Roll-your-own Z-Wave (or Zigbee) Smoke Detector

If you’re reading this thread, you probably already know that there are limited choices with regard to z-wave smoke detectors. Fibaro, First Alert and the other smoke detectosr offered are battery operated, not mains-powered (hard-wired). Without hard wiring, you don’t have the dead-simple (reliable) messenger wire to create a network of interconnected smoke detectors. In many jurisdictions, building code requires detectors be both hard wired and interconnected.

So I want BOTH, but I also want Z-Wave (or Zigbee). Apart from the obvious things like OH integration (and monitoring the detector’s backup-battery charge level (to avoid a chirping detector at 3:00AM) I want to TEST the smoke detector with Z-Wave (OH). Best practice is to test your detectors monthly. Does anyone actually do that? Of course not. Why? Because it’s too much of a PITA to get a ladder/chair press the button (and damage your hearing 12" away from the horn).

An OH/Z-Wave triggered test would allow convenient testing (enough to ACTUALLY do it). Furthermore, if detectors are interconnected, full testing wouldn’t require multiple people. A single person could test the ‘downstairs’ detector from the upstairs (and vise versa), ensuring that the interconnected-ness causes the others’ alarm to go off.

Yeah, but REALLY, how important is testing? I’m no expert, but I always thought the need was exaggerated. Recently I tested mine and… Nothing. it didn’t go off. I pressed the test button again… Nothing. I held up a smoldering match… Nothing. Then… I rebooted the detector, and… it works fine again. Who knew that smoke alarms could need to be …rebooted?! How scary is that? So… now I test.

For the holy grail of smoke detectors, I had the idea of rolling-your-own by way of embedding a Fibaro Smart Implant into a detector. Has anyone seen/tried this? The smart implant could find power from the detector’s own power supply (SMPS from mains), and the detector it could use one of its relays to momentarily close the test contacts (for remote testing via Z-Wave/OH). The Smart Implant’s sensor leads could be attached to the horn terminals, or LED lights, or even the messenger wire (9v DC to ground IIRC) to detect an active alarm by detecting the presence of voltage.

This roll-your-own idea seems like it might be worthwhile. EVEN IF some company were to come up with a mains-powered Z-Wave detector, it likely wouldn’t be quite the right product for all cases. For example, in a kitchen you want a Photoelectric detector (less sensitive than ionic detection but less prone to false alarms). Near a bedroom you want the high sensitivity afforded by a DUAL unit with BOTH Photoelectric and ionic detection (i.e. false alarms are less likely near a bedroom so maximum sensitivity is desirable). In a basement on the other hand you may want a COMBO unit that detects both Carbon Monoxide and smoke, or a CO-only unit located down low.

I’m hoping someone has tried this, and can share. Otherwise, indicate that you’re interested and I may build one and post about it.

We’re entrusting our children’s lives to these things, after all.

While “something is better than nothing”, there is a danger of being satisfied with an ineffective “something” instead of investing a bit more effort/money into something trustworthy.
You’ve clearly thought all this stuff through already, I’m not sure everybody does.

The idea of “adding value” to a standalone system seems like the way to go.
The zwave part is just adding convenience, the life-saving part should be independent of that.

I think this is how manufacturers get away with even selling zwave detectors - it’s viewed as a standalone detector/alarm, but with extra non-safety related features.

There is the matter of certification - if your insurers or building codes require certified fire systems, opening them up and soldering in some alien kit will probably invalidate the certification.
If we’re talking a wired network, it would be smarter to “eavesdrop” on that network rather than modify sensors. Can’t call it ‘non-intrusive’ if we’re looking to inject test requests, but at east we’re in the part of the system intended to be exposed to the rest of the world.

you can get Relay moduals for Kiddie brand smoke detectors that comunicate with the messanger wire
either to read the state and trigger something else or to trigger them if something else goes off
we use them in new town homes to trigger the smokes if the sprinkler flow switch is triggered or to trigger a strobe at the front porch

you could combine this with some relays and get close to what you want

the 9v wire can be used to trigger them

This would be my concern as well. Any hint of a homeowner doing something that prevented a smoke detector from working properly would be enough to invalidate an insurance claim. That’s the last thing you want in the event of a fire.

