Hello Community
I am searching for an smart Smooke/Fire-Detection-Alarm-solution.
My Snapmaker is placed in a small room in the basement. It is video surveillanced and switchable via Homekit (hombridge+Sonoff).
For my peace of mind i like to have a smart Smoke/Fire-Detector which will switch off the power of the Snapmaker in case of an alarm and send me an Message on my Phone or any other alarm upstairs.
Because of an existing Homeautomation system based on Homekit, I am very pleased to have an Homekit/Homebridge solution.
so at my actual point of Planing i like to install a normal Smoke detector in the room(already installed) and a google Nest protect inside the Enclosure of the Snapmaker.
Does anyone has experience with such a setting?
How does the Nest Protect work in such a hot Environment of the Enclosure?
Has anyone experience with Nest Protect and homekit?
Does anyone has another reliable solution for my needs?
the Fire Avert isnât wat i was looking for! their are a few Smokedetector with Relais available on the (german) market. even the great companies have some Solution (eg. Gira)
But the BlazeCut is a real good hint! Thanks for that!
Iâd be concerned about the laser setting off a smoke detector during normal operation. I suppose part of the benefit of a smart system is that you could deactivate it while the laser is in use, though.
In a friendâs house I installed a zwave switch up to a smoke detector relay wired into their hardwired home smoke alarms. Zwave goes through their HA system and sends them a text.
You might consider rolling your own with a smoke detector and z wave switch.
That could be wired to any off the shelf smoke detector relay output. Typical wiring, replace the switch with a 120V smoke detector with relay output, ignore load
i didnât specified my need enough.
so i try to write it down:
Smoke detection (Smoke senor)
Fire detection (Temperature sensor)
possibilty to switch the detection off, because of Laser and CNC.
Alarm should be delivered to devices via LAN/Wifi (no other hardwire connection from basement to Appartment possible! I do have a PowerLan-Connection+Access point from the basement to my router in my appartment.)
Homekit compatible (via Homebridge is ok), Than i can cut off the Power very easily because the Powerplug to the Snapmaker is already Homekit compatible
maximum price 150âŹ
i planed to install the Smoke detector within the Enclosure. Because i do work with Wood inside my Workspace and dust will not be avoidable. But maybe this could be handled via the âswitch off featureâ.
the BlazeCut would be a solution on top of an allert, because this is only working when âdie kacke am dampfen istâ
Solution which are not suitable for me:
Smoke detector with relais â No Allert at my Appartment when Fire/Smoke is detected
Arduino/ESP-Projects â i have designed a smoke detector running on an ESP+MQ-2, but i could not program it to run reliable. I had too much false alarms and when i simulated Smoke the smoke detector did not work.
Has anyone running an Nest protection within the enclosure?
@3DSW In the end, I donât believe there is a system that meets all your requirements at this point in time, mainly because of cost. Have you looked into a zwave wireless system? You mentioned HomeKit, do you already have a zwave hub? A zwave wireless system will meet all your requirements except for price⊠unless you already have a zwave hub.
Sorry I think I didnât got the Idea behind it. Read it again and now i have an imagination which can be adopted to my needs! Now i get the use of the FireAvert.
Instead of Z-Wave-Relais i can use a Shelly 1. I already have several of the Shelly working in my Homebridge.
I could use any Detector with an Relais and let the Relais work as a Switch at the Shelly.
I can program the Shelly to execute an Url when the Button is pushed (=the detector detects smoke) or monitor the status of the âswitchâ via Homebridge.
The Button in the Shell 1 can be configured as âActivation Switch - use it for motion sensor. Any input turns âONâ and resets Auto OFF timer.â with auto OFF = off.
I will think about that solution, sounds like a real good one. And much cheaper than the google nest protect.
You could just use a sonoff with a temp sensor (th10, th16 I think) so if the temperature goes over a set limit like 100 degrees C, you then switch off the snapmaker.
This thingiverse item also has a lot of useful info:
As for this one:
Invert the logic: Give it some kind of heartbeat. If the the heartbeat stops: have an alarm.
Then again, with the price point youâre aiming for, the only reliable solution I can think of is to just sit next to it and watch it.
One completely alternative route that you can take, but requires specific an more in depth knowledge:
build something similar to the spaghetti detective. Have a camera stream everything to your apartment. Either keep watching it, or build an visual recognition system that detects any anomalies.
This seems to me to be an easier problem that it might seem at first glance, with a caveat. The approaches Iâve seen are all about recognizing some particular failure condition. It would seem far easier to feed a 3-D model, to make the assumption that itâs being printed each layer all at once (usually true), and then to recognize anything that doesnât look like the model being built up a layer at a time. Trying to recognize spaghetti is training for a particular failure, but there are lots of ways of failing. Training for particular failures leads to false negatives in failure detection. Far better to look for the one good way to print successfully, because at worst you might get a false positive.
The caveat, of course, is that you need to feed your recognizer a 3-D model, and it has to be the same as the object being printed, etc. Probably the best way is to integrate the recognizer with a print simulator, so that the printer and recognizer are always working off the same G-code file.
If you have experience with building these kind of things, it might be a fun project. If you donât have any software development background, it might be a very challenging project
My personal approach to this would be to basically ignore the center of the buildplate anyway. You donât care about the model at all. What you do care about are changes in other parts of the videofeed. Like smoke and fire. Fire will start in all kinds of places, but not necessarily the model itself.
Choose a good angle of for the video feed that catches the entire machine (preferably the power supply as well).
Have some prints done and use that video as input for the ânormal situationâ and then create a model that detects anomalies. For example, part of the image is obscured by smoke. Next, generate an alarm.
You might even get away with just some plain statistics on the image data to create something in this case.
So weâre both saying more or less the same thing. Donât try to detect fire or smoke or whatever, but focus on detecting things that are considered out of the ordinary in general. (Although there might be pre-trained models for fire & smoke out there or other things that can be used for transfer learning). The difference: personally I wouldnât care for the model itself at all, but just ignore that and look for changes in the surroundings (where surroundings includes the toolhead as a likely cause of fire).
I agree, itâs not the most difficult challenge out there to realize, but itâs more than an afternoons work. (At least for me it is :p, with a background in software engineering and some high level AI experience, but not on image processing)
