We’ve been talking a bunch of home automation on the Podcast lately, and this week, in the Mailbag segment, a reader asked us about our setups. Neither Kristina nor I are poster children for the home automation movement: she has absolutely no smart anything because she didn’t want her data up in “the cloud”, and I have an entirely local system that’s really nothing more than a bunch of ad-hoc scripts that talk to an MQTT broker, everything fully DIY but held together with metaphorical duct tape. Neither of us are doing it right, but we’re doing it wrong in interestingly different ways.
Kristina thought, probably because of the range of commercial devices out there that tie you into using their remote data storage services, that giving up control of her data was necessary to use it. And it might be, if you insist that setting up the system be as easy as possible. But the tradeoff for this ease is a drastic reduction in simplicity. You shouldn’t need a remote server in some foreign country to turn your lights on and off. Adding “the cloud” into the mix brings a lot of complexity, mostly in the form of servers that have to be paid for somehow by whatever company is providing the service. It needs to be secure. You might even have to create accounts, remember passwords, and manage that whole deal. Sure, that’s easy enough, but it’s a lot of moving parts, and you can’t blame her for rejecting that complexity.
My system is hosted on a now-ancient OrangePi in the corner, and the network in question is an old WiFi router that it sits on. Nothing needs to leave my four walls, but actually some of it does – I bridge some of the MQTT topics out to an external server for my own amusement. There is no protocol, and no real “system” frankly. Each device in the network has its own topic, and I’m responsible for knowing what it means. The thermometer in the basement has an ESP8266 that transmits on the home/basement/temperature topic, and it puts out its temperature in degrees Celsius. It was the simplest system I could think of, but I have to write whatever software I want to log, display, or act on the data. Of course, that’s simple if you can write some four-liner scripts on the OrangePi broker, but it’s not easy enough that my wife wants to hack on it.
So if the full-buy-in commercial systems are easy but overly complex, and my DIY network is transparently simple but requires a level of hands-on that isn’t easy for “normies”, is there a middle ground? I know half of you are already screaming Home Assistant or Domoticz, and you’re also thinking of which client device libraries you like the most for all your DIY applications: ESPHome vs Tasmota, for instance. And you’re all right!
We are living the in the golden age of the home automation projects. Open-source software and firmware, combined with an abundance of online tutorials and worked examples, have made huge strides toward bridging the gap between simplicity and ease of use. You can set up a hub for everything on a single-board computer, upload the software of your choice, and you don’t need the complexity or loss-of-support liability of a cloud provider. At the same time, setup is easy enough if you’re willing to roll up your sleeves a little bit, and when it’s not, chances are good that someone else has already figured it out for you. These days, interoperability with popular commercial products is shockingly easy to boot.
I need to spend some time and rationalize my system: given the state of the art, it’s simply too simple, and taking a step into an open-source solution would make it easier to use for the rest of the family, without overly complexifying things, adding sketchy dependencies, or losing our data sovereignty. I haven’t finished exploring my options yet, but from what I can see, the community has converged on some goldilocks setups: not too simple or too easy, but rather just right. Thanks, y’all!
I only have a headless Pi3 with temperature sensors inside the house and outside, so I know better when to open all windows, especially in the summer.
No judgment from me, but home automation just seems like a nothing burger. I can’t see enough benefit to put any money toward it, much less time.
It can be well worth it if you have solar panels, feed in tarriff or not. For example a guy I know figured out that in winter it makes more sense financially to use the excess to heat his house in the day when he’s not even home, because the heat sticks around well into the evening. I think one of his heaters is a concrete block “storage heater”
The cost to achieve a similar level of evening warmth with gas fired central heating, is higher than the payout he could get from selling the daytime electricity to the grid!
Doesn’t really need home automation to do that, just a simple timer in the heating system thermostat you can switch in between summer and winter. I guess you can call that automation, but that’s not the Home Automation that this article is talking about with networked devices and whatnot.
it’s actually cheaper to buy a smart plug than a timer now.
That’s how I started.
And then you start writing scripts to automate the lights when you’re at home working, with another mode for when you’re on vacation to make the place look occupied. And then the Christmas tree, so that it looks nice in the morning, but doesn’t burn the house down over night….
Plus the Raspberry Pi and other stuff?
Very simple automation can be very cheap. Just a smart thermostat and a few smart plugs that can be connected to Google Home. You can even go for some schedule mechanical plugs for even more simplicity.