@Broncosis’s suggestion of the relay module seems like a good way to avoid that, since it doesn’t mess with the detectors directly. I think that would be the way to go, but I assume it won’t help when it comes to monthly tests.

I also recall hearing about a device that you put next to a smoke detector, and it triggers whenever the siren sounds. I suppose that could even be done with a Raspberry Pi and a microphone.

For what it’s worth, I have a wired Nest Protect.

  • It has both photoelectric and ionic smoke sensors, plus a CO sensor
  • It runs its own monthly checks and notifies me in the app and by email
  • When it detects smoke, it gives me a minute to cancel the alarm before blasting the siren
  • It has a great pathlight feature that softly lights my hallway at night when it detects movement
  • It even looks nice on my ceiling

The only downside is that it doesn’t work with openHAB. That doesn’t bother me though, because as a life-safety device it does everything I need on its own and with no fuss.

I don’t mean for this to be a Nest Protect commercial (oops, too late), but it just works well and doesn’t require me to think about it too much. It’s one of the most pain-free smart devices I’ve purchased, likely because Google has left it alone. It’s probably for the best that they haven’t come up with ideas to “improve” it that would basically turn it into another beta product.

They make Zwave and Zigbee versions as well. A number of camera vendors also support loud noise and siren detection (Wyze, Nest I think, Arlo, etc.).

I opted for wired Nest Protects as well (I even had to run a wire from an outlet to one of them). For me the deciding factor was the chimes and other noises it makes. The usual chirp sound smoke detectors make would terrify one of my dogs. We’d find her hiding in the closet hours after one of them would chirp because the backup battery needs replacement or it went off because of cooking. The NP’s replace the chirps with a chime which the dog didn’t mind (we lost her to cancer in July unfortunately). Even the self tests it runs use a chime instead of a chirp.

I also like how instead of just going off when I’m burning something on the stove is gives a nice friendly warning “Smoke detected in the hallway. The alarm is going to go off. It’s going to be really loud.” My Nest account is so hopelessly hosed right now though that I can’t even get these to appear in Google Home. But they still work with the Nest app and that’s good enough for me.

Over all they are a well designed product and I’m happy with them too.

1 Like

I wouldn’t recommend doing with something other than the certified relay pack if you want to be able to trigger them if you used a microphone to monitor them that is probably fine but as mentioned a non functional smoke detector could have serious implications on a insurance claim. but with a relay and a single input you could easy get most of what you want from a typical north American install I can’t speak for anywhere else

somehting like a shelly could probably do this a you would have to modify a sonoff so that probably not the best choice but there has to be some thing

1 Like

You’re remembering right. I’ve seen this one: The EcoLink Z_Wave Listening Alarm.

@rpwong and @rlkoshak make a pretty compelling case based on the feature set of Nest. Cloud-connected nature aside, I actually would be far more critical of its LACK of a wire for messaging/interconnect. Google seems to call this a feature. They say, “A message wire can’t communicate the nature of the emergency”.

My feeling is that nature of any emergency is secondary if you don’t know there’s an an emergency occurring. Am I right that the ‘3rd Floor’ detector wouldn’t alarm based on the ‘basement’ detector going off if your WiFi’s not working?

No doubt the nest app/cloud validates the detectors’ connectivity 24/7, BUT (bear with me here) history shows that the worst disasters usually occur as a result of multiple simultaneous failures. For example, a house fire is very bad, but a disaster is when the house burns AND the alarms didn’t warn the occupants.

Where humans ALWAYS go wrong evaluating risk is that they don’t consider that the multiple failures are usually positively correlated to the same causal event. In this example, it means if you have a house fire, that’s the precise moment when your Wi-Fi is LESS likely than usual to be working properly.

Here’s a failure scenario I’ve actually seen in real life: An open neutral wire of a typical residential/commercial split phase (or 3-phase) power service: If a tree branch, or ice disconnects the neutral wire only, and if the poles are unevenly loaded at the time of the open neutral (almost always), it can cause fires by way of a non-transient (i.e. minutes-long) voltage SURGE on one pole, and a simultaneous voltage SAG on the opposing pole. In this scenario fires can start when devices are fed an over-voltage (up to 2x nominal). Many modern electronics are friendly enough to NOT catch fire these days, usually by way of a simple MOV/Varistor PLUS a fusible component (sacrificing the device/components), However not all mains-attached devices fail safely. Meanwhile electronics (Wi-Fi) crash on the sagging pole (or vise versa).

