Duct Smoke Detectors for the NICET Exam: NFPA 90A Thresholds, Sampling Tubes, and HVAC Shutdown (2026)
Duct smoke detectors trip up NICET candidates: 2,000 vs 15,000 cfm, sampling tube rules, supervisory vs alarm, and why they never replace area detection.
Problem this solves
This guide solves the next-step problem for Fire Alarm candidates: it explains what matters, then gives you a direct way to test that knowledge with practice questions instead of guessing what to study next.
Why Duct Detectors Get Their Own Set of Questions
Duct smoke detectors are the one device on a fire alarm system that most technicians install, almost nobody loves, and a surprising number get wrong on the exam. Part of that is because they live in two codes at once. The requirement to have one comes out of NFPA 90A and the mechanical code — not NFPA 72. The requirement for how it is wired, supervised, tested, and annunciated comes out of NFPA 72. NICET writes questions from both directions, and candidates who studied only NFPA 72 walk in without the cfm thresholds that half the questions depend on. The other reason is that a duct detector does a fundamentally different job than every other detector on the job. A ceiling smoke detector exists to find a fire in a room and tell people to leave. A duct detector exists to stop an air handler from pumping smoke into rooms that are not on fire. That is smoke control, not detection of an occupancy hazard, and once you internalize that distinction the code language stops feeling arbitrary. Nearly every trick question on this topic is testing whether you know that duct detectors protect the system, not the space.
Try it now — no account, no download
Practice Fire Alarm questions free
Answer 5 real Fire Alarm exam questions with instant explanations before you keep reading.
Fire Alarm · Question 1 of 5
What is the primary difference between a conventional and an addressable fire alarm system?
Pick an answer to see the explanation + an instant AI breakdown.
The Two Numbers: 2,000 cfm and 15,000 cfm
If you memorize nothing else from this article, memorize these two thresholds — they show up nearly verbatim on exams. SUPPLY SIDE: greater than 2,000 cfm. NFPA 90A requires a duct smoke detector in the supply air stream of any air distribution system with a capacity greater than 2,000 cfm, located downstream of the filters and ahead of any branch connections. Both halves of that location rule are testable. Downstream of the filters means you sample after the air has passed through them, so a smoldering filter is on the detector's side of the airflow. Ahead of branch connections means you catch the air before it splits and heads off to individual zones, so one detector actually covers the system. RETURN SIDE: greater than 15,000 cfm AND serving more than one story. This is the one candidates blow. The return-side requirement has two conditions joined by AND, and exam distractors routinely give you a 20,000 cfm single-story system to see whether you call for a return detector. You do not need one — the system does not serve more than one story. Where the return detector is required, it goes at each story, prior to the connection to a common return and prior to any recirculation or fresh-air inlet connection. THE AREA-DETECTION EXCEPTION: return system detectors are not required where the entire space served by the air distribution system is protected by a system of area smoke detectors. Note carefully what that exception does and does not do — it can relieve the return detector, because area detection already sees the smoke before it reaches the return. It does not run the other way. Area detectors do not get you out of the supply-side requirement, and duct detectors never get you out of area detection. Watch also for the fact that your local mechanical code (IMC Section 606 in most jurisdictions) restates these thresholds and occasionally amends them. NICET tests the model-code numbers, but on the job the AHJ's adopted edition wins.
Free: the 20 facts Fire Alarm exams keep testing
Top 20 most-tested facts for your exam. Delivered instantly.
Supervisory, Not Alarm — And Never a Substitute for Area Detection
Two rules from NFPA 72 Chapter 17 carry a lot of exam weight. First, duct detectors normally initiate a SUPERVISORY signal, not an alarm. A duct detector's purpose is to shut down or reconfigure the air handler, not to evacuate a building, so where the building has a fire alarm system the duct detector is typically connected to produce a supervisory signal at a constantly attended location. There is an important carve-out: the supervisory signal at a constantly attended location is not required where the duct smoke detectors instead activate the building's alarm notification appliances. In other words, either it annunciates as a supervisory to someone who can act on it, or it is wired to sound the building — it cannot do neither. If the building has no fire alarm system at all, the duct detector still has to provide a local audible and visible indication, because someone has to know the air handler tripped. Second, a duct detector is never a substitute for required open-area detection. NFPA 72 states this outright, and the reasoning is purely physical: a duct detector only sees smoke when the fan is running. Shut the air handler off at night, start a fire in the space, and the detector inside the duct sees nothing at all. It also samples heavily diluted air — smoke from one room mixed with return air from an entire floor — so its effective sensitivity to a room-level fire is poor even when the fan is running. Expect at least one question phrased as a design shortcut, such as whether duct detection can satisfy area coverage in a small tenant space. The answer is no, and the reason to give is fan-off operation.
