Retrofitting Smoke Alarms With Bluetooth

Everybody should have a few smoke alarms in their house, and everyone should go check the battery in their smoke alarm right now. That said, there are a few downsides to the traditional smoke alarm. They only work where you can hear them, and this problem has been solved over and over again by security companies and Internet of Things things.

Instead of investing in smart smoke alarms, [Johan] decided to build his own IoT smoke alarm. It’s dead simple, costs less than whatever wonder gizmo you can buy at a home improvement store, and reuses your old smoke alarm. In short, it’s everything you need to build an Internet-connected smoke alarm.

Smoke alarms, or at least ionization-based alarms with a tiny amount of radioactive americium, are very simple devices. Inside the alarm, there’s a metal can – an ionization chamber – with two metal plates. When smoke enters this chamber, a few transistors sound the alarm. If you’ve ever taken one apart, you can probably rebuild the circuit from memory.

Because these alarms are so simple, it’s possible to hack in some extra electronics into a design that hasn’t changed in fifty years. For [Johan]’s project, he’s doing just that, tapping into one of the leads on the ionization chamber, measuring the current through the buzzer, and adding a microcontroller with Bluetooth connectivity.

For the microcontroller and wireless solution, [Johan] has settled on TI’s CC2650 LaunchPad. It’s low power, relatively cheap, allows for over the air updates, and has a 12-bit ADC. Once this tiny module is complete, it can be deadbugged into a smoke alarm with relative ease. Any old phone can be used as a bridge between the alarm network and the Internet.

The idea of connecting a smoke alarm to the Internet is nothing new. Security companies have been doing this for years, and there are dozens of these devices available at Lowes or Home Depot. The idea of retrofitting smarts into a smoke alarm is new to us, and makes a lot of sense: smoke detectors are reliable, cheap, and simple. Why not reuse what’s easy and build out from there?

46 thoughts on “Retrofitting Smoke Alarms With Bluetooth

  1. PSA: That looks like an ionization smoke detector. They usually have a ten-year lifespan, so if this one was originally installed in 2001, it is probably now passed its life span.

    1. Came here to make a comment regarding the lifespan of the smoke detector elements, glad someone else beat me to it. I don’t know of any smoke detector that doesn’t have a life span due to the sensing element. Old one’s would be good to experiment for this sort of thing on but I would never re-use one past it’s expiration date

      1. My understanding of the test feature is that it doesn’t (shouldn’t?) just activate the sounder but actually interrupt part of the detector path in the same way smoke particles would.
        If you don’t trust the test button then a can of special test smoke is the only alternative, burning a match or bits of paper will produce soot that will coat the inside of the detector and stop it from functioning properly.

    2. The Americium source has a half life of over 400yr. What should go bad after 10yr except the battery, which can be changed. There are so many things with expiration dates nowadays where it is really ridiculous, even table salt, which lay down for more than millions of years.
      Of course regular testing of the smoke alarm is important but the rest is just to generate revenue and increase landfills.

      1. It’s not the salt or vinegar or whatever themselves that in fact didn’t expire, but is the packaging what causes the trouble which degrades and releases harmful decomposition products.

        If you keep them in sealed full glass or ceramic recipients will last forever, as the 2000 years old amphoras filled with Hispanian olive oil found in a roman sunk shipwrecks near the coast of Italy.

      2. I guess the thing is, that it’s a little less active, so the alarm isn’t as sensitive… buuut, that could be fixed with recalibration… in theory…. maybe as we get into the era of sub-dollar a pop microcontrollers they will self recalibrate… and then you only have to worry about whether capacitor life makes it worth the bother.

  2. There is another way to do this. Make an ESP8266 PCB that slips into a USB wall wart for power. Have the ESP8266 listen for the siren. When it hears the siren send the message onto the Internet. Now you don’t void the UL safety listing of the smoke alarm. Plus it doesn’t use any batteries.

