Controlling devices without gateways? - Raspberry Pi

Hello everyone,

I was very close to buying a Google Home Mini and a SmartThings because I wanted to get into home automation, but decided to give openHAB a go when I learned about it in a video. I have a Raspberry Pi that isn’t used for anything so I went along and installed openHAB on it. I should mention that I am completely new to this.

My first goal is to control lighting in our living room, both floor lamps and ceiling mounted lamps connected to existing switches. Would be good if it was possible to use voice control and scenarios, and also a remote control for whenever you don’t want to talk to the GH.

However, it seems there’s not much I can do with the Raspberry itself, and its WiFi? I guess the next step is to buy a USB Z-Wave or ZigBee controller. But it seems like most (all?) the smart bulb require some sort of gateway, this is true in the case of IKEA Trådfri and Philips Hue. I fail to see the point of having a Google Home that talks to the Raspberry which in turn talks to the gateway, which then controls the bulb. The case where it will come in handy is if I want to control built in lights and other devices, I take it? Maybe it would be no different with the SmartThings hub…

Someone please give me your two cents on this topic so I can understand it better, and help me get started. Thank you in advance.

2 cents it is :slight_smile:

Your observations are mostly correct. OpenHAB is Open Home Automation Bus. It is a software which is reliant upon hardware. It is quite simply an automation engine/concentrator.

Hubs like smartThings and Mios/Vera are similar. They are software not unlike OH but on a proprietary piece of hardware and require their cloud services in many cases. The difference is that this hardware comes with some form of external protocol connectivity built in like Z-Wave, Zigbee, Insteon, Bluetooth… etc.

OpenHab comes simply as a software package and you byod.

Hubs are easier to get started as they are all-in one and have more user friendly GUI’s. But if you want a protocol they don’t support out of the box, you are getting bridges for them as well. Their rules processing engines are far more limited IMO.

OH has a literal crap load of bindings to support everything but the kitchen sink. Although, I’m sure you can automate that as well if you want but I digress.

Your pi is a great start. Next you need to pick the protocol(s) that make sense for the devices you want to automate. In your case lighting. I personally suggest Z-Wave but this isn’t the only method. If you go Z-wave, your next step would be to get a Z-Stick (or similar).

Then, I would replace the existing wall switches for those lights with a quality z-wave switch. (aeotech, linear, GE, 2gig, are all good names). If your floor lamps aren’t wall switch controlled, get a lamp module from one of the above mentioned vendors. They plug in-line and give you switching or dimming as you choose.

Once you have that setup, you can start working with bindings, things, sitemaps and rules to get them doing things automatically as you feel fits your desires. For that, read read read https://docs.openhab.org/introduction.html and when you get stuck, read here and ask. You will find the answers to the vast majority of newbie and beyond questions already asked/answered over the years.

Hopefully this helps and welcome to the addicting world of automation!

2 Likes

+1 to everything @moxified said so I’ll take a different tack.

The Home Automation industry is hopelessly fragmented these days. It seems like every vendor has its own protocol and its own little walled garden and there is an uneven overlap of capabilities across them all.

This is where openHAB steps in. It provides a bridge across all its 300+ supported technologies and APIs allowing you to orchestrate your automation regardless of the vendor or technology the device comes from.

However, there is only so much that openHAB can do given the closed nature of some of the technologies and the reliance on specialized wireless communication protocols. So sometimes you need to buy a USB dongle or RPi HAT to communicate with a technology (Zwave, Zigbee, 433MHz, etc) and other times you need to buy a hub (e.g. Hue, Tradfri, etc) and still other times you need to jump through hoops to get a developer key so you can access the technology’s cloud API (NEST, Google Assistant, etc). A lot of what openHAB can and cannot do with a given technology is driven by the features that the vendors expose or that enterprising hackers can reverse engineer.

As a result, Home Automation is very very complicated and frustrating and this is why openHAB itself has such a challenging learning curve. It is also why you will find lots of different ways to do the same thing. Ultimately, each and every individual Home Automation system is a custom one of a kind so you will need to be able to read and adapt generic solutions or parts of examples to your specific situation.

