As an Amazon Associate, Akermin earns from qualifying purchases. Product links on this page are affiliate links.
A gas sensor does not measure gas in general. It measures a particular interaction between a particular molecule and a particular sensing element, and the strength of that interaction changes with every molecule. So a threshold published without a named target gas is a measurement of the instrument in one unspecified situation, and it cannot be lined up against any other threshold. The naming matters more than the number: a figure attached to a gas can be verified, while a figure attached to nothing cannot be checked or compared.
The same sensor answers differently to different gases
Take four common sensing principles in turn. A catalytic bead burns combustible gas on a heated catalyst and measures the heat released, so its response follows the heat of combustion of whatever arrives and differs between methane, propane and hexane by a substantial factor. An electrochemical cell drives a specific reaction at an electrode, which makes it selective for a target such as hydrogen sulphide or carbon monoxide, and also gives it documented cross sensitivities where another gas can take part in the same reaction.
A heated diode releases and counts halogen ions at a hot element, so it is blind to anything without chlorine or fluorine in the molecule and responds at different strengths across the halogenated family. A metal oxide film changes resistance when reducing gases adsorb onto its surface, which makes it responsive to almost everything, including alcohols, solvents and changes in humidity, and therefore selective for nothing.
None of those four has a single sensitivity. Each has a surface of sensitivities across the gases it responds to, and a published figure is one point on that surface. Quoting the point without naming the axis it sits on is the omission this page is about. The broader treatment is in the sensitivity reference, and the instrument families themselves are set out in the combustible gas detector explainer.
Response factors are why one figure cannot cover a family
Instruments are calibrated on one gas and used on others, and the industry handles the gap with response factors. A pellistor calibrated on methane shown an atmosphere of pentane will under read or over read by a factor the manufacturer has measured and published. The displayed value is still a real measurement; it is a methane equivalent value, and the response factor converts it.
Two consequences follow for anyone reading a specification. The first is that the calibration gas belongs in the specification, because every reading the instrument ever produces is expressed in terms of it. The second is that response factor tables are specific to a sensor model and sometimes to a firmware version, so a table from one instrument does not transfer to another that happens to use a similar sensing principle.
This is the mechanism behind a very common field error: an instrument that reads correctly in a calibration bump test and then under reports a real hazard, because the hazard was a different molecule from the bump gas. Nothing is broken. The number simply means something other than what it appeared to mean, and the fault lies in the specification having been read as if the gas were irrelevant.
Percentage of the lower explosive limit is a gas specific unit too
Combustible gas instruments usually display percentage of the lower explosive limit rather than raw concentration, which is helpful because it expresses distance from a hazard rather than an abstract quantity. It is also gas specific twice over: once because the limit itself differs per gas, and once because the instrument has to be calibrated for the gas before the conversion is valid.
| Gas | Lower explosive limit, percent by volume | Volume at ten percent of the limit | Same quantity in parts per million |
|---|---|---|---|
| Methane | 5.0 | 0.50 percent | 5000 |
| Hydrogen | 4.0 | 0.40 percent | 4000 |
| Propane | 2.1 | 0.21 percent | 2100 |
| Butane | 1.8 | 0.18 percent | 1800 |
The right hand column makes the point sharply. Four instruments all displaying ten percent of the limit are sitting in atmospheres differing by nearly a factor of three in actual gas content. The displayed figure is the right one to act on for safety, since it measures proximity to ignition, but it is the wrong one to treat as a concentration, and converting it back requires knowing which gas is present. That difference is also why multi gas instruments list a separate configuration per channel rather than one global sensitivity.