    1. I thought of this years ago, back when I was designing an “Everything Is OK Alarm”, long before the ESP8266, but never got around to doing it. Tampering with a smoke alarm is something I don’t want to do, even if I have the technical knowledge to do it. So I figured it would be better to make a “Smoke Detector Sound Detector.” It really shouldn’t be that hard since the sound is very well defined and loud enough to overcome anything other audio that might be happening. A microphone and standard amplification circuitry, microcontroller, and some frequency detecting software, and you have a way to IOT your smoke detector without any risk of screwing it up, or incurring the wrath of insurance companies, apartment building management, etc…

    2. Or you can buy a smoke detector with an interconnect wire. Very, very common, if not the most common type. Any “wired-in” detector will have this. The intention is all interconnect wires are connected together in the house. When one detector is triggered, it pulls the interconnect high (9v) and all the alarms in the house go off. These detectors are designed this way and no modification is necessary.

      1. You can also buy smoke detectors that can be synced up over wireless. When one is triggered, it sends a wireless signal to the others and they go off aswell.

        But: i wouldn’t use either system. Synced/connected alarms suck. We had a bunch of the wireless ones in our hackerspace. They’re only good if you want people to scatter from the building ASAP. But if you want to figure out where the potential “fire” (or someone who can’t cook with olive oil, or someone testing his fog generator) is and whether you can “extinguish” it with a fire blanket, a fire extinguisher or just an open window, synced detectors are really bad. All of them go off, so to figure out “patient zero” you have to walk around 400 square meters of hackerspace instead of just going where the song of detector people comes from.

        1. If your area is large enough you shouldn’t be relying on interconnected but independent alarms you should have a system of detectors, sounders and call points with a separate control panel that tells you which detectors have activated and potentially if any are faulty.
          Saying that, most interconnected alarms have a feature where you can add a switch to the interconnect wire that will temporarily silence any detectors that are just acting as sounders. Even without this feature connected alarms will give you time to act before fire spreads and traps you inside so I think its pretty stupid to say they are a bad idea just because you find them occasionally inconvenient.

    3. Hi, I am interested in your idea here. I have no Knowledge of the board you are talking about. How would you teach the board to listen fr the Alarm sound and would it be possible to have a “teach” function for when a different alarm is installed. I want to turn an electric stove off when the smoke alarm sounds. So will communicate with another device to do that.

      Thanks for any advice you can give me.

  3. This is a safety device. If you screw up and it fails to function during a fire, you are potentially dead. Unless you are doing something like opto-isolation and a separate power supply, I wouldn’t even think of it. Imagine doing an OTA update, switching the input pin to an output by accident, and pulling it to the wrong rail. Don’t be dumb when people’s lives are on the line.

    1. You could always use this in tandem with your regular unmodified smoke detector. You’d still have the same protection you would otherwise, with the benefit of your IOT smoke alarm. Nothing says you can’t have two smoke detectors!

      1. Exactly this. If you’re worried, add another. They’re so cheap as to be disposable. I just added sprinklers in my garage and basement and a dry chemical suppression system (Haven) over my kitchen stove. I (gasp!) added a wire to the junction boxes of my wired-in smoke detectors so I could connect to the interconnect wire. When the smoke detectors go off, that wire goes to 9v, and my house will send an announcement to my phone. I’m okay with this arrangement, but I’m sure someone will berate me for doing this.

    2. I would agree with this 100%. I understand the value and interest in taking apart and hacking all sorts of things, but I would much rather buy a commercial IOT smoke detector. When I look at all the mistakes I make in creating a new thing and the bugs I find out days or weeks after the most recent code change, it’s not a big deal. With safety devices like fire alarms, it is a big deal.

    3. Use a series resistor, a buffer or just a transistor with 2-3 resistors to decouple the signals, if you have the fear of accidentally switching pin to output. If you have to do OTA updates. Its not like your phone, where the provider decides to do OTA updates.

  4. “Everybody should have a few smoke alarms in their house, and everyone should go check the battery in their smoke alarm right now. That said, there are a few downsides to the traditional smoke alarm. They only work where you can hear them, and this problem has been solved over and over again by security companies and Internet of Things things.”

    Know who else solved this problem? Everyone who makes smoke detectors as the networked feature you have described is required for apartment buildings.

