Gas Sensor Interface Circuit Design

A gas sensor turns a chemical concentration into a small electrical change, and the circuit around it has to do two jobs at once: drive the element under stable conditions and read a signal that moves slowly. Most field failures in these designs come from the drive side rather than the amplifier, so the heater and the reference conditions deserve as much attention as the analog front end.

How the Three Main Sensor Types Behave

A metal oxide gas sensor is a heated ceramic whose resistance falls in the presence of a reducing gas. Its resistance range is wide, from kilohms in clean air to a few hundred ohms in a strong concentration, so the front end sees a two or three decade change rather than a small shift around a fixed point.

An electrochemical gas sensor produces a small current, typically nanoamps to microamps, and behaves as a current source with a high output impedance. An infrared gas sensor measures absorption across a path and outputs a voltage that is close to linear in concentration. Each type sets the input impedance, the gain and the noise budget of the circuit.

A catalytic bead sensor changes resistance with combustion on a hot element, and a photoionisation sensor produces a current like the electrochemical cell but with a much larger standing current. The interface design follows from the source impedance, so identify that first and everything else falls into place.

The Heater Driver and Duty Cycle

A metal oxide gas sensor needs a heater, and the heater is the largest load on the board. It draws between tens and hundreds of milliamps at a low voltage, and the resistance of the heater changes with temperature, so a constant voltage drive produces a varying element temperature as the ambient moves.

A heater driver built around a chopper or a linear regulator holds the element temperature steady, which removes one source of sensor drift. Where power matters more than accuracy, a pulse width modulated drive with a fixed duty cycle is common, and the temperature then follows the ambient to some degree.

Pulsing the heater creates a second problem. The current step couples into the measurement path through the supply and the ground, and it can be larger than the gas signal itself. The cure is to measure when the heater is off, to keep the heater return current out of the analog ground, and to slew the drive rather than switching it in a few microseconds.

Building the Analog Front End

An electrochemical gas sensor wants a transimpedance amplifier with a very high feedback resistance, often in the tens of megohms, and a bias voltage that holds the working electrode at its recommended potential. The bias network has to be quiet, because any noise there appears directly as current error at the input.

A metal oxide gas sensor wants a load resistor and a buffer, or a divider followed by an amplifier with a gain that suits the measurement range. Choosing the load resistor is a compromise: a larger value gives more output for the same resistance change but slows the response and adds thermal noise.

The analog front end should include a low pass filter with a corner well below the sensor response time. Gas concentrations do not change in milliseconds, so a corner of a few hertz removes mains hum, heater artifacts and the noise of the reference, and it costs nothing in signal.

Understanding Sensor Drift

sensor drift is the slow change in output that occurs with no change in gas concentration. It has several causes: ageing of the sensing layer, contamination from vapours in the environment, and a shift in the heater temperature. Drift is the reason a circuit with excellent short term accuracy still needs periodic attention.

Drift is best handled by keeping the operating conditions constant and by measuring the baseline often. A sensor that is held at a fixed temperature and humidity drifts less, and a design that records a clean air reading at every power up can correct a large part of the residual change in software.

Temperature and humidity both affect the reading, and the effect is large enough to matter in a metal oxide sensor. A nearby temperature sensor and a humidity sensor let the firmware compensate, and the compensation should be based on measurements across the intended range rather than a single correction factor.

Calibration and Baseline Tracking

calibration establishes the relationship between the sensor output and the gas concentration. A two point calibration with clean air and a known span gas is standard for a metal oxide sensor, while an electrochemical cell is usually calibrated against a certified gas and then checked with a zero reading.

Because the response changes over time, the practical approach is a factory calibration followed by periodic zero checks in the field. The zero check is easy to automate: a valve or a filter presents clean air to the sensor, and the firmware stores the offset that brings the reading back to the expected value.

The span check is harder because it needs a gas source. Where a full calibration is not practical, the trend of the zero offset and the response to a known internal source give an early warning that the sensor is nearing the end of its useful life, and the replacement can be scheduled rather than reactive.

Temperature, Humidity and Interference

Cross sensitivity is the response of a sensor to a gas other than the one it is meant to measure. A metal oxide element reacts to many reducing gases, and an electrochemical cell may respond to a related compound. The datasheet lists the typical cross sensitivities, and the firmware can correct the largest ones if the other gases are measured.

Humidity changes both the baseline and the sensitivity of most sensors. A sudden change in humidity produces a transient that lasts minutes, which is why the humidity reading should be recorded alongside the gas reading and the compensation should include a settling term rather than a simple instantaneous correction.

Airflow over the sensor affects the reading as well. A sensor in a sealed enclosure with a diffusion path behaves differently from one exposed to a moving stream, and the calibration has to be performed with the same airflow that the product will see. Mechanical design is therefore part of the measurement chain, not a separate topic.

Alarm Thresholds and Hysteresis

An alarm threshold has to sit above the noise of the measurement and below the level at which the gas becomes a hazard. Setting it from the peak to peak noise of a long recording is more reliable than setting it from the datasheet resolution, because the noise in a real installation includes supply and thermal effects.

Hysteresis prevents an alarm that chatters at the threshold. A simple approach is to trigger at one level and clear at a lower one, with a delay of a few seconds on both edges. The delay also rejects the transient that follows a door opening or a heater cycle.

Where the alarm drives a relay or a horn, the load current must not share the ground of the measurement path. A separate return for the output stage, meeting the analog ground at one point, keeps the switching current out of the sensor signal. The general rules for that split are covered in our guide to mixed signal board design.

Layout, Contamination and Verification

Keep the sensor and its front end in a quiet area with a guard ring around the high impedance node of an electrochemical input. Flux residue, fingerprints and moisture on the board around that node create leakage currents that are comparable to the sensor signal, so the area is cleaned and often coated.

Verification starts with a burn in period. A new metal oxide sensor needs hours or days at its operating temperature before the baseline settles, and the measurements taken during that time should be discarded. A drift check over a week of continuous operation shows whether the design holds its zero.

The release checks that catch a missing guard ring, a contaminated assembly or a shared ground are the same for any precision measurement board, and they are collected in our PCB design release checklist. For the visual side of assembly quality, the points we look for on a finished board are listed in judging PCB quality.

FAQ

Why does my gas sensor reading drift upward over days? A metal oxide element needs a long burn in and is sensitive to humidity and contamination. Check the baseline in clean air and confirm the heater temperature is stable.

Can I share one regulator between the heater and the analog front end? It is possible but the switching current of the heater will appear in the measurement. Feed the front end from its own filtered branch and measure with the heater off.

How often should a gas sensor be calibrated? Most industrial sensors are zero checked monthly and span checked every six to twelve months. The interval follows the drift observed in the installation rather than the datasheet.