Sampling Tubes: Length, Orientation, and the 6-to-10 Duct Width Rule
The mechanical installation of the sampling tubes is where field technicians and exam candidates diverge, because the exam expects published good practice, not whatever got installed on your last job. LENGTH: the inlet sampling tube must extend at least two-thirds of the way across the width of the duct so it samples a representative cross-section rather than one lane of airflow. For ducts wider than 36 inches, use a tube longer than the duct width and support it at the far side — any tube 36 inches or longer must be supported at both ends, or it vibrates, sags, and eventually fatigues. ORIENTATION: rotate the inlet tube so the row of sampling holes points upstream, directly into the oncoming air; the exhaust tube's hole faces downstream. Get these backwards and the pressure differential across the sensing chamber collapses — the detector will pass a magnet test on the bench and fail to see real smoke in the duct. LOCATION: mount on a flat, straight section 6 to 10 duct widths away from any bend, transition, or obstruction, because air coming off an elbow is turbulent and stratified and will not give you a representative sample. When the mechanical layout makes that impossible — and it often does — that is a documented deviation, not something you quietly ignore. VERIFICATION: proper sampling is confirmed with a differential pressure gauge reading across the inlet and exhaust tubes. The working window most manufacturers publish is greater than 0.01 inches of water and less than 1.2 inches of water. If you are below the floor of that range, you likely have the tubes reversed, the wrong tube length, or too little velocity.
The Velocity Window: 300 to 4,000 fpm
Listed duct smoke detectors are rated for a specific range of duct air velocity, and the range to know is 300 to 4,000 feet per minute. Below 300 fpm there is not enough differential pressure to draw a sample through the sensing chamber; above 4,000 fpm the air moves past too fast for a reliable sample and can push the detector outside the conditions it was listed for. This matters more than it used to. Variable air volume systems throttle down at part load, and a duct that runs 900 fpm at design flow can drop under the low limit at overnight setback. When an exam question hands you a VAV system, a low-flow-listed detector is usually the intended answer. Field practice follows the same logic: measure velocity with an anemometer during commissioning at both design flow and minimum flow, and record both numbers. A detector operating outside its listed velocity range is not merely an underperformer — it is outside its listing, which is a compliance problem, not a performance opinion.
Remote Test Station and Remote Annunciation
Duct detectors end up in the least accessible places on a project — above hard ceilings, on rooftop units, inside mechanical shafts. Codes handle that with two related requirements that candidates routinely blur together. A REMOTE TEST/RESET STATION — the familiar key- or magnet-operated device with an indicating LED — lets a technician exercise a detector that is not readily accessible without renting a lift. It is a testing accessory. It does not by itself satisfy annunciation. REMOTE ANNUNCIATION is separate: where the detector is concealed, its location and status must be annunciated where it can actually be seen — at the fire alarm control unit, at an annunciator, or at another approved location — with the specific detector identified. An addressable duct detector on an SLC handles this natively, because the panel reports the device and its custom location label by address. A conventional duct detector on an initiating device circuit needs a separately labeled remote indicator, because 'IDC 4 supervisory' tells nobody which of five rooftop units tripped. That is one of the cleanest practical arguments for addressable duct detection, and NICET has been known to ask it as a design-judgment question rather than a lookup.
Duct Detectors vs. Air-Sampling (ASD): Don't Confuse the Two
Both involve tubes. They are otherwise unrelated devices, and the exam does put them in the same answer set. A DUCT SMOKE DETECTOR is a spot-type sensing chamber mounted on the outside of a duct, drawing a passive sample through inlet and exhaust tubes using the pressure differential the fan itself creates. It has no aspirator of its own, it has one sensing point, it operates at standard spot-detector sensitivity, and its job is smoke control at the air handler. An AIR-SAMPLING (ASPIRATING) DETECTOR — the ASD or VESDA family — actively pulls air through an engineered pipe network with its own aspirator fan, past a high-sensitivity detection chamber, with calculated hole sizes and a transport time requirement of 120 seconds or less from the most remote sampling point. Its job is very early warning in an open area, a cabinet, or a high-airflow space, and it does count as area detection. Summarized as four contrasts: passive tube versus powered pipe network; standard sensitivity versus high sensitivity; smoke control versus early warning; not area coverage versus area coverage. If you want the ASD side in depth, it has a full treatment in the [air-sampling smoke detection guide](/blog/nicet-fire-alarm-air-sampling-aspirating-smoke-detection-asd-transport-time-2026).