But, as moxified said, openHAB supports the largest collection of technologies of any other home automation system with orders of magnitude more than the commercial hubs like Vera and Wink.

It sounds like you are starting off right. Start small and grow from there. I too started with a couple of lamps and a DIY garage door opener remote. Good luck!

1 Like

Hello again,

thank you for excellent replies guys. Sounds good, I’m sold. It’s really a shame that the industry looks as it does, but it’s understandable, considering it’s quite new and every company wants to grab market shares.

Getting back to the title of the thread: are there actually any good dimmable bulbs that I could control only using the Raspberry Pi with a ZigBee or Z-Wave controller? I should mention that I live in Germany. Or would I be better off just getting an IKEA TRÅDFRI gateway and bulbs for it? I’m not a fan of having a bunch of different proprietary devices that essentially do the same thing, and take up space.

Like I mentioned, I want to use the Google Home Mini for voice control, be able use the existing wall switches, and/or some sort of remote. Eventually I’d also want to have motorized blinds. Is this currently doable with openHAB? Are there any simple remote controls for this purpose?

And lastly, could you recommend a combined Z-Wave/ZigBee USB controller?

Really appreciate all the help!

If you go Zwave anything with a zwave logo on it will work with a couple of exceptions. Only the development branch of the binding supports the security command class so you have to use that for locks. If you have a new device, you can add it to the database and wait a couple days for it to make it into the snapshot build.

Zigbee is a little experimental right now. Its a work in progress. You will need to search the forum for what people have and have not been successful with.One of the problems with Zigbee is there is no guarantee that two Zigbee devices will be compatible with each other. :frowning: But there is a nice dual controller from Linear (Linear HUSBZB-1) that supports both Zigbee and Zwave and both the respecitve OH bindings work with it.

I don’t know the state of this support at this time. There is work to have a Google Home binding but the last I read there was a problem on Google’s end that was preventing it from working. See Google Home - Actions (API) now available! for details. There are other approaches as well. For example:

Switches will largely eliminate smart bulbs as an option. When you flip off the switch the bulb will no longer have power and can’t respond to new commands.

Remotes are no problem. You have wall panels, small little remotes, and smart buttons to choose from there. You also have the option of your phone apps or a wall mounted tablet.

I do recommend though that if you find you have to use your phone to do standard home activities (e.g. turn on/off the lights) you are heading down the wrong path. IMHO the home automation should be as automated as possible and just do the right thing based on events and sensor data and not require a remote or UI to trigger. Obviously, this is not always possible but it should be a goal.

Lots of people do as well. There are some zwave devices but I think most people use KNX for those.

Without further details (the devil is in the details) I see nothing asked for thus far that is impossible.

The Linear device I mentioned above is the only compatible one I know of.

2 Likes

Not to throw yet another direction but that’s the nature of this field as outlined by Rich.

I use a vera edge as my zwave controller and communicate via ethernet and the mios binding. Couple of reasons. I started on a vera and after I became very frustrated with it crashing trying to keep up with my advanced automation scenes and LUA code, I found and decided to migrate to OH. My initial plan was to use a zstick (I even have one new in box I never opened). But I found that the zwave implementation in Vera’s system is IMO superior to that of OH’s at this point. It also supports security devices out of the box where it wasn’t even an option 2 years ago in OH. Also, I use VM’s for my OH instance. Passing usb devices to vm’s is unreliable at best.

Point is, Vera makes the vera plus which supports zwave and zigbee in the same unit. You could put that hub in place and communicate with it over ethernet and the mios binding. Just an option. It probably will ultimately set you back more than the linear or two separate sticks but adds some additional benefits. http://getvera.com/controllers/veraplus/

I’ve been running OH on a VM with the USB controller passed in for about a year.

