If you were to start all over again with your Smart Home & Home Assistant journey today - what would this journey look like in todays day with current technology

Social
1

Hi all, I'm new to the community.

By the end of the month I should be moving into my very own, first home. It has always been a very cool idea and aspiration to create a smart home when I purchase my first house, and now that time is coming.

I've been doing a fair bit of research into this, learning about HA, Matter, Zigbee, Thread etc that are completely new to me and things I have never heard of a few years ago until now.

With my research I seen amazing ways people have utilised HA and created many things that I want to see in my house, but a journey has to start somewhere and for some it started 3, 5, 10 years ago - but for me it starts by the end of the month!

I have seen video's of other people's setups and although they appear great and functional, it gets me thinking whether or not if they started with the tech available now - what would they do different if anything at all? Which is sort of why I don't just want to copy someone's setup that may have taken them 2 to 3 years if not longer to create, when there may be newer tech and better ways of doing it.

The reason why I want to proceed with HA is simply due to not wanting to be locked to set ecosystems, I love the idea of HA being something you run locally and its not controlled or limited by the big companies that have been doing this sort of thing for a while like Apple, Amazon, Google etc.

So the big question is, if you started all over again - what would you do? What would you buy? I hope to achieve from this post not only a nice interaction with some of you, but for it to also serve as a guide for me on the best way to start my foundation from the best device to setup HA with (HA Green, Raspberry Pi, Mini PC's etc) and then develop it further beyond that.

1 Like
2

Congrats on your soon-to-be home! The best guide I can offer is to recognize that home automation is a journey, not a destination. Things you assume will be awesome before you start will turn out to go largely unused. Things you didn't anticipate up front will turn out to make you smile every day. And those things will change over time. This applies to both hardware and your automations.

So don't worry (too much at least) about planning and trying to figure it all out ahead of time. Instead - get started with something/anything, and then tinker. Find what works for you (and critically - the others in your household.) Embrace change and evolution. Enjoy the ride!

3 Likes
3

Ethernet not USB coordinators

4

I will try to stay general.

You will make mistakes, for example: buy something that doesn't work as you want or end up buying something better and never use the first thing again. My advice is to start slow and build up over time, so you get a feel for how things work.

Likely your first automations will be in one of three domains:

  • Lighting.
  • Climate (Heating/cooling) and/or water heaters.
  • Security - Door locks, cameras, door/window sensors.

Some of the hardware will overlap, you can use IR sensors both to turn on lights and as part of a security setup.

Think about what happens when things goes wrong, other threads will suggest:

  • Buying smart thermostats where you can control the set points from HA, but the actual control of the heating/cooling is done directly by the thermostat, so that if HA goes down you will have local control.
  • Preferring smart switches (with dumb bulbs) for the same reason - the smart switch can still be used directly in the worse case.
  • This point also rolls into support for guests - you don't want to explain to everyone how the lights work.

There is some consensus on the forums about "better" options, but at the end of the day you need to figure out whats right for you - you may make different decisions to the majority on this forum.

I would also say your (tech) background and experience with smart homes will also affect your choices.

A note of caution: I would focus on "useful" before focusing on "cool", once you build a reliable system that makes your life easier (that is stable) you can start to add new features over time the setup will probably start to look cool due to the shear number of things it can do for you, but having things keep failing is not cool, its just annoying.

Case in point my system is not very complex, I control lights in 5 rooms and 2 AC units I have a few door and window sensors which I use primarily to turn off the AC when said doors or windows are opened, its not a super "cool" system.

However I do use the lights to notify me of things - the light next to the window turns blue to tell me to open the window (I have outside temp sensors) - once opened the AC turns off and the light goes back to normal.

I could give the opposite advice I have two USB coordinators and have heavily favored Zigbee so even if my WIFI goes down most of my automations keep running.

But there are pros and cons of both approaches and the specifics matter less than the overall strategy / what you are trying to achieve.

TL;DR make a plan for simple things you would like to achieve to make your life easier.

Be open to updating your plan - I no longer use manual switches for hallway lights (i.e. rooms I am only passing through) instead I rely on IR sensor to turn the lights on - that wasn't a goal I originally had.

Not everything needs to be automated I have a button next to the front door that the last person to leave presses - sure I could do presence detection based on phones etc - but I just don't see the value add and it's a nice reassurance to see the house go into "away" mode before I leave.

5

I already gave a much longer answer where I tried to stay objective.