    A smoke detector that goes off in the basement is required to trigger all of them.

    Buy canned smoke if you really want to test the entire signal chain.

  5. I hooked an electric imp to a smoke detector, and was showing it off at work. I forgot that I had broken the web page when I was messing with it over the weekend, but found it was still a closed loop system when my wife called and told me to knock it off.

  6. Theoretically, you should be able to run a few loops of insulated wire around the battery terminal and detect the current surge when it starts going off, then no physical connection.

    1. that is actually a smart idea, You would obviously have to make sure to differentiate between low battery beep and the alarm actually sounding, but that wouldn’t be too hard as low battery just pulses every few minutes.

    2. Please rethink this. If you warp the detection coil around the conductor, the directions of the current are 90° of and it will not work. You could put the battery wire through a ferrite/laminated core which carries a sense coil. But thats way to complicated and uses probably too much power for things like amplifiers and signal analysis.

      1. Ah whoops, was having some transductance problems with high voltage cables that were actually internally coiled the other year and it’s broke my thinker.

        However, snap on ferrites can be got.

  7. im glad someone is modding and learning,
    but i have issues with this.

    1) most of these things are sensitive, it does not matter how hard you try to UN-technify the name; bluetooth is a RADIO TRANSMITTER and may interfere with a smoke alarm, unless its isolated AND SEVERAL INCHES AWAY! remember that the user’s manual might mention something about ACCEPTING INTERFERENCE INCLUDING INTERFERENCE CAUSING UNDESIRED OPERATION. aside from the deadly possibility of not being set off (unlikely) your probably going to get constant nuisance trips, resulting in a smashed detector. a smashed detector is useless lol

    2) this hack should only be done on SINGLE-STATION alarms, but due to the new laws, single-station alarms are limited to extra alarms, this reduces opportunities for hacking, as new houses have MANDATORY inter-connected alarms with mandatory dual-power (AC with batt backup). i believe for reliability, weight, cost, ect they are powered with capacitive droppers, this would make the ENTIRE circuit a schock hazard INCLUDING THE INTERCONNECT-WIRE, just worth noting the voltage in case someone tries to update it while running straight from unisolated 120/230vAC.

    3) due to above, signal/interconnect wire may present a schock hazard, i do not know if this is low-voltage, but it may actually be low-voltage relative to the ASSUMED potential of the other unit’s also-HOT-chassis. so you may measure 5v to detector chassis and again 5v to chassis of other unit (when wired in-phase) but may register a high-voltage relative to REAL-GROUND andor neutral.

    4) all of the units i have seen use three conductors; Hot, Neutral, and Signal, but no ground reference, if mains power are split and a load switches on, the voltage drop on the neutral may go past safe limits for both arduino and george, swhy we have seperate ground on most stuff.

    5) some places have laws about modified andor unapprouved fire/smoke alarms, those laws only apply when it is the primary unit for each mandatory area. if it is an extra unit you can do as you please, just dont modify the primary multi-station dual-power units and you should be fine.

    6) a smoke/fire alarm at less than arm’s length miiight damage your hearing if for prolonged periods… debug at your own risk (or disable speaker)

    7) there IS a HUGE difference between a smoke alarm and a fire alarm unless it is marked as having both opto(visible-particles) and ion(invisible-substances). different places and settings may mandate differing detectors, check your local regulations when in doubt.

    1. and whatever you do, do NOT scrape, drill, sand, heat, bend, impact, or otheriwse damage your radioactive sample, it is illegal to do it on purpose and ingestion of any amount of this dust particles may cause cancer or other ills.