HVAC Shutdown: The Function vs. the Device
A duct detector activation has to do something, and what it does is an emergency control function. On activation, the detector shuts down the air handling unit or, in an engineered smoke control system, drives fans and dampers to their smoke control mode. That command runs through a relay or an addressable control module, and the module has to be located and powered so the function still happens under the fault conditions the code anticipates. Two distinctions matter on the exam. First, fan shutdown is the FUNCTION and the duct detector is only one of several possible INITIATORS — an area smoke detector, a sprinkler waterflow switch, or a smoke control sequence can all command the same fan. A question asking what shuts down the AHU may have a correct answer that is not the duct detector. Second, shutdown is not automatic restart. Restoring the detector does not put the fan back in service by itself; restart is a manual operation per the approved sequence of operation, and the record of completion should say so explicitly. The wider set of these functions — elevator recall, shunt trip, door holders, and HVAC — is covered in the [emergency control functions guide](/blog/nicet-fire-alarm-emergency-control-functions-elevator-recall-hvac-door-holders-2026).
Testing and Maintenance: Prove the Airflow, Not Just the Electronics
Duct detectors are tested at the same annual frequency as other smoke detectors under the NFPA 72 Chapter 14 inspection, testing, and maintenance tables, but with an added performance step: it is not enough to prove the sensing chamber responds. You have to prove air is actually being drawn through it. That means checking the sampling tube pressure differential and confirming that activation performs its intended emergency control function — the fan actually stops. Duct detectors also foul faster than ceiling detectors, because dust that would settle harmlessly on a ceiling device is force-fed through a duct detector's chamber every hour the fan runs. Expect more sensitivity drift, more nuisance trouble conditions, and shorter service life than the rest of the system, and budget for it on maintenance contracts. The exam phrasing to watch for is the difference between verifying the signal path and verifying the sample path: a magnet or a remote test station activation proves the electronics and the wiring back to the panel, not that the tubes are drawing air. Proving the sample path takes smoke or an approved aerosol introduced into the duct. For the full ITM schedule across device types, see the [NFPA 72 testing frequencies guide](/blog/nicet-fire-alarm-testing-frequencies-nfpa-72-itm-schedule-2026).
How to Study This Topic
Build a single index card with the four numbers: 2,000 cfm supply, 15,000 cfm return AND multi-story, two-thirds of the duct width for the sampling tube, and 300 to 4,000 fpm for velocity. Those four values plus the supervisory-signal rule account for the large majority of duct detector questions on a NICET fire alarm exam. Then drill the reasoning behind them, because NICET increasingly asks why rather than what. Practice saying, in one sentence each: why the supply detector goes downstream of the filters, why the return threshold has two conditions joined by AND, why the fan-off case kills the substitution argument, and why an addressable detector solves an annunciation problem a conventional one does not. If you can explain the reason, the number stays put; if you only memorized the number, a rephrased question will take it back from you. Then get reps against real questions. Free NICET-style fire alarm practice questions are at [VoltExam's fire alarm question bank](/questions/fire-alarm), and the structured path through NFPA 72 chapter by chapter is at the [fire alarm study guide](/study/fire-alarm). Download the Fire Alarm Prep app to drill duct detection, initiating devices, NAC and SLC design, and the rest of the NICET Level I and II knowledge areas on your phone — between service calls, on breaks, wherever you actually have ten free minutes: [VoltExam Fire Alarm Prep](/apps/fire-alarm).
Studying with VoltAI
Stuck on a Fire Alarm question? Ask VoltAI why.
VoltAI is a built-in AI tutor that explains every practice answer in plain language — citing the actual code, protocol, or standard — and answers your follow-ups until it clicks. Most prep apps just tell you the right letter; VoltAI teaches the reasoning the exam actually tests, so you pass by understanding, not memorizing. Free to try, no account.
Try VoltAI free →Your next best step
Turn this guide into practice
Use the article to understand the topic, then do a short web practice session to find your weak spots. Paid web access is optional after the free preview.