With respect to the "what would I actually do", frankly I know what works for me I would just buy that again:

Hardware

  • I ebay'ed a second hand mac mini (around $70 USD with shipping).
  • ZBT-2 coordinator.
  • Bunch of Zigbee
    • Smart Bulbs (Philips, Sengled, Aqara, Third Reality)
    • Smart plugs
    • Door/Window Sensor
    • Sonoff IR Sensors
    • Sonoff MM Radar sensors
    • Sonoff Temp sensors
  • Broadlink RM4 Mini (with optional temp/humidity sensor). **

** - The RM mini was a surprise win, I bought the extra temp module because it was only a couple of extra bucks but it turned into a full solution for controlling my AC.

Software:

  • Installed Ubuntu on the Mini
  • Run mosquito on the Mini
  • Run HA in docker.
  • Run Zigbee2MQTT in docker

Networking:

  • IOT has it's own VLAN (Fire-walled from my main) with its on WIFI SSID

A final note on Zigbee bulbs, every manufacture has their own set of quirks (including price).
I would likely end up buying different brands for different use cases.

Here are some of the quirks I have found in the bulbs I own:

1 Like
6

As stated above, it is a journey, not a destination.
I started 5 years ago almost by accident. My neighbor had a smart lock that he could unlock with his phone. I thought that was cool, so I started searching for locks with that capability. Based upon the visual appeal of the keypad, and the fact that it had the ability to be controlled remotely, I opted for a Yale Z-Wave door lock. Then I found out that my neighbor was using his alarm system app to unlock his door. I did not need or want an alarm system, so I set out to find the least expensive way I could get my lock into a Z-Wave hub that could also talk to my phone. I ended up with a HomeSeer Home Troller Pi.

I paired my lock, and I could open my door with my lock. But it was a little hit and miss. I found out I needed a mesh to repeat the lock's signal. So I bought 3 Z-Wave light switches so I could control the outdoor lights on the front of my house while also providing the repeaters my door lock seemed to need. So basically, my choice in door lock dictated my choice in my first set of light switches. I got to know Z-Wave pretty well.

Then I built a steel shop building next to my garage. I wired the three lighting circuits with Z-Wave switches and put another Z-Wave door lock on it. But no matter what I tried, this lock would not reliably respond to Z-Wave commands. The steel building was blocking the radio signals. So I got a Z-Net IP Z-Wave interface that I connected in my shop with Ethernet, and I had a second Z-Wave network just for that outbuilding. Worked great. At this point I'm still a dedicated Z-Wave only guy.

Then I wanted to control a celling fan with lights in my guest room. But this room only had a single-gang box. One dumb switch powering the fan/light combo. The only way to control the fan was with the pull chain. After quite a bit of searching, I found the Inovelli Blue Fan Canopy Module that could be bound to the Blue On Off switch to control both the fan and the light with a single smart switch. These are Zigbee devices. So I got myself a Zigbee dongle and the canopy module and switch. Just like my choice of locks pretty much forced me to jump on the Z-Wave bandwagon, this single-gang fan/light dictated my introduction to Zigbee.

I have since added more than a dozen Z-Wave and Zigbee switches and dimmers, a Z-Wave motion/light/temperature sensor for outside measurements, and a couple of Z-Wave multi-relays to control and monitor my garage doors. I have no motion or presence sensors indoors, and I have no use for them. My system is not a smart home in the way most people think about that term. I don't want my smart home predicting what lights I want on or off and controlling things based on my moving about in my house. Mine is really a glorified dusk to dawn lighting timer. And that's exactly what I want.

I was on that HomeSeer Home Troller Pi for a few years. Home Seer HS4 is a .NET application. On the Pi it runs on the now deprecated Mono library. I wanted to be on a supported native platform for the system, so I moved my HS4 installation to a modest Ryzen-based mini pc. Since the mini pc ran Windows 11 Pro, I had Hyper-V available. Since Home Assistant is free, I decided to spin up a VM running HAOS. Then I went about the task of learning how to replicate my HomeSeer automations in Home Assistant. It took me almost a month before I was comfortable enough in Home Assistant to have replicated everything I needed, and I shut down HS4.

I thought I was a HomeSeer guy, and Home Assistant was going to be just a fun experiment. Turns out, I like HA way better, and I haven't fired up HomeSeer in months. Nor do I plan to.

About a week ago I was reading about Proxmox and how it was superior to Windows 11 Pro running Hyper-V. My Windows server was running fine, but it's memory usage was over 80% constantly with spikes above 90%. I know the sysadmins at work would treat this as an emergency, so I formatted the SSD on the mini pc and ported my VMs over to Proxmox. Now I'm running a HAOS VM and a Debian 13 VM. The Linux VM is my Docker server where I'm externally hosting containers for Mosquitto, Zigbee2MQTT, and two Z-Wave JS containers. All my radios are external to Home Assistant. And I have a LXC running rclone to sync my nightly VM backups to OneDrive.

That's been my journey so far. Every time I make a change or add a new device I say the same thing. I'm done. I can't imagine another thing I would want to automate or control. And then, inevitably, something pops up a few weeks later and I start playing some more.