I had some issues when I first started but learned that that ESXi 6.5 has a buggy USB driver. I learned on this forum how to downgrade that driver (it’s like two commands) and it has been rock solid ever since. The only problem I’ve had is remembering to plug the dongle into the same USB outlet each time I remove it to include a new device. If you were running on ESXi 6.5, see ESXi 6.5 and Zwave Help, in particular @smar’s reply

At some point, I will probably do Share Z-wave dongle over IP (USB over IP using ser2net / socat ) guide with a spare RPi 0W or perhaps that demo unit Banana Pi M2 Zero they sent me to test running OH on. With this I can move the controller around on battery power when necessary and it would be a first step towards experimenting with some failover scenarios. This is also a good approach (or getting a commercial hub like a Vera) if your server is locked in a closet in the basement or otherwise in a bad location to have your USB controller relative to your ZWave/Zigbee devices.

@Qliver, there are lots of options and approaches. The best ones will highly depend on your specific situation.

Totally “ditto” your comment. TOTALLY. Using your phone as a switch is nice but definitely Plan B- for an automation solution.

viz. " I do recommend though that if you find you have to use your phone to do standard home activities (e.g. turn on/off the lights) you are heading down the wrong path. IMHO the home automation should be as automated as possible and just do the right thing based on events and sensor data and not require a remote or UI to trigger. Obviously, this is not always possible but it should be a goal."

Sorry, soapbox warning :wink:

I don’t think this is a fair statement.

This is a totally fair statement because of the IMHO :slight_smile:

If we were strictly talking about automation then yes, your first statement would be correct. Automation by definition is artificial intelligence. His stated requirements more closely mimic a remote control.

I don’t think people are doing it wrong if they want a simple remote system maybe with a little gentle AI peppered in. Maybe they will expand, maybe they wont.

It’s a completely different commitment, aptitude and financial level to be able to yell at siri to turn on a light compared to having sensors hidden (hopefully) around your house making calculated decisions about when to do something.

My parents, for example, liked my automation system and wanted one. I setup a vera to do what they wanted. Quite simply, a programable thermostat and some light timers could do their “automation”, but they wanted to be able remotely control it when away. I would argue they didn’t do it wrong; they did it exactly how they wanted it. They have been very happy with it for the last 2 years. My pop still thinks the best feature is turning lights off on my mother from work.

The point of all this is that the OH community is diverse and full of unlimited levels of talent, commitment, finances, desires etc. OH doesn’t cater only to the hard core automation people and I don’t think it is fair to try to type cast them into our way of thinking.

Don’t get me wrong, I’m not trying to call you wrong or chastise… just pointing out something I feel is an important distinction. I wasn’t going to at first but the ditto pushed me over the edge. :smiley:

You’re an invaluable member of the community so thanks!

I don’t think that is a fair statement. :wink: AI is one approach to automation but it is not even the most commonly used approach. A lot can be very well automated with deterministic state machines and the like. No AI required. In fact I would venture to say that the vast majority of automation, particularly in the home automation space, has no AI involved. I’ve no AI involved in mine. I’ve even turned off the learning features on the Nest.

That is perfectly acceptable and not at all what I’m addressing. You missed a very important qualifier in my first statement: “you have to use your phone to do standard home activities.” Meaning there is no other way to control, automate, or interact with your lights and HVAC then not only are you missing many of the opportunities that home automation supports but you are creating an interface that is MORE awkward than the traditional non-automated interface. I’m not saying nor do I mean that the phone UIs should never be used nor are they always the wrong answer. But, if your phone is the only way to iteract with basic controls (not wall switches, no remotes, no voice control) you might be making things “cool” but it is not really an improvement over a non-automated house for day-to-day use.

That is all I’m saying with both statements.

You would be surprised at how many people set up their home automation where the only way to control it is through their phone.

And you can apply the “IMHO” to any assertion made by any member of any forum unless it is a concrete fact (e.g. you can send a command to all members of a Group by sending a command to that Group). That is all it is.

It was never my intent to typecast anyone nor am I thinking that everyone should do it the same way or to the same level. But I do still maintain that if your home automation approach makes using your home by you and guests MORE difficult than the non-automated way, you should consider alternative approaches. Afterall, I know of few users for whome the ultimate goal of all of this is not to make using their home easier to use and/or enable uses that would not otherwise be possible.

