I Ripped Out a £6k Lighting System | Hacker News Reader

I Ripped Out a £6k Lighting System

robdobson.com

230 pointsiamflimflam1309 comments

I wouldn't buy a house with such a system in it, and I doubt anybody else will, either. Who wants to become an expert on these systems just to turn the lights on? Not a chance.

I considered these systems when my house was built, and am very glad I didn't get them. I turn the switch on, the lights go on, I turn it off, the lights go off. I replace a bulb now and then.

That's it! I'm very happy with it.

But what I did do was run cat5 and RG6 to every room in a star configuration, which has paid off very very nicely. I did all that work myself because the electrician didn't understand words like "crosstalk" and I didn't want to find out after I moved in that the wires in finished walls didn't work.

the electrician didn't understand words like "crosstalk"

Can you explain what the problem is? I also run cat5 (or maybe even cat6, don't remember) to every room, most even multiple times, just put it in walls/cailings/whatever worked and all 40 or so sockets work just fine.

I turn the switch on, the lights go on, I turn it off, the lights go off.

I run a star configuration for switches and light points, where all switches are on 12V DC and use a 'patchboard' to hook them to impulse relays which drive the lights. Still switch on/light on, but you can select which switch does what, easily hook up miltiple switches to multiple light points etc. Not domotica, but still much more dynamic than traditional build-once-change-never.

I'm curious to understand the advantages/benefits of being able to reassign switches/lights. Surely a light/group-of-lights needs a switch/set-of-switches, but once it's done, what's the imperative for change?

but once it's done, what's the imperative for change

If it's truly done there's no reason for change. However getting there, to what for us is the most optimal configuration, is what was the main incentive for me. The way it is now it requires almost no ahead planning. I.e. for most rooms we had no idea on beforehand how we were going to pupulate them with furniture and which activity we'd do where, so I just went 'cable for lighting in the center, at the side on each wall, and some extra' and then put the switches in sensible places i.e. usually next to room entrance. Then during the next years hook up as needed and adapt if needed. E.g. couple of months ago we needed an extra desk for working form home and one extra for my GF's new hobby, so we just put some desks to a wall and instantly can get a light above it. With a fixed configuration you'd either need to think of that in advance, or use the light centrally in the room which makes you sit in your own shadow. or add lights you just plug in the wall socket but that's extra cable, and especially for desks it's hard to find a suitable one. Also throughout the years I've made small changes here and there to change things like being able to turn on the light in the bathroom from the bedroom, or turn on light in the kitchen from the place where we eat so you don't have to look for th kitchen switch in the dark. Sure those are just nice-to-haves but pretty convenient. I've been and lived in quite a lot of houses where the whole configuration was pretty akward. Though I get some people don't care about that.

You get some things wrong when designing a huge project ahead of time.

Crosstalk is when two wires are close together, one induces a voltage in the other (it's how transformers work). The trouble with house wiring is if you run a phone line parallel and close to an A/C line, it will induce a 60 cycle hum in the phone line, which will drive you utterly mad.

The electrician was installing the cat3 phone wires in the same holes as the A/C wires. It was guaranteed to induce a hum.

The correct approach is to keep the low voltage wires as far from the A/C as you can, and if they cross, keep the crossing at 90 degrees. Don't run them in parallel unless they are separated by feet (plural). If you have to, you can run them in grounded metal conduit as a shield.

I probably went overboard on this, but doing it inadequately meant 60 cycle humming on the phone lines and an utterly ghastly bill to rip the walls up and fix it.

He also was installing cat3 (POTS cable) and R59 cable (1970's cable TV stuff). I pulled that all out and threw it in the trash, putting in cat5 (cat5e and cat6 are better, but weren't available then) and quad-shielded RG6 cable.

Getting the best cable available cost maybe 20 or 30 dollars more, so it's crazy not to do it. The network and cable speeds have all increased dramatically since I installed it, and the cables have all worked like a champ.

Even the cat5 cables are happily carrying gigabit Ethernet.

I have a couple plugboards in the basement where all the wires terminate.

I did the whole thing for less than a grand. If an electrician had done it, it likely would have been 10 or 20x more.

Here in The Netherlands and probably other parts of Europe, you are not allowed to run electricity and network or phone cables through the same tubes. Also the number of wires allowed in a single tube is limited.

Many houses have concrete walls, floors and ceilings, so you don't run extra tubes easily...

Reconfiguring switches is next to impossible and via existing switch wires you can dim/switch the lights but not change colour(temperature).

So here the original switches are removed and the connection is bridged (always on) and battery/kinectic powered ZigBee switches control the lights directly, even if the gateway is down.

Conduit isn't very common in US residential construction. Electricians will generally cut a hole in the drywall, drill a hole in the floor plate and tack the cabling to the framing.

In houses than are typically built with a wooden frame. From a European pov, that sounds like a recipe for fire.

Americans love building single-storey detached houses on big lots. And when they build an upper floor, they love complex roof systems where almost every upper floor window has a first floor roof or porch below it.

This means problems like fires spreading to adjacent buildings and fires trapping people where they can't escape are much reduced; you'd never see a repeat of the great fire of London in an American suburb because there's practically a fire break between every single home.

This is made affordable by America's vast tracts of land and high rates of car ownership.

