Sensor Calibration Wizard

Last updated March 19, 2026

Overview

Over time, sensors can drift — a temperature probe might read 2°C too high, or a pH probe may shift after weeks in solution. The Sensor Calibration Wizard lets you correct these inaccuracies directly from the dashboard without touching your hardware or Node-RED flows.

Calibration in LoopString is a display-side transform. Your raw sensor readings stay untouched in the database. The correction is applied after the data arrives from your Pi, just before it appears on the dashboard. This means:

  • Your historical data is not retroactively changed.
  • PID controllers and alarm thresholds continue to operate on the original raw values.
  • You can reset to factory (uncalibrated) at any time with no data loss.

Calibrations are saved to your account and automatically synced back to your Pi so that Node-RED can also apply the correction when needed.

How to Open the Wizard

There are two ways to open the Calibration Wizard:

  • From a sensor card: Click the three-dot menu on any sensor card on the dashboard, then choose "Calibrate."
  • From Settings: Go to your device settings page, find the Calibration section, and click "Calibrate" next to any sensor in the list.

How to Use

The wizard walks you through a short sequence of steps. The number of steps depends on which calibration mode you choose.

Step 1 — Select a Sensor

If you opened the wizard from the settings page (not from a specific sensor card), you will first be asked to pick which sensor to calibrate. Choose from the list of sensors currently reporting data from your device. If a sensor is not listed, it may be offline or not yet reporting.

Step 2 — Choose a Calibration Mode

Pick the correction method that matches your situation:

  • Single-point offset — Shifts every reading by a fixed amount. Use this when your sensor reads consistently too high or too low across its entire range (for example, always 1.5°C above the true value). This is the simplest and fastest mode. Available on Hobby plans and above.

  • Two-point linear — Corrects both the zero point and the gain (scale factor) of the sensor using two known reference measurements. This handles sensors that are off by different amounts at the low end versus the high end. Available on Maker plans and above.

  • Multi-point polynomial — Fits a curve through three to eight reference points. Best for sensors whose error is non-linear across their operating range (for example, non-linear analog voltage sensors). Available on Pro plans and above, and only for sensors using the Generic Analog Sensor part.

  • Reset to factory — Removes any existing calibration and reverts the sensor to its raw uncorrected readings.

Step 3 — Capture Reference Points

For each calibration point, you will see the sensor's current live reading. Place your sensor in a known condition (for example, an ice bath at 0°C, or a calibration solution at a known pH), wait for the reading to stabilize, then enter the true reference value.

The wizard captures the current sensor reading automatically when you move to the next point. You do not need to type the raw value — only the known true reference value.

For two-point calibration: You will go through this step twice — once at a low reference point and once at a high reference point. The further apart the two reference conditions are, the more accurate the correction will be.

For multi-point polynomial calibration: You can add between three and eight reference points. Use the plus and minus buttons to add or remove points. More points generally produce a more accurate curve, up to the physical limits of your sensor.

Step 4 — Preview

Before saving, the wizard shows you a live preview of the corrected value alongside the uncorrected raw reading. You can see exactly how the calibration will change what appears on your dashboard. If the preview does not look right, use the Back button to adjust your reference values.

For multi-point calibrations, a curve chart shows how the correction is applied across the full operating range.

Step 5 — Apply

Click "Apply Calibration" to save. The correction takes effect immediately on the dashboard. Your Pi also receives the updated calibration parameters within a few seconds.

Configuration Options

Single-Point Offset

  • One reference point required.
  • Enter the true value at one known condition.
  • The wizard computes the fixed shift: calibrated = raw + offset.
  • Best for: sensors with stable gain but a fixed zero-point drift.

Two-Point Linear

  • Two reference points required.
  • Enter a low reference value and a high reference value at two separate known conditions.
  • The wizard computes both a slope (gain) and an offset: calibrated = (raw × slope) + offset.
  • Best for: sensors that need both a zero correction and a sensitivity correction.
  • Important: the two raw readings you capture must be different from each other. If you capture both points at the same sensor reading, the wizard will warn you and not proceed.

Multi-Point Polynomial (Generic Analog Sensor only)

  • Three to eight reference points required.
  • The wizard fits a polynomial curve (quadratic by default, cubic also available) through your reference points using a least-squares algorithm.
  • Use the degree selector in the wizard to choose between quadratic (smoother, less overfitting) and cubic (tighter fit to your points).
  • Best for: analog sensors with non-linear response curves, such as analog pH probes or custom voltage-divider sensors.

Raw ADC Mode (Generic Analog Sensor)

The Generic Analog Sensor part works by reading a raw voltage or milliamp signal and converting it to an engineering unit (such as ppm or psi) using a configured range. When you calibrate one of these sensors, the wizard automatically captures the pre-conversion raw signal (in V or mA) as the input for calibration, rather than the post-conversion engineering value.

This ensures the calibration math is applied at the right point in the signal chain and avoids scale mismatches. You do not need to configure this — the wizard detects it automatically.

