Wow, I did not know this was a problem. So the USB drives are more robust than the cards?
I assume it will utilize one of the four USB ports on the PI?
Is there any difference in the installation of the OS or the performance of the HA system after installation?
Sorry for the questions as this is the first I have heard of this.
Just a little warning… If you are going to use hass.io, I do not believe usb drives are supported for the os.
The larger the sd card, the less wear on each write to it (more balanced wear leveling). So for the cost difference between 32GB and 64GB…
Luckily I haven’t lost a SD card yet but I switched two of my Pi’s to a HDD already. I didn’t recognized a big performance jump but for me it was more to be on the safe side.
I don’t use hass.io, so yeah please check Dixey’s comment 
I did not realize hass.io was something different. When selecting a catagory this seemed to be the best place to put it.
I am exploring just regular HA. Should I move this thread?
I apologize for my ignorance.
You’ve probably seen this, but there is another interesting discussion going on here:
I run a ‘test install’ in a Docker container on my Synology NAS - and if I weren’t using the GPIO ports of my RPi3 for some Reed switches, I probably would have moved my live system on there as well by now.
They are nearly the same thing, but Hass.io runs on its own OS. This makes updating and backing up your system much more straightforward, but does limit your access to the hardware vs. just installing home assistant on a raspberrypi running Debian.
Others can probably elaborate better on the difference, I am running home assistant in a virtual environment on Debian stretch but am considering switching to Hass.io. It’s the default approach recommended by the documentation now.
I think you know this already but what you listed there? That’s just the admission fee for an HA system.
The cost of an “HA system” is much, much higher than the sum of those parts. If bitten by the HA bug, in a few years expect to look back and marvel at how much time and money you’ve sunk into your new obsession.
Unless your idea of an “HA system” is an RPi running Home Assistant with one smart switch, prepare yourself to part with many more bucks in the near future.
FWIW, I’ve had Home Automation as a hobby for over a decade so I know it doesn’t begin and end with an initial cash outlay of $100. 
2 Likes
You mean the assortment of $40 cameras that were unreliable and failed after 6 months so you had to replace them all with $100 cameras?
You mean the $20 per window and door sensor for the ones you didn’t have to take a soldering iron to to make compatible?
You mean the $40 per reliable motion sensor?
You mean the $200 home thermostats?
You mean the $200 zwave combo locks? (oh yeah, and the hub)
You mean the (okay… what’s your time worth) entire weekends spent debugging and making some obscure sensor talk right?
You mean those costs?
4 Likes
I see you’re not new to this thing called Home Automation.
Oh yeah, all that and, in my case, untold hours learning APIs and writing software drivers. Sometimes it’s learning a whole lot more than you ever wanted to in order to accomplish one silly little thing. Then there are the hours spent surfing forums to learn new things, ask questions, and try to help others in a bind. Fortunately, most of the time it’s fun (and if it ain’t then it’s time to find a new hobby).
You are on the right track!. I moved from several years of using Smartthings to HA back in July.
Could not be any happier.
I have exactly the same setup you listed. Here are some of my devices that work together very reliably:
- Gen 1 Phillip HUE bridge that controls all my Zigbee bulbs.
- 43 Z-Wave devices (motion sensors, wall/plug switches, door sensors, etc).
- 2 Konnected.IO boards to merge my traditional home security sensors with HA. <-- This is a killer solution!
- 6 WiFi WEMO devices for areas that are too far for ZWAVE (front gate control, landscaping lights, etc)
- 2 NVIDIA Shield TVs for media consumption
- 1 Logitech harmony hub for my media center and some IR only devices (eRod curtains)
- Ubiquiti UNIFI devices (Gateway, Switches, Cloud Key, and 4 WiFi APs) - used for device tracking
- 6 Blink cameras
- 5 Foscam cameras
Some observations based on my experience:
- 64Gb SD cards night be overkill. 32Gb is more than enough but it won’t hurt.
- You will have to decide if going with hass.io vs. Hassbian. I went the Hassbian route because the granular control it affords you. If you don’t mind tinkering with YAML files, I would go Hassbian.
- Lastly, this can get expensive very fast. The setup you listed is just a start.
Have fun!
The Konnected system is why I am interested in testing HA. Well, that and the fact that Vera has not fixed Geofencing on Android for many months and doesn’t seem to want to.
I have a Konnected system sitting here along with a new SmartThings controller ready to test.
I was just wondering the cheapest and most complete way to go with HA in case I decide to move my current home Z-Wave system from Vera to another controller.
One other question I have. Does HA need the internet to work? Vera is pretty independent but has no built in clock for automation making scheduled tasks not work if the internet is down.
Thanks.
HASS can run completely free of the internet if you wish.
As mentioned further up, would definitely budget something like a USB SSD caddy and a cheapo SSD instead of an SD card. I was encountering read/write errors every 6 months on good quality (Samsung/Sandisk) SD cards; maybe I’m just running too much stuff/logging too much.
Got an USB SSD Caddy and a 128GB SSD for about $40 and its been perfect. Cloned my SD card directly to the SSD and then set the Pi to boot from it.
I’ve been on HA for only about a month but had great luck with the new Zooz Z-Wave Plus S2 USB Stick ZST10, instead of the Aeotec stick. Not much but saved me about $10-$15 on the initial build out.
Do you have some links of the caddy / ssd you got?
keep your vera and use it as a z-wave controller only. Will save you $50 on the z-stick and many prefer to have the z-wave controller separated from the machine running home assistant. Certain benefits including not having to wait 2.5 minutes for the z-wave network to load on every reboot. those minutes add up when you’re doing multiple reboots during a debugging/automation creating session. Having them separate also adds one more level of fall back when HA goes down due to your newest experiment - you can still set the temp on your thermostat and control the lights etc.
If you’re using hassio, or any docker install, you don’t get those waits as you can reboot HA independently of the device and the add-ons keep running.
Best advice to save money is to only automate things that are actually worth automating. People often go mad and try to do everything which will add up stupidly fast, as an example, if you’re going into a room that you only want the lights on when you’re in there and walking past the switch as you come and go, you really don’t need motion sensors and smart bulbs.
Just take it slow and do things you’ll get an immediate long lasting benefit from, others can wait until you really know what you want and then when you find a bargain.
With regards to hardware to run HA on, if you have an old PC laying around or just one that is on all the time, you can put HA on that and have a play for no cost whatsoever, run it as a VM, far easier than you think, and it won’t even disturb what else you have on it. Then, as above, when you’re sure what you want spend some money if you really have to. Given it runs headless, even a laptop with a broken screen would get you going.
2 Likes
I’b be interested in some info, too.