I don’t mean to go off on a tangent, but I am of the same mind as @rossko57 favoring the simple and effective for safety, and complex and fancy for situational awareness.

I also agree with @Broncosis view that the ideal way for to integrate alarms for situational awareness for a messenger-wire type system is to use the certified relay pack (such as the First Alert RM4). However that still leaves the ability to press the test button remotely (to test/hush)

I agree with everyone who voiced the concern that an insurance company might just LOVE to deny a claim as a result of ‘tampering’ with a safety device. The irony is of course that a the “tampering” in this example would be totally benign since it’d be a method for remotely pressing the test button, i.e. with a relay. Still, it’s a valid concern. I suspect that even if you taped a servo-operated robo-finger-presser-5-Million to the detector, it’d look to a claims adjuster like tampering.

Lack of a wire doesn’t mean they don’t communicate with each other. They communicate wirelessly directly to each other (not through the cloud) using Weave.

No, the basement detector will communicate with all the other NPs in the house wirelessly using Weave 802.15.4 riding on WiFi 802.11 b/g/n. And like Google says, it doesn’t just transmit “alarm” but the exact nature of the alarm or other information too. For example, if the NP near the kitchen is about to go off because I’ve burned the bacon again, if I press the button the other two NPs in the house will announce that the one on the main floor has been silenced.

NP to NP communication doesn’t go through the cloud. And the devices have battery backup so even in the event of an Internet and power outage, the alarms will still work and still communicate with each other, for at least as long as the battery lasts which should be several days. Unless there is some reason why the Weave signal is being actively jammed these should be every bit as reliable as a wire. And if it were not as reliable as a wire, the trade off seems to favor the NP in terms of features.

Which is why the NPs don’t rely on your home network WiFi to communicate. They rely on Weave which is router/gateway less so there is no third device that as to be on and working for the NPs to communicate. They can discover and message each other directly, as long as they have power, which is covered by the large battery they have on board in the case of a power loss.

I do risk analysis for a living. I did my homework before choosing NPs. Are they perfect? No. But they are very well designed and they clearly consulted experts in fire as well as fault tolerant and reliability engineers.

Everyone else should do their own homework too, but part of that is to actually read the docs instead of making assumptions.


I’m guilty of this, and you’re exactly correct here.

Today I read that Underwriter’s Labs new “217 Standard 8th edition” was supposed to go into effect in 2021, but has been delayed because of Covid until June 2022 (for now). I believe that detectors sold after that date must meet new standards to be listed. A few companies appear to have compliant products, Nest is among them. Also a company called USI Electric (Universal Security Instruments) in their detectors using “Sensing Plus” branding.

I’m glad I went down this smoke detector rabbit hole, because it brought the new UL standard to my attention. It seems like a meaningful improvement. In essence: Earlier warnings, fewer nuisance alarms, more resistance to anomalous mains power events. Apparently “escape time” has decreased in the last 40 years from 17 minutes to just 3, largely due to changes in finishes, materials, how they burn, and toxic products of combustion.

Reading between the lines, I see something like: Increased sensitivity is good, but can be counter-productive because nuisance alarms make the humans disconnect smoke detectors. Nuisance alarms can be reduced with the use of microelectronics (ASICS/DSPs etc.). Microelectronics require better power supplies to be stable/effective.

Getting back to Nest: I’m wondering if @rlkoshak or anyone looked into Nest’s OpenWeave to see whether it’s feasible to do any local messaging over WiFi to nest devices?

My understanding is that Weave runs on top of Thread which itself is kind of a modified Zigbee protocol so you would need both the software stack but also hardware that can talk Weave. I never have looked to see if someone has hacked a Zigbee coordinator’s firmware to talk Weave or not but I expect that would be the first step.

In the long run I fully expect that Weave will be replaced with Matter which, like Weave, also rides on Thread (but also supports WiFi and Zigbee). But who knows when that will happen.

This topic was automatically closed 41 days after the last reply. New replies are no longer allowed.