What the listings name and what they leave open
Three products show the full range of disclosure. The Aituzero rechargeable refrigerant leak detector states a semiconductor sensing element, three selectable sensitivity levels, sensitivity up to 3 grams per year given also as 0.1 ounces per year, and a list of refrigerants comprising R134a, R410A, R22, R1234yf and R32. That is far better than most, and it still leaves the key question open: the threshold is one figure and the list has five members, so which refrigerant the figure belongs to is not stated in the listing bullets.
| Listing | Threshold published | Target gas attached to the threshold | What is missing |
|---|---|---|---|
| Aituzero refrigerant leak detector | 3 grams per year, also given as 0.1 ounces per year | Five refrigerants listed, none singled out | Which of the five the figure was measured on |
| Inficon combustible gas detector | Not stated in the listing bullets | Combustible, as a category | Threshold, calibration gas and units |
| Plug in natural gas and carbon monoxide alarm | Not stated in the listing bullets | Natural gas, described qualitatively | Alarm concentration for either gas it covers |
The Inficon 718-202-G1 combustible gas detector goes further the other way: its listing bullets carry a product category, a country of manufacture, a manufacturer name and a package size, and no measurement specification of any kind. The plug in natural gas and carbon monoxide alarm covers two gases whose hazards are measured in different units, flammability for one and toxicity for the other, and states a responsive sensor and alerts without a set point for either. Two gases in one housing need two thresholds, and a single adjective covers neither.
Questions to ask of any sensitivity claim
Ask which gas the figure was measured on, and treat a list of gases as a coverage claim rather than an answer. Ask what the instrument was calibrated with, since every reading it produces is denominated in that gas. Ask whether the figure is a threshold of detection or a threshold of alarm, because a needle that moves and a buzzer that sounds are different events. Ask what the air was doing, because still air figures are ceilings rather than forecasts.
If the answers are not in the listing, they are usually in the manual, and a manufacturer who publishes response factor tables has already answered most of them. Where nothing is published at all, the honest position is that the instrument is unspecified rather than poor, and it should be chosen on other grounds. The roundups of handheld gas sniffers show how unevenly this information is distributed across the market.
Frequently Asked Questions
What is a response factor?
It is the multiplier that converts a reading taken on an instrument calibrated for one gas into the concentration of a different gas actually present. If an instrument is calibrated on methane and you take it into a propane atmosphere, the displayed number is a methane equivalent reading, and the response factor tells you what to multiply or divide by to recover the truth. Manufacturers publish tables of them, and the tables only apply to the specific sensor and firmware they were measured on.
Why is percentage of the lower explosive limit gas specific?
Because the limit itself is a property of the gas rather than of the instrument. Methane becomes flammable in air at around five percent by volume while propane does so at around two, so the same displayed percentage of the limit corresponds to very different quantities of gas in the room. Two instruments showing twenty percent of the limit in two different atmospheres are reporting two different concentrations, both correctly.
Can one instrument cover a whole refrigerant family honestly?
It can cover a family where the sensing principle responds to a shared feature, which is why heated diode instruments legitimately claim all halogenated refrigerants: every one of them carries the chlorine or fluorine the element reacts to. What cannot be claimed honestly is a single sensitivity figure across the family, because the response per unit mass still varies between molecules. A family claim and a threshold claim are different claims.
Does calibration gas have to match the gas being hunted?
For a quantitative reading, yes, or you must apply a documented response factor. For a locate and repair task, a calibration on a related gas is often adequate because you are comparing a reading at the joint against a reading in clean air, and both are displaced by the same factor. The error becomes unacceptable the moment the number is used to make a safety judgement rather than a positional one.
Why do some detectors list many gases but only one figure?
Usually because the figure was measured on whichever gas gave the best result, and the list was assembled from what the sensing principle can respond to at all. Both statements can be true at once while the combination misleads. The way to read such a listing is to treat the list as coverage and the figure as a best case on one unnamed member of it, then ask the manufacturer which one.
What does a complete sensitivity statement look like?
A named gas, a threshold with its unit, the condition under which the threshold was demonstrated, and ideally the standard or procedure used. Something of the form of a stated mass flow of a named refrigerant detected in still air at a stated probe distance is complete, because every term in it can be reproduced. A bare number with a list of gases beside it is not, because nothing in it can be checked.