Anyone who has characterized gas sensors knows the classic workflow: set one concentration, wait for the response, remove the gas, wait for the baseline to recover, then set the next concentration. A full response curve can consume a full working day, repeating the same manual actions dozens of times.
Universities and research laboratories abroad solved this problem decades ago — not by asking operators to be more diligent, but by letting the gas mixing system execute a pre-programmed concentration sequence: which concentration at which time, how long to hold it, when to switch back to zero gas, how many cycles to run.
This article explains why programmed concentration profiles have become a standard approach in gas sensor testing, and what the gas delivery/mixing stage must deliver to make such automation trustworthy.
A Brief History: Programmable Gas Mixing Is Not New
Figure 1. A programmed concentration profile: concentration steps, hold times, zero-gas purges and cycle count are all defined by the program.
1980 — University of Washington: the programmable gas mixer
In 1980, Jonathan Jacky at the University of Washington School of Medicine published a programmable gas mixer in the Journal of Applied Physiology. A microcomputer controlled one on/off solenoid valve per gas; each valve was rapidly pulsed open and closed, and the concentration of each gas became proportional to the fraction of each cycle the valve stayed open (pulse-width modulation).
Any arbitrary time-varying concentration waveform could be pre-programmed into computer memory and reproduced with accuracy and repeatability.
Aspect
The 1980 Washington approach
Control
One solenoid valve per gas, microcomputer-driven PWM
Principle
Duty cycle ≈ concentration, no feedback needed
Waveforms
Pre-programmed step, ramp, sine, arbitrary profiles; up to ~4 Hz
The insight was simple and durable: translate "concentration value" into "how long the valve stays open"— and a concentration-vs-time profile becomes a sequence that can be programmed, edited and repeated exactly.
1987 — U.S. Naval Research Laboratory: unattended operation
In 1987, Grate and colleagues at the Naval Research Laboratory described an automated vapor-generation and data-collection instrument. It managed up to 12 vapor sources (bubblers or permeation tubes), diluted each to ~1–10,000 mg/m³, delivered programmable flow, and could generate two-component mixtures.
The entire system was fully automated by a small personal computer and permitted completely unattended operation during elaborate vapor-exposure sequences, automatically storing test conditions and sensor responses.
Note the phrase "completely unattended operation" — in a 1987 paper. The reason is practical: sensor characterization routinely runs for hours or days across dozens of concentration points and repeats; manual operation is neither realistic nor repeatable.
The Modern Version: Saarland University's Automated Gas Mixing System (2014)
In 2014, the Laboratory for Measurement Technology (LMT) at Saarland University (Helwig et al., Measurement Science and Technology 25, 055903) published a computer-controlled gas mixing system for automated characterization of gas sensors. Its design goals:
·Generate trace gases (e.g., benzene, naphthalene) using permeation furnaces;
·Pre-dilute test gases to cover a wide concentration range;
·Run automated test procedures under a LabVIEW-based user interface.
Pre-dilution covering a 1:62,500 concentration range within one test procedure
Controlled variables
Up to 6 test gases, humidity, oxygen content, total flow — and their variation over time
Control interface
LabVIEW graphical interface; automated test procedures
Concentration level
ppb range and lower, enabling sensor analysis at trace concentrations
The key sentence is: up to six test gases, humidity, oxygen content, total flow and their variation over time can be controlled. In one test run, concentration, background humidity and oxygen can all change according to a time program — precisely the "controlled atmosphere script" a sensor test chamber needs.
2014 — University of Freiburg: real-world conditions in the lab
Also in 2014, Kneer et al. at the University of Freiburg (IMTEK) published an apparatus that emulates real-world conditions in the laboratory for MOX (metal-oxide semiconductor) sensor characterization: precise control of trace gas concentrations, oxygen and humidity levels (all key to MOX surface chemistry), precise sensor temperature control, and per-device power consumption measurement.
Sensors are tested not in an idealized gas, but in a controlled atmosphere set to the target scenario.
2020 — Qatar University: the "calibration version" of programmed gas mixing
If Saarland solved characterization, Qatar University solved calibration. Benammar et al. (Sensors, 2020) proposed a dedicated smart calibration rig for electrochemical gas sensor nodes, with parallel/batch calibration and full automation:
Element
The Qatar University smart rig
System
Gas mixing system + temperature control + test chamber + process-control PC
Calibration
LabVIEW platform controls the calibration environment, automating all phases
Data & models
Logs sensor data, fits a best-fit equation per sensor, uploads it to the node for next deployment
Communication
IEEE 802.15.4 (ZigBee), same protocol used in field deployment
Here the mixing program is not just "deliver a concentration" — it is the timeline of the entire calibration procedure: what concentration at what time, how long to hold each level, how to cycle, all determined by program.
Figure 2. Four decades of programmed gas mixing in sensor research, from the 1980 programmable gas mixer to the 2020 smart calibration rig.
What Programmed Concentration Profiles Solve
Across four decades of independent research, the value proposition is consistent:
Value
What it delivers
Evidence
Repeatability
The same program can be re-run exactly, eliminating operator variability
Jacky 1980; Helwig 2014
Coverage
One program covers a wide concentration range (e.g., 1:62,500) or many components without changing bottles
Grate 1987; Helwig 2014
Unattended operation
Long exposure sequences and overnight runs need no operator presence
Grate 1987
Multi-variable linkage
Concentration, humidity, oxygen, temperature change together on a time program, emulating real conditions or calibration flows
Kneer 2014; Benammar 2020
"Multi-variable linkage" deserves emphasis. Real-world gases do not hold a constant concentration — concentrations fluctuate, humidity changes, oxygen varies with environment. To study sensor behavior under these dynamic conditions in the lab, the mixing system must control multiple variables and let them vary over time.
Saarland's system writes "variation over time" directly into its controlled-variable list; Freiburg's apparatus controls oxygen and humidity alongside trace concentration. Both answer the same requirement: controlled does not mean "fixed concentration" — it means "changing by program, on schedule."
What the Gas Mixing Stage Must Deliver
Figure 3. Where programmed gas delivery sits in the sensor test chain - the gas path is unchanged; the program, interface and communication layer do the work.
Condition provided by the mixing stage
If unstable, what data is corrupted
Concentration steps on a time program
Distorted response curves; wrong sensitivity and response-time calculations
Zero-gas purge and cleaning programs
Baseline drift; carry-over from the previous concentration contaminates the next point
Votre message doit contenir entre 20 et 3 000 caractères!
english
français
Deutsch
Italiano
Русский
Español
português
Nederlandse
ελληνικά
日本語
한국
polski
فارسی
বাংলা
ไทย
tiếng Việt
العربية
हिन्दी
Türkçe
indonesia