There are some quality of life improvements with smart devices, but most of the time there’s no real point in networking them together even locally because there’s no real need to automate anything – a well designed house doesn’t need constant monitoring or adjustment.
So, most things sold as home automation are simply gadgets and nerd toys.
It kind of is, there really is no amazing benefit in home automation, which is the reason it has never really caught on. Another big part is the cost and non-support from the commercial players. Who wants to set up an expenssive system for a little bit more convinience, when in 10 years max you need to buy a whole new system or in bad case in a year or 2.
And one big cost is electrical work, if you need to make any changes to the electrical connections in the house, that stuff costs a lot.
But i do mine myself, and it mostly just monitors stuff, it’ll control the heating and turn off water when i’m not around. Maybe a security system, you know. At this point i’m not going to touch the electrics in the house. And that’s with a free miniature industrial PC. The sensors and such i’ve bought. It takes time though, it’s not plug and play in any way.
Pretty much all the things home automation system does can be accomblished without a central system, except for gathering the data. Now it might not be very valuable data, but perhaps something could be found there to make things better.
The ultimately best feature i see for houses is humidity monitoring of the structure of the house. But that would require big sensor network in the structure built-in, not just in the room air. Humidity and fire damage are the most expenssive damage a house can suffer. We have smoke alarms, but we don’t have a structural humidity alarms. Kind of like bearing vibration monitoring system in some industrial machines, which helps detect bearings which are about to go bad. Also I don’t understand why cars don’t have one, it’s a cheap system.
But yes, it takes time and money, much less so if you use more time and DIY, but you obviously need to consider what the benefits for you are. Also, if you are going to sell the house, the system either needs to be ripped off or it has to be documented so well, that the new buyers aren’t scared of it.
It never caught on 🤣🤣🤣 it’s good tk see you can get internet from under your rock on mars!
I use a smart ceiling to monitor structural humidity. If my ceiling is solid white, the humidity is acceptable. If gray or yellow spots appear, humidity is excessive. It’s an easy visual alert spanning almost the whole house. 😅
If you live in your parent’s basement and don’t pay the utility bills then automating your energy use is a nothing burger. If you are interested in moving your electricity usage from on-peak to off-peak to save a 300% cost increase based on time-of-day consumption then automation can be helpful. If you are in the habit of leaving the windows open in winter with the heat on, and/or, if you leave the windows open with the A/C on in summer, then automation and energy efficiency and cost management isn’t an issue for you because you are independently wealthy by inheritance, you have a sense of entitlement, and your slaves work for free if they know what’s good for them.
You can save money that way, but the tradeoff is living like it was a different time zone. I want my dinner and TV pretty much the same time as the rest of my neighbors.
That’s why I signed a flat rate contract, which costs more per unit than the average price you’d get on a variable rate contract, but I pay less at peak when I’m actually using most of my electricity so the actual average is less. I also don’t get the surprise surcharges when the system prices in neighboring countries peak and the utilities decide to jack up the rate despite having no shortfall of electricity here.
I miss my X10 system 😢
My lights are still operated by X10 controllers 😊
it was darn expensive. Nowadays SPE beats the ass of X10
Meh. Started doing this supposed H/A stuff back in the late 1980s using current loops on twisted pairs and 8051 controllers sprinkled around my little universe.
Tried wireless for a few years, now I am back to using 20mA loops for the house, and 300/400MHz for the two greenhouses and four weather stations because I don’t wanna string cabling up and down the hills. None of this stuff uses any formal I/O spec or any of that craptastic commercial hardware.
The many published protocols, commercial modules, controllers, etc are burdened with flakiness and convoluted code. Most people do not need H/A – it just adds another layer of complexity to your life. You may require a security system, but you seldom need H/A.
I second current loops. Shielded twisted stranded pairs 4-20 mA kinds.
Was looking for good reliable documentation for amateur DIYers, seem simpler in practice (humble home, wide tolerances, so I won’t need industrial grade sensors) than CAN bus and sure been around longer than anything to prove it Just Works When It Needs to Work. As a nice side effect of being analog, it allows for cheap analog VU meters (after obvious voltage drop from suggested 24V) to supplement the ubbity uppity MCUs.
For now I am going with Attinys for ADC/number-crunching, as I have plenty of spares left after other projects, but could be upgraded to some Pi Pico if/when merited. Oh, found this nifty board that has rp2040 (pico) paired with dedicated tcp/ip stack chip, presto, instantly online, future expansion. (i’ve tried ESPC3 before – similarly, works kinda sorta, just too power hungry for my humble needs, later, maybe, if I consider WiFi).
