an echo. A pressure probe ignores every one of those problems because it does not look at the water. It sits underneath it.

	Ultrasonic, from the top	Pressure probe — this module
How it measures	Times a sound echo off the surface	Feels the weight of water above it
Tank walls	Sound bounces off them — false readings	Doesn't care
Ripples, foam, debris	Scatter the echo	Doesn't care
Condensation on the face	Blinds it	Lives underwater anyway
Mounting	Needs clear line of sight straight down	Just drop it in
Cost	Cheaper	More
Power	Runs off the board	Needs 12–36 V — hence the boost converter

Why you would want this

How it goes together

What you will need

Wiring

Water bus — QDY30A

Module 2 as this guide builds it

Ian's deployed farm-node (water + valve)

The valve side

Build it, step by step

Set both converters before anything else is connected

Wire it up

Load the code

Zero it

Check it actually moves

Wire the relay and the valve

Build the safety cutoff before you leave it alone

Install on your tank

Point it at the base station

Supporting files

When it doesn't work

How this fits with the others

Why a pressure probe and not an ultrasonic one

The source