A gas detector that eventually reads the right value but takes too long to get there may miss the window when a fast response matters most. T90 is the number that captures that speed. It measures how quickly a detector reacts once gas reaches it, and it tends to get worse as detectors age or their gas path gets blocked. This guide defines T90, explains why it matters for catching real releases, and covers what slows a detector down and how the number is checked during testing.
T90 response time in one line: T90 response time is the time a gas detector takes to reach 90% of its final stable reading after being exposed to a step change in gas concentration. It measures how fast the detector responds to a real release, and a slow T90 means gas can be present for longer before the alarm level is reached. Aging sensors, dirty splash guards, long sample lines, and calibration caps left on all degrade T90.
When a gas detector is suddenly exposed to a known concentration, its reading does not jump instantly to the final value; it climbs toward it over a few seconds or tens of seconds. T90 is the time taken to reach 90% of that final reading. Related figures like T50 mark the time to reach half the final value, and together they describe the shape of the response curve. T90 is the common headline because it corresponds to being nearly at the true reading, which is what matters for crossing an alarm threshold.
The reason T90 matters is that detection speed is part of safety. A release does not wait, and the sooner a detector reaches its alarm level, the sooner ventilation, isolation, or shutdown can begin. Two detectors that both eventually read correctly are not equal if one reaches the alarm point in seconds and the other takes far longer; during that lag, gas keeps accumulating. In a fast-developing high-pressure release, a sluggish detector effectively shrinks the margin the whole safety scheme was designed around.
T90 also depends on how gas gets to the sensor. A diffusion detector waits for gas to reach it by natural movement of the air, so its response includes the time for gas to work through any barrier in front of the sensing element. A sample-draw system actively pulls the atmosphere through tubing to the sensor, which can help or hurt depending on the tubing length. Either way, the specified T90 assumes a clean, unobstructed path, and anything that impedes that path adds to the real-world response time.
The most quietly dangerous cause of slow response is a sensor aging out. As an electrochemical cell or catalytic element degrades, it may still eventually reach a reading but take longer to get there, and it can lose sensitivity so the final value is lower than it should be. A detector that once had a brisk T90 can drift to a sluggish one without any obvious sign in clean air, which is one reason functional testing with gas, rather than just watching the live zero, is essential.
Physical obstruction of the gas path is another common culprit. Splash guards, dust filters, and sinter caps in front of the sensor can clog with dirt, paint, ice, or corrosion, slowing the passage of gas to the element and stretching the response time. In sample-draw setups, long or partly blocked sample lines add transport delay, so the detector cannot react until the gas has physically traveled the tubing. Regular inspection and cleaning of these barriers keeps the response honest.
The most avoidable cause is human: leaving a calibration cap or a weather protection cover on after maintenance. A cap that seals the sensor from the atmosphere does not stop the detector from reading a comfortable zero, but it prevents gas from ever reaching the element, so the real-world response time becomes effectively infinite. This is a genuine and recurring field error, and it is exactly the kind of silent failure that a functional gas test catches and a glance at the display does not.
T90 is checked by exposing the detector to a step change of known gas and timing how long it takes to reach 90% of the expected reading. This is often folded into calibration or a functional test: apply the gas, watch the reading climb, and note whether it reaches value within an acceptable time. A response that is noticeably slower than the detector's specification, or slower than it was on previous tests, is a warning that the sensor or its gas path has degraded, even if the detector eventually reaches the right number.
Typical response times vary by sensor technology. Catalytic and infrared combustible sensors tend to respond quickly, while electrochemical toxic cells can be somewhat slower, and the exact figures depend on the specific detector and its guards. Because these values differ, the meaningful comparison is a detector against its own specification and its own history, rather than against some universal target. A trend of increasing T90 across successive tests is a clear maintenance signal.
This is where recorded test data pays off. Logging each detector's measured response time over time, alongside its calibration and bump results, lets a maintenance platform surface detectors whose response is creeping upward before they become dangerously slow. Merobix keeping that history, together with live fault status from the detectors, means a remote team can spot an aging or obstructed detector at an unmanned site from the trend rather than waiting to discover a sluggish response during an actual release.
Both describe how fast a detector responds to a step change in gas, but they mark different points on the rising curve. T50 is the time to reach 50% of the final reading, and T90 is the time to reach 90%. T90 is the more commonly quoted figure because reaching 90% is close to the true value and therefore relevant to crossing an alarm threshold.
A calibration cap seals the sensor from the surrounding atmosphere. The detector can still read a normal zero, so nothing looks wrong on the display, but gas from a real release can never reach the sensing element. The effective response time becomes infinite, meaning the detector would never alarm on a genuine leak. This is why functional gas testing, not just watching the reading, is essential.
A technician applies a step change of known calibration gas to the detector and times how long the reading takes to reach 90% of the expected value. This is often done as part of calibration or functional testing. A response that is significantly slower than the detector's specification, or slower than in previous tests, indicates a degraded sensor or an obstructed gas path that needs attention.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.