"Which one is the test switch?" My helper was standing on a rooftop curb holding a keyswitch plate in one hand and a flashlight in the other, and honestly I had to think about it. Because on a duct detector job, "test switch" can mean three different things, and people use the word for all of them.
So let's sort that out first.
Three things people call a test switch
- The magnet spot on the detector itself. Most duct housings have a marked place where a test magnet trips the unit. Some have a small built-in test button instead.
- A remote test station. A plate mounted somewhere you can reach from the floor, usually with an LED and a keyed switch. It lets you test a detector that lives 14 feet up in a ceiling or behind a duct wrap.
- A plain toggle on a relay or a fan control panel. Not the same thing. This one is about shutdown, not detection.
Most searches for this phrase are really about number two. The remote station is there because NFPA 72 and most AHJs want a way to see and test a concealed detector without a ladder and a prayer. Exact rules vary, so check your AHJ for what they expect on yours.
What the remote station actually does
Generic version. The station has an LED that follows the detector's alarm state, so you can see from the floor that the unit in the duct tripped. And it has a switch that makes the detector behave as if it sensed smoke.
How it connects is the part to be careful with. The station lands on the detector's remote indicator and test connections, usually with a handful of conductors (the count depends on the model). The LED side is driven by the detector. The test side closes a path the detector reads as a test request. That's the whole idea.
I'm not printing terminal numbers here on purpose. They change by manufacturer, by model, sometimes by revision of the same model, and a wrong terminal on a detector is a bad afternoon. Pull the installation sheet for your exact detector and your exact station. They are often sold as a matched pair for a reason. For the general wiring picture, the duct detector wiring diagram article covers how the pieces relate.
Running the test
Here's how I do a functional test, in order.
Tell whoever owns the building what is about to happen. Fans will shut down. Smoke dampers may close. Somebody will notice.
Put the panel in test or disable the dialer and any releasing outputs first. Then, at the station, turn the keyswitch to test (or hold the magnet to the detector). The panel should show an alarm or a supervisory condition, depending on how it was programmed, and the LED on the station should light.
Then the part people skip. Walk to the air handler and look at it. Did the fan actually stop? Did the damper actually close? A panel that says "duct alarm" proves the detector talked. It does not prove anything happened to the air.
Reset the station, reset the panel, confirm the LED goes out. Put everything back in service.
What I write down
Here is a log from one anonymized job, a 6 unit rooftop retrofit. Times are from keyswitch to fan stopped, measured with a phone stopwatch, so treat them as rough.
| Detector | Location | Test method | Panel response | Fan stop time | Result |
|---|---|---|---|---|---|
| DD-1 | RTU-1 supply | Remote keyswitch | Supervisory | 4 sec | Pass |
| DD-2 | RTU-2 supply | Remote keyswitch | Supervisory | 5 sec | Pass |
| DD-3 | RTU-3 return | Magnet | Supervisory | 4 sec | Pass |
| DD-4 | RTU-4 supply | Remote keyswitch | Supervisory | no stop | Fail |
| DD-5 | RTU-5 supply | Remote keyswitch | Supervisory | 6 sec | Pass |
| DD-6 | RTU-6 return | Magnet | Supervisory | 5 sec | Pass |
DD-4 was the useful one. The panel was happy, the station LED lit, and the fan kept spinning. The relay was wired but the fan control circuit on the mechanical side had never been landed. Without looking at the fan, that one would have gone in the report as a pass.
Test switch versus a real sensitivity test
Worth saying plainly, because it trips people up. The keyswitch or magnet is a functional check of the detector and the circuit it's on. It is not a sensitivity test. NFPA 72 has its own requirements for sensitivity and for how often, and the method has to be one the manufacturer lists for that detector. Some say canned aerosol is fine and some say it will foul the sensor, so read the sheet before you spray anything into a housing.
Airflow matters too. A duct detector sampling tube in a duct with the wrong air velocity or the tube pointing the wrong way is a detector that looks fine and sees nothing. Check the tube, check the orientation, and look at the housing's own airflow or velocity requirement for your model. If you need the design side, duct smoke detector requirements goes through where they're required in the first place.
When it doesn't test
A few things I've actually found.
- The station was wired to the wrong detector. Labels matter. Two RTUs side by side, plates swapped.
- The keyswitch was left in the test position by the last tech. Panel shows a trouble or a stuck alarm and everyone assumes the detector is bad.
- Station cable run too long for what the sheet allows. Check the limit for your pair.
- Magnet wasn't strong enough or was held at the wrong spot. Use the magnet that comes with that brand and put it where the housing marks it.
Anyway, the retyping and cross checking of all those device IDs into a test report by hand is the part that always cost me the evening, so full disclosure, I built FireDeck to take some of that off my plate.
Look at the fan. That's the whole test.