Arduino: read water quality with the TDS module

TDS (total dissolved solids) is the “is my water filter actually doing anything” number. The Gravity analog TDS kit is a probe plus a small interface board, it plugs into an ADC pin like any other sensor, and this build prints a compensated ppm reading to the serial monitor once a second.

The trap is expecting chemistry from a conductivity meter. TDS does not identify what is dissolved; it estimates the total (e.g. it counts the good minerals and the bad salts as the same “stuff”). It also drifts about 2 percent per degree C, so an uncompensated reading is a mood, not a measurement. This tutorial wires a DS18B20 alongside and compensates properly.

What you need

Needed

• Arduino Uno or Nano
• Gravity analog TDS sensor kit: probe + interface board (about $20)
• DS18B20 waterproof temperature probe plus a 4.7k ohm resistor (about $3; the compensation math is worth more than the sensor)
• Breadboard and jumper wires

Nice to have

• Soldering iron, solder, stand, and helping hands (header work on the interface board if it ships bare)
• Wire stripper, multimeter
• Anti-static wristband, magnifying goggles, soldering mat
• Distilled water and one known reference solution (e.g. a 342 ppm conductivity standard) for a one-point calibration check
• Small cups, and a stand or clamp so the probe hangs in the sample without touching the cup walls

Wiring

Connection	Arduino
TDS board VCC	5V
TDS board GND	GND
TDS board A (signal)	A1
DS18B20 red	5V
DS18B20 black	GND
DS18B20 yellow	D2, with the 4.7k resistor to 5V

The signal output stays under about 3.4V even on a 5V supply, so the Uno ADC reads it directly. Keep the probe cable away from motors and relay coils; pump noise shows up as phantom ppm.

Install

One library pair, from the Library Manager:

• Arduino IDE >> Sketch >> Include Library >> Manage Libraries >> search “OneWire” >> Install
• Same path, search “DallasTemperature” >> Install

The code

// TDS with DS18B20 temperature compensation. Prints once per second.
#include <OneWire.h>
#include <DallasTemperature.h>

const int PIN_TDS = A1;
const int PIN_ONEWIRE = 2;
const float VREF = 5.0;
const float ADC_MAX = 1024.0;
const float CAL = 1.0;    // nudge this against a known solution

OneWire oneWire(PIN_ONEWIRE);
DallasTemperature temp(&oneWire);

float readVoltage() {
  int s[10];                                   // sample 10, trim, average 6
  for (int i = 0; i < 10; i++) { s[i] = analogRead(PIN_TDS); delay(5); }
  for (int i = 0; i < 9; i++)
    for (int j = i + 1; j < 10; j++)
      if (s[j] < s[i]) { int t = s[i]; s[i] = s[j]; s[j] = t; }
  long sum = 0;
  for (int i = 2; i < 8; i++) sum += s[i];
  return (sum / 6.0) * VREF / ADC_MAX;
}

void setup() {
  Serial.begin(9600);
  temp.begin();
}

void loop() {
  temp.requestTemperatures();
  float tC = temp.getTempCByIndex(0);
  if (tC == -127.0) tC = 25.0;    // sensor missing: fall back to nominal

  float v = readVoltage();
  float comp = 1.0 + 0.02 * (tC - 25.0);          // 2% per degree C
  float v25 = v / comp;
  float tds = (133.42*v25*v25*v25 - 255.86*v25*v25 + 857.16*v25) * 0.5 * CAL;

  Serial.print(tC, 1); Serial.print(" C   ");
  Serial.print(v, 3);  Serial.print(" V   ");
  Serial.print(tds, 0); Serial.println(" ppm");
  delay(800);
}

Sanity ranges, so you know when the number is lying: distilled water lands near 0-20 ppm, most tap water between 100 and 400, hydroponic nutrient 800-1500, and seawater is far above the module’s range (0-1000 ppm).

What you learned

• TDS is conductivity wearing a ppm costume: the board measures conductance, the standard curve converts it, and temperature bends the curve.
• Trimmed-mean sampling (drop the extremes of 10 reads) kills most of the ADC noise without a filter library.
• Compensation belongs in the voltage, before the curve, not in the final number.

When something breaks

• Reads 0 ppm in actual water: probe connector loose, or the signal wire is on the wrong A-pin. Wiggle the BNC-style connector; the reading should appear instantly.
• Negative or exploding ppm values: the DS18B20 is not answering (check the 4.7k pull-up; -127 means “not found”), or the probe is in air.
• Number creeps up over minutes: the probe needs 2-3 minutes to stabilize in a new sample, and biofilm or mineral scale on the electrodes does the same thing. Rinse in distilled water and dry between samples.
• Jumpy by tens of ppm: USB power noise, or the probe cable runs next to a motor or pump wire. Separate them; add the trimmed-mean if you have not.
• Reads high against the 342 ppm standard: nudge CAL (e.g. measured 380 against a true 342 means CAL = 342/380, about 0.9).
• Probe left in water for weeks: biofilm grows on the electrodes and every reading drifts. Rinse weekly; this is a sampling instrument, not a permanently installed lab probe.

What to build next

• Put the reading on a screen at the tank with the LCD I2C tutorial.
• Combine this with the MQ-2 alarm’s buzzer pattern: out-of-range ppm, same beep logic, new meaning.
• The ESP32 plant monitor covers the irrigation side, and esp32 MQTT publish/subscribe logs every reading to your own broker (self-hosted, no cloud subscription and no third-party dashboard in the path).