I mean, there is no doing it wrong. If your system works for you, do it. I just find Home Assistant to check all the boxes for what I need, so I do it. None of my stuff is cloud-based, which is just a choice I’ve made as I integrate and build my own smart devices. It has to be able to be LAN-only or I don’t purchase. HA making it easy to have everything in one spot and have them interact is quite enjoyable when making things automate how I want them to.
I was all in with Home Assistant at one stage. Now I’m mostly Matter with Scrypted for the cameras. I would like to drop that as well when I can get some good Matter cameras. Hoping for at the very least we get Matter support in Scrypted.
I dove head-first into IoT platforms about 12 years ago, especially the ESP devices. I whipped up some test environmental monitoring, put a MQTT server on my home Linux box, put a sample webapp on my hosted webserver, flashed some SonOff devices with tasmota, etc etc. I even prototyped a whole web-based remote boat-monitoring system, til I found that another developer had gone there first and did it better.
Anyway my IoT stuff rarely went beyond experiments and proof-of-concept. I never finished a fully working home-automation system. Why? Because it’s just my wife and I, and our house is small and uncomplicated and everything is working fine. About the only time I wish for home automation is when I’m going to bed on the second floor, and I can’t remember if I switched the basement light off or not.
So this admission tarnishes my geek cred, I know. But we haven’t yet imagined a compelling reason to do more home automation and online monitoring. If anything, remote monitoring for security and to get notifications of intrusions, system failures, flooding, power-outage, etc would clearly be useful.
That’s what I use those 433MHz wireless plug adapters for. You don’t have to know whether you remembered to switch it off, you just press the “off” button and now it is. You do need a duplicate remote downstairs though, so you don’t have to run upstairs to turn it back on.
It’s not even automation, it’s just a dumb remote switch.
Doesn’t understand home automation (or anything, really), so writes about why home automation isn’t necessary.
I’m waiting for someone to explain it to me, so I could agree with it.
So far, crickets.
You could argue that the wireless plug adapters, or any thermostat (even a bimetal coil) is a form of automation.
But most of the home automation builds I see are the sort of thing done purely for the cool factor rather than a practical reason. They’re the home electronics equivalent of my heavily modified 1966 Dodge Dart, not the Kia I use for everyday transportation.
There are a few standalone automated devices that have a possibile benefit. My house is on a heat pump so trying to back down the temperature when nobody’s home does not work particularly well, but I can see the benefit of a remote controlled programmable thermostat for other heating options. I’ve thought about a few timer controls for outside lighting.
But most things in the house aren’t things I need to interact with unless I’m already near them. And I really don’t see why I unrelated devices should be tied into a common system except in off-grid situations.
Outside of temperature control, the home automation scene hasn’t rolled out any must-have items for the general public. Just a few interesting toys.
It may be automatic, but is it home automation? Usually we don’t count it as home automation if it’s just individual things operating without central coordination.
I find that the more things you tie under one system, the more things will fail at once when it inevitably malfunctions.
In reply to Dude’s reply: The central coordination is the least defensible and least understandable part of home automation. Central coordination only makes sense if you are living off grid and need to ration power, or if the devices are capable of working together.
So until a smart fridge is capable of physical handing off a pizza to a smart stove…
For me, it is DIY at its most fun — most of the time!
I use Tasmota for most of my sensors, smartplugs, and some light bulbs, but have started to dabble in Zigbee (dabble?? I have 35 devices already!). All of it is tied together with Node-Red — which I discovered long before Home Assistant — and a local mosquitto (MQTT) broker and, recently, a Zigbee2MQTT server.
I still haven’t found ways to do in HA what I have been doing in Node-Red…
I recently ran across some comments in my Node-Red config that were dated a decade ago!!
Almost everything stays in house. I monitor temperatures and humidity room-by-room, energy usage for the large appliances and 3D printers (and far too many other minor devices), energy production and usage of my solar panel setups, and control a basement dehumidifier as well as some auxiliary heaters. The Node-Red dashboard makes it simple (easy?) to chart everything. It’s not a perfect system, I’m constantly tweaking it or adding new stuff.
So, there is no wrong way to do home IoT or automation, as long as you can understand the various pieces, but you can go deep down that rabbit hole when you DIY…
I have no “real” home automation to speak of, only my Honeywell Evohome thermosthat. I find the “Your thermosthat has lost connection” e-mails more useful than the actual being-away-and-heating-my-house, which I do once every two years…
I have a homebuilt home entertainment system based on two raspverry pi’s, one with lirc and kodi and one with NFS and an external hard drive, which is so cumbersome to operate that when a friend housesat, I just pulled out the HDMI cable and told her to plug that into her laptop.