But that was a longwinded way of getting to your question. If I started over today, what system would I build? I would build exactly the system I have today. Mini PC running Proxmox with a HAOS VM, a Linux VM, and a RClone CT. As of right now, for me, my system is about as perfect as I can imagine.

What about Matter and Thread? Since my Z-Wave and Zigbee are working fine, and I have no current need for new devices, I have no need or desire to try that protocol. Plus, to this day, my favorite door locks are still the Z-Wave models I already own. They just look better on my doors than any of the Matter locks I've seen so far, so I'd still opt for the Z-Wave/Zigbee route even now.

1 Like
7

Begin with smart lighting. The Most important questions:

  • What is my budget?
  • on/off only or dimming
  • only white or warm/cold white lights or full color
  • Where do you want what type of (smart) light (bulbs, LED strips, etc.)
  • Does it scale? ( WiFi bulbs are not a good idea for a whole house). What product types are available
  • How can I control my Smart Lights via wall switches?
  • How to automate lights using motion and presence sensors, etc. This is the part that makes your home smart
  • Nobody should be able to disable your smart bulb by pressing the old style light switch

You want to be able to control your lights via

  • wall switch
  • HA Dashboard
  • Voice Control
  • Automation by motion and presence sensors, time, brightness, etc.

After answering These questions you can choose your product. Start with a few bulbs, Test it a few weeks, and if it works scale…

In my Smart Home in Germany I use Philips Hue bulbs and motion sensors only and Aqara
Presence sensors and the Friends of hue Enocean Type wall switches

I use ca. 100 Hue products at the moment (bulbs, strips, motion sensors, wall switches, lamps, everything) and to speak with Paul Hibbert „it just works…“

Install Home Assistant OS on a mini PC. Do not use the docker Version. With HA OS you can use AddOns to add functionality (e.g. a MQTT Broker) besides integrations. Its more powerful than a HA Green or a Raspberry Pi. Given your experience, Proxmox is too complicated (I run Proxmox with HA VM personally). At least for now

P.S.: Keep it simple. If a brand works, stick to it. Complexity comes by time and progress :grinning_face:

1 Like
8

Care to explain? I'm starting out too and chose a cheap SONOFF USB dongle and to focus on ZigBee. I do have one WiFi/MQTT device (Glow energy monitor, which ironically connects to the UK energy company smart meter using a separate ZigBee network).

@MrExzibit For the OP, I also suggest you consider if you have important aims (such as independence as you mentioned), and what you'll need to achieve this.

I began with the same (independence, privacy) , plus "ooo automation looks fun" but no idea what that meant in practice. That soon morphed into a HomeLab (search Mastodon for that hashtag) so I can build something independent and resilient over time, and now Home Assistant is an important part of that rather than the centre.

My main aim though is to have fun and not turn this into "work". Hope you enjoy it, and your first home. Congratulations!

1 Like
9

Ethernet coordinators gives you more flexibility in terms of placement and also does not lock to a single hardware device. If you virtualize you have to option of moving your workloads from host to host without having to bother with USB devices and where they are plugged in.

4 Likes
10

I would argue that smart dimmers / switches with dumb bulbs is a better option than having loads of smart bulbs (if you don't need colored lighting). It saves you money, you don't have to worry about family members turning the smart bulbs off by accident and with some dim2warm bulbs you also get some temperature control on the lighting. They also work when HA / Zigbee2MQTT / whatever is down out of the box.

1 Like
11

At the end of the day coordinators are cheap (around $50 USD) so you could buy both or plan to buy the other later. I first bought a Sonoff USB then replaced it with a ZBT-2 (USB) the primary reason was that I wanted to move from ZHA to Z2M and frankly I didn't trust the migration scripts.

If you are going to move coordinators I would suggest having both networks up at the same time because it means you can migrate slowly - making sure things are still working.
Additionally I found that (almost) all device would slide from one network to the other simply by putting the new network in join mode and kicking the devices off the old network.
Some of the battery devices were a little more obstinate, but a press of the pairing button got them to move.

With respect to placement of the antenna thats only really a problem if you want to run HA in a server rack (ProxMox or similar) if you have HA Green or a stand alone machine, you can just place that where it needs to be - that said don't plug USB coordinators straight into the back of a mini pc - it's much better to get a USB cord (a couple of feet is fine) so that the antenna isn't blocked by the metal in the PC case.

Device lock in is a question of whether the device allows you to change the hardware ID of the co-ordinator, most ethernet coordinators have that feature, but some USB ones have it too.

12

Or you could buy an adapter that you can migrate to without having to re-pair devices. Zstack to Zstack for example.