    2. On point #7, I think you know what you’re talking about, but didn’t delineate it very well in the explanation presented here:

      SMOKE detectors generally fall into one of two categories: ionization technology and optical technology (note, I have encountered several other types, but they are quite unusual and out of the scope of this post). In ionization technology, as has been explained, a small amount of radioactive material is used to generate an ionized atmosphere between two plates, the more ion pairs generated, the great the current. So long as a current above a certain threshold is maintained, the alarm does not sound. Smoke particles, and they do have to fall between two extrema in sizes, but regardless, smoke particles which are visible can fall in this range. Same with optical technology, so long as a beam path remains unobstructed by smoke particles (“visible” or “invisible” — you never stated what the size limit was for what constitutes visible/invisible), so long as the beam path remains a certain % unobstructed, then sufficient light can reach the receiver and the alarm is not tripped. As soon as a sufficient amount of smoke particles* enters the ionization chamber, then the number of ion pairs produced drops to below a predetermined threshold and the current moving between the two plates drops commensurately, resulting in an alarm condition. Likewise, when a sufficient amount of smoke particles enters the beam path in an optical detector, the amount of light reaching the receiver unimpeded drops, and the detector enters an alarm condition.

      Where you erred was a comparison either between the two types of SMOKE detection technologies, already discussed in my previous paragraph, or between the components of a fire that are the focus of the detector device: SMOKE or HEAT. To that end, a detector can have sensors to detect by either optical or ionization techniques, the amount of smoke present and/or it may have sensors to detect the level of heat present.

      HEAT detectors are most often set to a static point, above which a sensor is designed to trip the alarm. Say, a temperature of 140 degrees F for 5 seconds. Another type that is reasonably common, but usually confined to more commercial settings or settings where hazardous materials that a fire situation would complicate enormously, is the rate of rise. A sensor actively monitors the environment for a temperature change that occurs in just a few seconds, for instance, smouldering rags that suddenly become engulfed in flames would drive that location from a slow heat increase, maybe 1 degree F per minute whilst smouldering, to a rate of rise of 25 degrees F in a little under a second or two. Either way, the alarms have a type of fire that is better detected than the other. And within each category, SMOKE or HEAT, they have a further specialization.

      (*note, not gases, but particulate smoke components like minuscule ash, uncombusted fuel material, etc.)

      Just clarifying…


  8. I’ve also been planning something like this for a while.
    We have a large house (many detectors) with high ceilings (PITA to replace the batteries and mute the detectors when my neighbor lights his log-burner).
    My plan is to hard-wire all the detectors to a UPS-based power supply. This relieves the battery-replacement problem and would then leave the 9v battery space in each unit free to hold an ESP8266-based add-on.
    This could allow for remote alarming and remote muting as well as other NEST-alike features (environmental sensing, etc.).

    I have some of the same concerns expressed in other posts though; modifying smoke alarms may not be legal in NZ – especially the power supply…

    Like the idea of modding a ‘secondary’ unit as a POC though.

    1. From my experience living in an apartment/condo, our smoke alarms were all wired to a circuit in the ceiling, and only used the 9V battery as backup when the power was out. They were not part of a central fire system like the multistory apartment buildings, but they pulled power from the power when they could.
      As for muting the detectors at will, not sure they make them with that feature out of the box. Good luck :)

  9. The comments on hackaday are rather hilarious.

    If it’s not a hack – just listen to screams of sacralige.

    And if it IS a hack the Occupational health and safety committee get roused from their slumber.

    It’s a neat project and great way to experiment with ble and writing firwate and producing your own App.

    And if half the OHS fanatics actually looked at his project page you’d see most of your concerns are answered.

    Btw has anyone tried to ring his mother/neighbor/friend :o

  10. Is Bluetooth really a good solution for this?
    All commercial rf smoke detectors I know of use a classical 2 way rf link, either relying on a standard like zigbee, homematic, Z-Wave, or a proprietary protocol, and many of them can be easily integrated into your home automation system.
    If it has to be DIY, I’d most likely use something like MySensors.

  11. This is a brilliant idea. I am deaf and actually have a specially suited smoke alarm that will alert me on my pager via what I think is RF 868 MHz frequency made by “Bellman and Symfon”. Whilst it works, its brilliant, but expensive and my pager doesn’t appear to work greatly anymore. Would love to be able to make some sort of device to pick up those signals and relay them to my phone, PC or whatever. Btw, the same system also alerts me when doorbell goes off, baby monitors sets off and telephone rings via a similar device as the smoke alarm does.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.