So I’m feeling the gist of this post.
the main use i have for home automation is automating the switching on off of loads at my off-grid Adirondack cabin, which I drive to in an EV that must, at some point, be charged. i also largely heat electricity with a heatpump-based hot water heater that is another thing that can be timed to run only when there is lots of sun. i am in the process of adding a weather forecast API so I can aggressively draw down the battery on the mornings of days that will be sunny so I hit 100% battery charge as little as possible, because after that, the electicity is wasted. my biggest ways to waste electricity are to run a minisplit and even resistive heaters in the basement (which, in this climate, can always benefit from heat) but am open to suggestions such as bauxite refining or hydrogen generation. see my totally custom system (which runs on the server, in the microcontroller, and in the browser) here:
https://github.com/judasgutenberg/Esp8266_RemoteControl
“Kristina thought” can hear it clacking in the dark
I had an full x10 system at the old house built in 1998, but when I moved to this house built in 1970 none if it would work. I could not get the controller to work on a lamp module in the next room or the flood lights. I have 4 tapco lamp modules, a light switch in the back patio on timers. Also a myq garage unit and a light switch that controls the lights in the garage that works with it l. MyQ, tapco and Simple safe are all different apps that dont talk to each other. The ring door cam does work with Myq and knows if the garage is open, ring is another app. One app to control everything would be nice, but Myq and Simple Safe are not Matter compliant or any other nonpritory protocol. My front door Sledge dead bolt I think had some protocol but uses a lot battery power to turn it on so have not used it. Bought some peice of junk controler from Amazon which said it did stuff but does not see my Matter complant tapco modules or anything else. Thinking of getting Home Assistant
Home automation is a ART at this stage. There is no one solution that fits everyone.
Automation is for piece of mind and cost savings. If it is not delivering on that front, something isn’t right. Remote control is secondary but important feature and current state is useful.
I mainly have Tuya devices and the different manufacturers quality define if they are replaced in a year of use because of failure or not. Often they are sub $20 each, so no big $ expense when needing to replace but annoying with automation configuration.
I tried HA on a SBC and used it till the uSD gave up. I wanted to have the power monitoring capabilities of devices and assumed power usage of some devices like lights also. Powercalc plugin indicated it could do the second part but overall it didn’t feel like a polished system and it needed TLC to keep being informative.
HA wasn’t delivering on the piece of mind or cost savings, so I have not configured it again. It was more of a time sink.
Currently the Tuya app SmartLife supports all brands products within their eco system and has Google Home intergration. It passes the easy to use test for manual remote control.
However the devices can randomly stop responding on Wifi and need to be power cycled. Randomly lights go into programming mode and flicker till they are set up again otherwise they go offline.
My first bit of home automation was in the late 90’s when I got a PLC from work. It was being scraped due to Y2K issues. I just set the clock back a few years and played with basic lighting controls till I moved out and junked it. Years later I saw some smart plugs that were cheaper than plug in timers and got me right into Home Assistant and all sorts of systems. Now if it isn’t Matter it gets ignored.
I need less things in my life so I can have more time with things that matter more to me. We only have so much time on this planet. Spending it as the system admin of an iot light switch is not how I want to use my time.
Commercial stuff goes out of support in a few years and turns into e waste. Open source stuff means you have to explain it to your significant other and be responsible for it when it doesn’t work correctly.
Keep it simple stupid.
I like home assistant. I do try to keep everything I can from accessing the internet. I got almost everything I have inside zigbee2mqtt with a few wifi devices as exceptions. I’m generally not in favor of using “AI”, but writing code for home assistant automations is an amazing way to get it setup.
Just came here to say that Claude Code as a Home Assistant MCP (so it can interface directly with it), and is getting really good at writing automations and troubleshooting issues.
I found that a real Simple House Automation like a Raspberry pi with a temperature sensor can give you a real heads up on how your house works. Our house is from the UK 1930’s and although we have insulation (mostly in the roof) the house sinks heat away in the winter like you couldn’t believe. So you can see this by monitoring and graphing the temperature really simply. On our temperature waveforms it looks like a sawtooth waveform. Ironically its not cool in the summer and accumulates, WTF???
So Automation can be used as useful monitoring, mind you the climate change conservative UK outlook is for 45 degree summer norm in 2056.