Question:
Was TIG welding a aluminum tank this weekend and I triggered the smoke detector in the back part of my house. However, the closest two smoke detectors didn't even chirp.

All three are the same make and model and purchase date. I tried using a butane powered torch placed near each detector to get them to go off but wasn't successful.

I don't smoke so I can't blow smoke into them. Ideas?

These units are less than 3 years old and batteries are probably 1 year old. I'd hate to think their crap already.

Answer: Yep, that's your key right there - Smoke detectors are just that: detectors of *SMOKE*, not heat, not open flame, but the particulate mess that a fire generates. A clean burning fuel such as butane produces (at least in terms of stuff a smoke detector can "see") next best thing to nothing. A match will probably also be insufficient. Smoldering paper/rags/grass or similar material should work wonderfully, as will the "canned smoke" that can be bought from a beekeeper's supply store, or an ultra-fine mist such as you can get out of a spray bottle. Aerosol cans are likely a bad thing for the health of the detector - They'll proabably trigger it just fine, but the stuff they're spraying can cause clogging of the sensor chamber, or contaminate the sensors inside it in such way as to render them "blind".

Basically, to make a smoke detector, take a chamber, and pass a beam through it. Place sensors that can detect the beam in such a way that their "line of sight" is at a 90 degree angle to the beam source (They can't "see" the beam source) Now add some circuitry to make a loud obnoxious noise when one or more of the sensors starts seeing the beam. Presto... One smoke detector.

As long as the beam isn't scattered by something, no alert. As soon as enough "something" enters the chamber to scatter enough of the beam to activate the sensors

In smoke detectors, the beam source is usually a dot of Americium (I think I blew the spelling on that), a low-level radioactive source, mounted behind a pinhole in a shield plate, and in turn mounted in a typically circular chamber. One or more sensors that amount to geiger counters are placed around the circumference of the chamber, and the chamber is allowed to "breathe". As long as the radiation beam (I forget which flavor of radioactive particles are involved - probably alpha, if I'm remembering right - the ones that can be shielded against with a sheet of paper) remains unscattered, the "geiger counters" in the sides of the chamber see no radiation, and the alarm remains silent. Once enough particulates enter the chamber to scatter the beam enough that it becomes detectable by the "geiger counters" (which are, if you recall, situated such that they're at right angles to the undeflected beam), the alarm goes off.