If you see a "Needs redeploy" warning on the capture step for a Generic Analog Sensor, it means your Pi's Node-RED flow is not yet sending the raw signal value. Redeploy your flow from the Configurator to resolve this.

Use Cases

Correcting a Drifted Temperature Probe

A grower notices that the dashboard reports 23.4°C but a calibrated reference thermometer placed next to the sensor reads 22.0°C — a consistent 1.4°C high offset. Using the Single-point offset mode, they place the sensor in a container of water alongside the reference thermometer, wait for the reading to stabilize, then enter 22.0 as the reference value. The wizard computes a −1.4°C offset. After applying, all future readings from that probe are shifted down by 1.4°C on the dashboard.

Recalibrating a pH Probe with Two Buffer Solutions

A food and beverage producer calibrates their pH probe using two standard buffer solutions: pH 4.0 and pH 7.0. Using the Two-point linear mode:

  • They place the probe in the pH 4.0 buffer, wait for stabilization, and enter 4.0 as the reference.
  • They rinse the probe, place it in the pH 7.0 buffer, and enter 7.0 as the reference.

The wizard corrects both the offset and the slope. The result is accurate readings across the pH 4–7 range. The correction is automatically synced to their Pi so the Node-RED flow also uses the updated calibration.

Linearizing a Non-Linear Analog Sensor

An engineer using a custom analog pressure sensor notices the output voltage is not perfectly linear across the operating range. Using the Multi-point polynomial mode on the Generic Analog Sensor part, they capture five reference points by connecting a calibrated pressure gauge at five known pressures. The wizard fits a quadratic curve through the points and shows a preview chart confirming the correction closely follows the reference values. After applying, displayed pressure readings closely match the calibrated reference gauge.

Troubleshooting

The sensor I want to calibrate is not in the list

The sensor list shows only sensors that are currently reporting data. Common reasons a sensor might be missing:

  • The Pi device is offline. Check the device health panel to confirm it is connected.
  • The sensor was recently added to the flow but the flow has not been deployed yet. Go to the Configurator and deploy the flow to activate new sensors.
  • The sensor key in Node-RED does not match what the dashboard is subscribed to. Check the Configurator for any validation warnings about signal mismatches.

The live reading on the capture step is not updating

The wizard shows live data from the dashboard's real-time connection to the Pi. If the reading appears frozen:

  • Check that your browser has a network connection and the device shows as online.
  • Reload the page and try opening the wizard again.
  • If you are on a metered connection, sensor push intervals may be slower than normal (30 seconds or more at the Hobby tier). Wait for the next push before capturing.

The preview value looks wrong or the wizard shows an error

  • Two reference values are the same raw reading: For two-point calibration, the wizard requires that the two raw readings be different. If you captured both points at the same sensor output, move the sensor to a genuinely different condition for the second point.
  • Reference value was left blank or is not a number: Make sure you have entered a numeric reference value for every capture step. The preview will not compute until all required points have valid entries.
  • Polynomial fit produces a poor curve: If the curve preview does not match your reference points well, try adding more reference points at intermediate values, or switch to a higher polynomial degree. Avoid using more points than necessary, as very high-degree polynomials can produce unexpected behavior between reference points.

The calibration appears to have no effect after applying

  • Wait a few seconds for the next live reading to arrive after saving. The correction is applied to incoming data, so the display updates only when a new sensor reading comes in.
  • If readings still appear uncorrected after a minute, reload the page. Occasionally the browser cache holds an older calibration state.
  • Check that you applied the calibration to the correct sensor. If multiple sensors of the same type are connected, their names in the list come from the sensor registry. Sensor display names can be changed in the Configurator.

A "Needs redeploy" warning appears for a Generic Analog Sensor

This means the Pi's active Node-RED flow is not sending the raw pre-conversion signal value alongside the engineering value. This is required for polynomial and multi-point calibrations on Generic Analog Sensors. Go to the Configurator and redeploy the current flow to your device. Once the Pi restarts the flow, the raw value will begin arriving and the warning will disappear.

Known Issues and Limitations

  • PID controllers and alarms use raw values, not calibrated values. If your setpoint or alarm threshold targets a calibrated reading, you will need to adjust the setpoint or threshold to account for the calibration offset. For example, if a temperature probe reads 2°C high and you apply a −2°C offset calibration, a setpoint of 25°C should still be set to 25°C (the target engineering value), but you may need to verify behavior against your actual process conditions.

  • Historical data on the Analytics page is not retroactively corrected. Calibration is applied to live incoming data only. Historical readings stored before calibration was applied will still show uncorrected values in the Analytics chart.

  • Polynomial calibration is only available for the Generic Analog Sensor part. Other sensor types (I2C, UART, MQTT, Modbus) use offset or linear modes only.

  • Each sensor can have only one active calibration at a time. Applying a new calibration overwrites the previous one. The previous calibration's reference points are not retained.

calibrationsensorsaccuracysettings
Previous
Anomaly Detection
Next
Notification Center