An example may be informative. Take turning in the lights when you get home. I can see five high-level interaction approaches:

  1. Not automated: come in the door and flip the switch on the wall. This is the baseline for which the automated solutions should be as easy or easier.

  2. Voice: come home and shout across the room to turn on the lights. Not really better but no worse overall. It’s a little awkward and may not be suitable in all situations (e.g. there is an infant in the house, don’t wake the baby).

  3. Tags: NFC, RFID type tags or QR codes where you use with your phone to read the tag and it causes the phone to generate some sort of event that OH reacts to turn on the lights. This is starting to become a little more awkward than 1. Warning bells should start to go off and choosing this approach should need further justification.

  4. Fully automated: sensors (could be just Network binding seeing the phone join the network, geo tracking with OwnTracks, IFTTT, or iCloud, hardware sensors like PIRs and BT beacons if you want to spend some money) determine when you get home and turn on the light without being asked. This is even better than 1 as it doesn’t need any separate interaction with a UI to trigger the lights. Just go about your business and the house reacts appropraitely.

  5. Phone app only: come in the door, pull the phone it of my pocket, bring up the app, navigate to the lights, turn on the lights. Ugh. This is way more awkward than 1.

So when I say one is heading in the wrong direction I mean they are heading towards 5. Don’t head towards 5. 5 is a step backwards. Keep the ability to do 1 or add in 2, 3, or 4. That can drive decisions of what to buy. For example, you can keep 1 by using a smart switch instead of a smart bulb and get the ability to remote control and keep 1.

That is all I mean by both statements, though the second statement is more of the goal one should shoot for. In this example, one can achieve 4 (i.e. the light knows to just turn on) with no further investment in hardware sensors and only a small investment in time. And, as I said, it is not always possible to get to this goal nor does it always make sense. But if you end up with 5, you should look for ways to improve the automation.

OK, well, I didn’t want to start something. I just don’t think your meaning is as clear as you think it was.

I totally agree with you here. Again, I just don’t think your two previous back to back statements were as clear about your meaning as this paragraph and in bullet number 5. In any event, the point is now clear one way or another.

I’m talking Artificial intelligence not Artificial learning. By dictionary.com’s definition: “the capacity of a computer to perform operations analogous to learning and decision making in humans,…” your rules using sensors are making decisions as analogous to humans.

Think the mars rover. It has a processor, a boatload of sensors and a gazillion lines of code… it’s considered AI. Your system is basically the same with less sensors (possibly) and far less code. What makes your OH setup not AI and NASA’s rover AI?

From an educational standpoint; I would be curious to know at what point programming crosses from not AI, to AI. It doesn’t appear that there is a quality agreed upon standard. I can think of a few definitions:

Machine needs to be able to write or change its own code
Code allows for variables
Code allows for random selection
Code interacts with data from sensors and does more than just converts format or re-displays it
Must pass Turing test
Combinations of the above…

Since this discussion peaked my interest, I was reading around and found this article: http://www.astronomy.com/news/2016/08/how-does-mars-rover-curiositys-new-ai-system-work. Based on NASA’s explanation of Curiosity’s AEGIS AI system (as they call it) it simply is an algorithm that looks at sensors (mostly pictures it seems). I would argue my automation system and yours, from what I have seen of your stuff around the forum, qualifies as similar type of behavior just with motion detectors, reed switches and other sensors.

In conclusion; since there isn’t a solid definition of what qualifies as crossing into AI… I guess we are both right… IMHO :+1:

You can’t separate the “learning” from the definition. If the system doesn’t both “learn and makes decisions” it is not AI.

I like this quote from the wikipedia article better:

Colloquially, the term “artificial intelligence” is applied when a machine mimics “cognitive” functions that humans associate with other human minds, such as “learning” and “problem solving”

The quote comes from one of the important AI Text Books taught in colleges (it might be out of date by now though and not used much).

