Gas analyzer calibration procedure

 How to Calibrate a Gas Analyzer

If you've spent any time around a kiln, boiler, or any combustion process, you already know the drill: 

the gas analyzer is the one instrument nobody thinks about until it starts lying to you. And it will lie to you eventually 

 every sensor drifts, every optical bench gets dirty, every electrochemical cell ages. Calibration is just how you catch it before it costs you.

Here's the thing most manuals don't tell you plainly:

 calibration isn't one procedure, it's three habits

 zero

span

 a sanity check 

done consistently, on a schedule you actually follow.

Why It Drifts in the First Place

Before touching any buttons, it helps to know what you're fighting:

Electrochemical cells (O₂, CO, NOx sensors) slowly consume their internal electrolyte. Output drops over months, sometimes faster in hot, dusty environments.

NDIR (non-dispersive infrared) benches drift when the optical path gets coated with dust or moisture, or when the IR source ages.

Zirconia O₂ cells drift with temperature swings and cell aging — this is the one that fools people because it can drift and still respond fast.

Sample lines and filters clog or leak, which looks exactly like sensor drift but isn't.

If you skip this step and jump straight to calibration, you'll sometimes "calibrate out" a sample-line problem instead of fixing it, and it'll come back within a week.

Step 1: Zero Calibration

This sets the analyzer's reading at the bottom of its range, usually using pure nitrogen (for O₂ analyzers, zero air is used instead, since O₂ analyzers need a truly oxygen-free gas to find zero — nitrogen at 99.999% purity is standard).

Practical steps:

Isolate the analyzer from the process sample which may be done automatically through a SOV.

Feed zero gas at the manufacturer's specified flow rate (usually 0.5–1.5 L/min — check your unit).

Let the reading stabilize.

 Don't rush this

NDIR and electrochemical cells can take 3–5 minutes to settle, zirconia cells often less.

Adjust to zero using the analyzer's calibration menu (most modern units  do this via a soft-key .

Step 2: Span Calibration

This sets the top of the range using a certified span gas , a known concentration close to your expected process reading (not necessarily full scale).

Practical steps:

Feed the span gas at the same flow rate used for zero.

Wait for stabilization.

Adjust the reading to match the certificate value on the gas cylinder.

Cross-check: 

cycle back to zero gas briefly. If zero has shifted after spanning, something's off — sample conditioning, cell fatigue, or a leak and it's worth chasing down rather than re-zeroing and moving on.

A note on gas certificates: 

span gas cylinders have a certified accuracy (often ±2%) and a shelf life. Using an expired cylinder is one of the most common reasons a "calibrated" analyzer still reads wrong.

Step 3: The Sanity Check (Cross-Reference)

After zero and span, verify against a second source if you have one — a portable analyzer, a lab sample, or a known process condition (e.g., ambient air should read ~20.9% O₂ dry). This catches errors that a clean zero/span cycle won't: sample line contamination, a partially blocked probe, condensation in the line, or a converter (for NOx units using NO2-to-NO conversion) that's degraded.

How Often to Calibrate

There's no universal number, but here's a working rule of thumb used across cement, power, and refinery plants:

Analyzer type

Typical interval

Zirconia O₂

Weekly to monthly

Electrochemical (O₂, CO)

Monthly

NDIR (CO, CO₂, NOx)

Monthly to quarterly

After any maintenance, sensor replacement, or sample line work

Immediately

Environments with heavy dust, high moisture, or temperature swings (kiln inlets, cooler vents) push you toward the shorter end of these intervals.

Common Mistakes Worth Naming

Calibrating at the wrong flow rate. Most drift complaints trace back to someone feeding gas at whatever flow the regulator happened to be set to, not the analyzer's spec.

Not letting the reading truly stabilize. A reading that looks stable for 30 seconds isn't the same as one that's stable for 3 minutes.

Ignoring a zero shift after span. This is diagnostic information, not noise — don't just re-zero and walk away.

Skipping the sample system. A perfectly calibrated analyzer downstream of a leaking probe or clogged filter will give you perfectly wrong numbers.

Using out-of-date span gas. Always check the certificate date before assuming the analyzer is the problem.

The Bottom Line

Calibration isn't glamorous, but it's the difference between an analyzer you trust and one you're quietly second-guessing every time it disagrees with what your process is telling you. Zero, span, cross-check, log it, and stick to the interval — that's really the whole job.