Use existing hubs for new sensors

Good evening,

I know, it’s s newbie and perhaps stupid question, but I was nto able to find a clear point of view and I’ve still lot of doubts on the topic.

In my home, I’ve hue devices with their internet hub, Tado devices with their internet hub and an Amazon Echo 4th generation. As a matter of facts, I’m technically full of zigbee hubs.

Well, the fact is that I realized that all those devices shares the same protocol only after I started studying and using HA. I guess I could have saved some bucks with the Tado hub (Hue ones came at home before Amazon Echo actually), but…well…that’s life.

ok, the point now is that I want to add some new sensors from Aqara (among others) but I was wondering if I can do that WITHOUT adding their proprietary hub. For the aforementioned reasons, ofc, but also because i don’t want this chineese hub to “phone home”.

Assuming that philips hue and tado won’t allow me to use their zigbee bridge for 3rd party devices, I was wondering if this can be done with Amazon Echo instead. Do I need some special hardware/software to do so? If this is possible, have i to register the device into the Alexa app to be discovered by HA?

tnx for support and have a nice day!

You can ditch both the Hue hub and avoid buying the Aqara hub by using either zha or Zigbee2MQTT with your choice of coordinator (CC2652 based for Z2M, CC2652 or EZSP based for ZHA).

I suspect the Tado stuff has to remain on their hub just because it’s not designed to interact with the devices any other way.

1 Like

Tnx allot, but I’m afraid I’m still a bit lost here. Does it means I need an additional set of hardware (Conbee?) acting as a controller?

Yes, not the ConBee, but yes.

You buy a Zigbee coordinator (typically a USB stick) and connect that to Home Assistant.

OK! I see there are several RPI hat solutions, too. This would end in a better look :slight_smile:

This would work also with homemade sensors (those using for ex. mySensor framework)?

The hat solutions could be a terrible mistake if you use a Pi4 - because USB 3.0 has significant interference issues for Zigbee (and Z-Wave).

It also makes it harder to move away from the Pi to a decent platform.

Home made sensors are unlikely to be using Zigbee, but if they are this will work (assuming you ensure they’re Zigbee 3.0 compliant). Home Assistant does have an integration for MySensors.

Actually, it’s a Pi3 B+.

I like the idea of using Pi because I’ve severa lying around here and because I like playing with the GPIO. For example, i just soldered a very small pcb piloting some RGB leds to reflect the status of HA.

Is it not recommended to run HA with a Pi3?

It’s an ok choice, as long as you understand that SD cards tend to have horribly short lifespans. If you must run it on an SD card get one designed for this kind of use - like the Samsung Endurance or WD Purple ranges, or an industrial card. The Pi3 also tends to suffer from power problems.

Running it off of a USB connected SSD would be better, though you’d want to use one that could be powered externally, or use a powered USB hub to avoid power problems.

If you’ve got something better then I’d always recommend that over any Pi.

Yes, i’ve a samsung endurance somewhere here, I can reuse it I guess.

SSD could be another option, Perhaps I could design something sturdy for the power line. A 12v LED switching power supply and a stepdown should accomplish it. But it’s a bit more time consuming I would say

About the storage, I was wondering if it would be better to mount /config as a samba share and having the persistant storage into the NAS, to having proper redundancy. I don’t know how often HA uses the /config directory, actually. I mean, I see there is a (sqllite) database inside the directory. Perhaps this could led to a bottleneck…

You couldn’t do that with Home Assistant OS, but if you’re happy with Docker just use Container (native Docker) and you can do that.

HA will write there all the time, between the log file, history database, and many other files for internal state tracking. You can of course run an external database for history, such as MariaDB/MySQL, and turn down (or up) the logging.

This sounds very promising!

I would avoid using Docker on RPi, actually. Better a natibe install. Is Home Assistant OS a barebone linux fla

I mean, the yaml files are very small and probably only be accessed during bootstrap. The database is used just for history, so it will be ok to move to an external MariaDB instance elsewhere in my LAN. Internal states are probably better to stay locally, I don’t think it would be critical to lose them due to SD failure.

Thus, I just need to redirect the backup onto an external location acting as a registry, better in the NAS for redundancy.

Yes, I think it could work :slight_smile:

It’s an appliance like setup - you can only run software in the form of add-ons and can’t modify the base OS. If that is enough for you then it’s a good choice.