To quote this Popular Science article:

but artificial intelligence is now giving rovers and orbiters the ability to collect and analyze science data, then decide what info to send back to Earth, without any human input… The scientists taught AEGIS how to recognize bedrock, which they’re interested in because it contains clues into Mars’ past ability to support life.

It is unfortunate that the astronomy article you looked at did not make that point because it is a key one. It isn’t simply an algorithm, the rover was taught how to recognize bedrock. They didn’t hardcode “if you see this this and this in the picture it is bedrock”. They gave it a bunch of pictures of bedrock and the software learned how to recognize bedrock on its own. That is what makes it AI.

Generally, it is when the behavior you want the software to do is taught to the software rather than hard-coded into it. There are lots of techniques but in general, you do something like the following if you are training the software to recognize something like AEGIS does:

  1. set up the software to be trained
  2. give the software a ton of properly labeled training data, in this case, lots of pictures that are known to and known not to contain images of bedrock
  3. the software learns how to recognize the thing it is being trained to recognize by adjusting internal weights or connections between logical nodes or the like (it depends on what learning approach is used)
  4. when training is over, you present new data to it and it determines if it contains what it was trained to recognize or not

The software developers often have no idea how the software actually works (if that doesn’t cause you even a little bit of concern I don’t know what will). For all intents and purposes, the software is a black box. All the developers can say is it is usually right when presented with pictures of rock, but we can’t even guess how it decides.

The non-AI approach to the following would be:

  1. decode the data into some sort of format that can be processed
  2. write a bunch of feature extraction algorithms to detect lines, shapes, colors, etc and their relationship to each other
  3. write a bunch of conditionals that see what features are in the data and makes a predetermined decision, e.g. if there is are horizontal lines and a tannish color and a flat surface it is bedrock

In this approach, given a picture, the developers can predict 100% what decision the software will make and how it came up with that decision.

From the decision making perspective, it’s artificial intelligence when one, for example, trains the system with the rules and strategies for a game and the software takes that input and learns how to play that game. In Deep Mind the Go-playing AI’s case, it plays thousands of games against itself and “studies” recorded human games by masters. At each step, it develops a better and better way to play the game until it becomes shockingly good. At each stage of the game, it is deciding the best move to make based on this learning. But the developers of Deep Mind have no idea how it works. And the strategy often causes the AI to make moves that no human ever would make that turn out to be unexpected winning strategies. This is how Deep Mind managed to beat the best Go player in the world.

The non-AI approach would be to set up a set of rules that lead to a strategy of “if the board looks like this, put the stone there”. One reason why Chess and Go are such hard problems is the number of possible boards and the number of possible moves is so large even very powerful computers have to take shortcuts rather than explore all the possible moves and this is why the hard-coded strategies like this rarely are as good as expert human players. But, as with the pictures above, the developers of the software can predict with a high degree of accuracy (it depends on how much randomness is thrown into the aforementioned shortcuts) what moves the software will make and can explain exactly how the software came up with that decision.

So why use AI at all? We don’t really understand how it works all the time and we definitely cannot explain how it makes the decisions it makes. We use it because it is SOOOO much better at this stuff than anything that a human can code manually.

Sadly I disagree. At least in the realm of engineering, robotics, computer science, etc there is a real distinction between AI and what we do and it is the learning part. Our Rules don’t learn and make decisions based on what they have learned. We hard code the decisions into them ahead of time. There is no learning.

I can’t speak to the general populous and how AI is used in general as I live in the engineering/computer science world.

OK. Well, this isn’t the place to continue this discussion so I’m going to agree to disagree without any further points so as to close this out.

Guys I feel like this is getting out of hand and going off-topic.

Let me elaborate on what I am planning for my home automation, because I see that there are many speculations. First I want to mention that I am working with industrial automation (PLC programming etc.) so I feel like home automation is something that I really should get into. I think I also have a pretty good idea what is a good solution, and what would be stupid.