I specifically said that this is something needful when you virtualize - you cannot migrate a virtual machine from one host to another seamlessly when you have a USB adapter you depend on. It has nothing to do with hardware ID :slight_smile:

13

I run dockerized directly on Linux (no ProxMox).

I can move my installation from (Linux) machine to (Linux) machine simply by:

  • Stopping the docker containers
  • Copy the config directories (Z2M and HA) to the new machine.
  • Move the USB adapter.
  • Download the same or newer versions of the docker containers.
  • Fire up the HA and Z2M docker containers on the new hardware.

There is nothing magical about USB coordinators at the end of the day they are "Serial" devices as long as you can talk to them they work the same irrespective of what they are plugged into.

It's also possible to point multiple HA instances to the same Mosquito broker - it's not ideal since you might end up with double execution if both HA's have the same automation, but it has its uses.

We may be talking about 2 different problems, but you should note the hardware ID of your existing coordinator (before it dies) then when you buy a new one you need to check you can change the hardware ID or you will be forced to repair your entire (Zigbee) network - I can't speak to non Zigbee networks.

14

Ethernet coordinators do not use Wifi. They use ethernet, like the name suggests.

The coordinator should be placed as centrally in the house as possible, so that you avoid as many additional hops via routers as possible. If that central room is somewhere like a corridor, it's gonna be harder to hide your server + coordinator, even if you're running on a Green or a Pi.

That's not to mention the additional cabling - you'll need power, ethernet & a USB cable to prevent interference (and that's assuming you're not using an external SSD on something like a Pi which will require a powered USB hub). That's a lot of cabling to pull & hide vs the single ethernet cable required for PoE coordinators.

1 Like
15

Exacty my point. You MUST move the USB adapter - on the ethernet based ones you don't. If you have HA running virtualized on some hardware tucked away in the attic this is a big pain point.

Again, what I am talking about has nothing to do with hardware ID. It's the concept of not being bound to a USB device that I mentioned as a bonus on the ethernet based adapters.

16

I wasn't explicit in my previous answer, but the point I was making was the same, if you have a need to run the HA somewhere else (which could be because you don't have somewhere to hide it) then an ethernet based coordinator may be required.

The key point I was arguing was that if you DO have somewhere good and central to place your HA machine, you don't need an ethernet coordinator.

With respect to cabling in my case I could run HA over WIFI (since all my IOT devices are WIFI anyway) so that creates a wash on cabling as long as I have power.

17

No-one has said you need an ethernet coordinator - how ever starting from scratch which is the topic of this thread, it is a better option, especially if you want to place the coordinator centrally or you want to virtualize and be less hardware-dependent.

Having IoT devices on wifi has nothing to do with how well (or not) a home assistant server behaves on a wifi connection. If you have the option to use cables that is always preferred over wifi.

18

I will go back to how I started my involvement in this thread.

For some people there will be clear reasons to pick USB vs Ethernet based coordinators.
For others there won't be a strong reason one way or the other.

If you happen to have an ethernet drop next to HA you may as well plug HA in.
If you don't have an ethernet drop next to HA (and your chosen HA installation supports WIFI) then WIFI will probably work fine for you.

I purchased the ZBT-2 because I thought I might switch it to thread mode at some later point, I don't know which way I will jump at that time I might move to a Network coordinator (to handle the Zigbee network) or I might stay with USB - to me is not a clear win to move to a network coordinator.

19

Matter and Thread are new to everyone, and in my opinion not ready for wide adoption. Out of the box, Matter devices provide hub agnostic connectivity. But Matter is a semi-locked protocol in that to get advanced features in Matter devices, you have to buy the manufacturer's hub and/or subscribe to their cloud service. I've seen nothing in Matter devices that can't be equaled with Zigbee and Home Assistant automations.

Over a million installations of Home Assistant, and I'll wager that no two are alike. You design your system around your preferences and needs. Use others' as inspiration. If you try to copy them, you will be asking for headaches.

First, kill all the ISP-provided routers. They come from the low-bidder and are not up to the task of a diverse smart home. Most ISP provided routers only have enough RAM to handle 30-40 clients. I found this the hard way when WiFi devices dropped off the LAN and new devices wouldn't come online. I finally fixed that by buying commercial-grade router, switches and wireless access points. I use TP-Link Omada and have over 125 WiFi and Ethernet clients. (Omada is the budget version of Unifi)

Starting over, I would use more Zigbee devices.

Most start with the Green or a Raspberry Pi, and some usually move up to an X-86 mini computer. Or they just start there with an Intel NUC or Beelink mini-PC.

Smart Bulbs- unless you need color control, smart bulbs are dumb. There is nothing but color that smart bulbs can do that dumb bulbs and smart switches can't.

20

Big mistake. Your server would occupy most of your WiFi bandwidth and make devices sluggish in response.