I never wanted to control only using a phone, that option has a very low WAF, and frankly doesn’t feel like much of an improvement. I want existing wall switches to work as they did before, but with the additional possibility to use some sort of remote control (like one of those you get with 433 MHz power plugs). It should also be possible to turn on/off one or more lights with voice control (Google Home, already have one). The third option would be to control the home from the phone when I am at work etc. Some lights should also turn on depending on door or motion sensors.

Phase 1:

  • Smart bulb in a tripod floor lamp. Dimmable.
  • Smart wall switch (Z-Wave?) under the existing switch, connected to a ceiling lamp. 3 x dumb LED bulbs hanging from the ceiling. Dimmable.
  • Smart power plug with a LED strip around the projector screen connected to it.

Phase 2:

  • Blinds in the living room which will go down when saying “Movie time”, remote controlled or when it gets dark outdoors.
  • Door sensor for the home entrance door which turns on the hall room lights if it’s dark outdoors.

Phase 3:

  • Change our current radiators to more efficient electrical radiators, and connect them to smart thermostats.
  • Food dispenser for our cat :slight_smile:

It would also be desirable that the system didn’t stop working because of the Internet being down. This is where I would start, but I am sure new ideas would pop up all the time.

The more I read about openHAB the more it sounds like there are many things that are not finished or at an experimental stage, that’s its far from user friendly, and requires a lot of spare time (which is something I don’t have). I’m not saying it’s a bad solution, or too difficult to get into. However, currently I am thinking whether I should get a SmartThings for now, just to quickly get started.

Yup and sorry, I started it and and that’s why I shut it down…

Curious, What do you mean by under? You looking at the micro dimmers that fit in the box with existing switch?

Do you have any desire to change color from OH or just turn them on and off? Dovetail: do you have your theater automated at all? I leverage a logitech harmony hub and some arduino stuff so that lights and equipment go into modes automatically. You could easily control the ir of standard led controllers.

Wow, this is a statement that surprises me. Where do you live? Electricity is one of the most expensive methods to heat, at least in this area.

Lots of quality thermostat options. The zwstat can control electric as well as fuel based systems. You could get one or two to get started with your existing system and just switch them over to electric when you change you systems in the future.

Most of the hubs and certainly OH will cover you there.

Well, I think that is the nature of this type of open project. It will never be “complete” and for that I am grateful. There are lots of parts that are very “complete”. Just avoid the things that are not.

Yeah, it isn’t the most user friendly but you get used to it. As for time consumption. That’s somewhat up to you. Once you get your system setup and running reliably, it doesn’t require much for interaction. The time consumption is often that a lot of us like to extend and tweak. I still spent considerable time on my Vera based system prior to switching to OH.

Starting with a hub isn’t a bad idea at all if you want a quick start but I suspect you will outgrow it. I know you are looking at the smartthings and that’s a fine choice as far as I know. One thing to consider before picking a hub if you go that route is that OH does have a Mios binding which exposes most of a Vera hub to OH. You could use it to migrate or extend your hub over time. I’m not aware of any such integration with smartthings. Just food for thought.

Yeah, I meant one of these: https://www.fibaro.com/en/products/switches/
Lets you use the existing switch and makes it “smart”, also allows dimming. Maybe “behind” would be more accurate.

Nothing is at the moment, and I hadn’t thought of doing anything. But this could be interesting to do at a later stage. I just want to turn the strip on and off.

Yeah, tell me about it… I live in Germany, the apartment building is nearly a 100 years old. No gas nor central heating. We currently have dated storage heaters, and I believe we would be better off with some new electric heaters with smart thermostats.

Thanks for sharing, I’ll have to think about it then. You’re right that I will probably outgrow it, and I expect that solutions like openHAB will continue to develop, and allow you to tailor the setup to your needs, more so than the commercial alternatives. But then I will most likely just need to change the hub.

I recently encountered an MQTT Bridge project on GitHub that bridges from ST to MQTT. OH has good support for MQTT so that could be one way.

Just to expand on this a little bit. If you don’t care about color, you will probably want to use a smart outlet or switch rather than a bulb. The problem with bulbs is if someone physically flips the switch the bulb is completely offline. That can often be a usability problem and often runs counter to your @Qliver’s statement “existing wall switches to work as they did before” . I personally have no smart bulbs for this very reason.

EDIT:
Just one anecdote about ST from another thread. This is a common refrain for people coming from ST. They leave the platform because it is limiting and not very stable. I’ve not seen the same complaints about Vera so if you go with a hub that might be the one to choose.

Dear All,

I’m also new to OpenHab and home automation but this is I believe an important question for any newbies like me. I’m also has concepts and much more questions in my mind regarding the gateways. Maybe this is a fine place to ask help.

First of all let me tell you as this area is fearly new to me please excuse me if I use wrond terms sometimes. As I understand a gateway is typically a USB device we plug in to our Raspberry (for example, or a computer’s USB port) and it will communicate to the smart devices over wireless protocol. We need these devices based on what manufacturer smart bulbs/plugs/whatever we use. So if I have 5 Philips Hue bulbs and 5 ZWawe compatible smart device I may need both dongle connected to my Raspberry. Am I right?
If I see well there are dongles which are able to handle multiple type of smart devices like Linear HUSBZB-1 but basically I have to choose one dongle and pledge myself for that ways.

Question1: I saw a video regarding Raspberry and an 433MHz RF transmitter and receiver. Practically in this video the RF transceiver is teachted to know the wireless signal of ON and OFF command for a smart plug. Isn’t it possible to do the same with all of my smart devices? Just track the signals and if it is done just give out the commands. Or are these smart thing secured against this? Because if you are an attacker you can do the same and get the control of someone’s house with this method.

Here comes my other question in hand: I do not really trust in these wireless things. How secure are they? I do not really have knowledge in this area. What is your experience?

I am just building my flat so I have the possibility to add more wires to my wall. I know its not so dynamic in the future but what about if I just put more tubes for wiring what I want to control, lead all these wiring tubes to a central place then I put a Raspberry there with a few relay boards and maybe some dimming circuits.
I know it’s not so dynamic. It will hard to extend later and I cannot connect every smart device on wire. Also mybe one RPi will not able to handle all wires so more will be needed.

There are pros and contras. Maybe a mixed solution can be optimal.

What do you think about these questions? Are these totally stupid concerns or does it worth to think on the them?

Thank you in advance!

Kind regards,
G

Usually we call these “controllers” or “coordinators” or “dongles”. A “gateway” typically refers to a separate stand along box that OH would interact with over the network. Examples of gateways include the Hue Hub, Wink Hub, Vera, etc.

That is correct. You would need a Zigbee controller (I think Hue is standard enough that a generic Zigbee controller will work) and a Zwave controller, or one of the very few combo USB dongles that have both onboard.

That is what I currently use, though I’ve not deployed any Zigbee devices yet.

You can have as many dongles as you have USB ports. However, at some point the various wireless transmissions in such close proximity will start to interfear with each other.

No because in the case of Zigbee and Zwave it is a whole two-way mesh networking protocol. 433MHz is just single transmit and receive messages sort of like controlling your TV with a remote control. You can’t just record and play back the signals for Zwave and Zigbee. Furthermore, you woul need a transceiver that you can plug into your computer that operates at the Zwave and/or Zigbee frequencies. By the time you have done that you may as well just buy the controller/coordinator.

The are as trustworthy as any other wireless technology. Zwave is perhaps a little more secure than Zigbee, but overall both have their strengths against certain types of attacks and weaknesses against other types of attacks. Though while I have seen some theoretical attacks pulled off in a lab, I’ve not read of any attempts let alone successful attempts done in the real world.

If you are dealing with security and you don’t have good additional mitigations in place I wouldn’t recommend wireless sensors for an alaram system, for example. Though if you do have mitigations in place or you are not in that risky of an area using wireless is no problem whatsoever.

Wires will always be more reliable and overall the devices will be cheaper than wireless ones. If you have the opportunity to run wires in walls, do that. At least do that for security related stuff like window and door sensors. For not so vital controls like lighting it doesn’t matter quite as much.

1 Like

Thank you very much for the fast and very detailed answer!