I'm curious how you'd compare the Great Chicago Fire (that happened ~200 years later)?

https://en.wikipedia.org/wiki/Great_Chicago_Fire

The same day, a forest fire ravaged a number of small towns:

https://en.wikipedia.org/wiki/Peshtigo_fire

I think that is very location dependent. Here in the Midwest, Romex still isn’t a thing and most residential construction still requires (effectively or otherwise) 1/2” electrical conduits for all high voltage wiring. Older construction might be the horrible BX/AC instead.

You said Midwest, but are you sure you don't just mean Chicago? Chicagoland is the single place I've heard about residential work requiring conduit.

I had a house with mostly Armored Cable and came to appreciate it. Especially after I opened up a floor and found an exposed conductor on some Non-Metallic that had been chewed by a squirrel.

NM wiring is extremely common in every state (except for Illinois) and NM is allowed without restrictions in multi-family stick-built housing up to four stories. You must live in Chicago or some other restrictive AHJ.

There are still existing functional knob and tube wiring systems that are 100 years old

> (except for Illinois)

Guess where I am ;-)

(Not in Chicago/Cook, though.)

I still have some knob and tube wiring in my home (which is 96 years old). I've ripped out most of it over the years, but it is amazing how reliable it actually is.

Were you able to do that without digging into the walls? My house had a bathroom and the kitchen upgraded by the previous owner and we did the basement but most of the house remains on knob-and-tube.

I've gradually added GFCI outlets to each circuit. They're not grounded, unfortunately, but they do trip if something were to go wrong. And it allows us to plug in 3-prong plugs without using those horrible 2-to-3 adapters.

We had to change home insurance companies because Amica blacklisted us the instant they heard there was knob-and-tube wiring. A few other companies have the same approach. But it's not like homes are spontaneously combusting everywhere because of it.

We ripped open the walls. This was done in the context of a larger remodel and then a rebuild of the back of the house after a fire.

Even if you aren’t using conduit you cannot use the same pathways for line-voltage and low-voltage wiring. You need to drill separate holes thru the studs to run NM in one and Cat6 in the other. You also need a divider in a multi-gang box if it contains line-voltage and low-voltage wiring.

Source: US National Electrical Code

> Here in The Netherlands and probably other parts of Europe, you are not allowed to run electricity and network or phone cables through the same tubes.

Presumably that is not because of interference concerns but because of the danger of mixing high voltage and low voltage circuits in the same conduit, right?

As others have noted, conduit isn't usually used in houses in the west.

Yes, it is for safety reasons.

Sorry should have been more clear; I'm an eletronics engineer so I know what crosstalk is, it's just that I didn't understand how this would be an actual problem with CAT5. Phone line, that's something else though.

If an electrician had done it, it likely would have been 10 or 20x more.

Hmm, that is like twice the amount one would typcially pay here. Still, don't froget the working hours you put in there :) plus the fact that usually technicians get their materials for cheaper than you do (also because in my case it's only 6% VAT instead of 21%). I did many such calculations when renovating and for some things it was actually cheaper to get a technician.

They're maybe getting it for cheaper than you do, but if you think that's what they're actually charging you, you're deluded.

The typical electrician gets it indeed for cheaper than market rate, then charges it to you retail rate plus 20% to account for the "trouble" of sourcing and purchasing it.

if you think that's what they're actually charging you,

I never claimed that.

The typical electrician

Maybe you rather mean 'the typical electrician I met' or so because I know for a fact the rest of that sentence does not apply to the ones I met myself.

From my experience in trades, you just double the price of all materials.. which conveniently is roughly the same price homeowners will see if they spot-check at Home Depot.

Indeed, UTP5 is more or less crosstalk immune within for practical uses, STP5+ moreover so.

That's an easy one. Switch on the light, network disconnects. We had this happen at our office. There's obviously much more happening on an electrical network other than 50/60 Hz, like transient response to switching loads on and off, RF noise from spark loads, dimmers etc.

In digital cinema the projector and the playback server does a handshake and key exchange before you can play encrypted content.The rest of this story might not be entirely correct, I've been out of the business for a few years and this happened maybe 5-6 years ago, but it's a bit fun and kind of relevant.

This usually works without any issues, but on one type of projector they had used a non-shielded network cable internally in the projector and if the projector lamp was lit juuuust at the right moment, the handshake would fail and the film would not play. Normally it wouldn't be a huge deal as the handshake happens just before the film and the lamp is usually lit on the beginning of the show long before the film starts. HOWEVER, there was this one cinema chain that had standardized their playlists in such a way that they for some stupid reason turned of the projector lamp in between the trailers and the film (I think to "introduce" the film) and then turned it on again just before the film with the result that the handshake failed 3 out of 10 times or so.

Long term solution was to replace the shoddy cable. Short term solution was to change all their playlist templates to blind the lamp instead of turning it off. Like they should have done to begin with.

Never had this myself for network, but I can imagine if the wires are close enough together and the light switch sparks enough (or draws enough power) it could happen. Actually this is our 'standard' way of intentionally messing up serial or USB communication: place wire next to the serial line, then try to generate as much transients as possible to induce errors in the bit stream. Still doesn't always work.

> technicians get their materials for cheaper than you do

Sure, they buy materials wholesale, but they bill you full retail for it. I installed around a mile of wiring :-)

But I also got it exactly the way I wanted, the wires run where I wanted, and I bought professional quality tools to do the work, and still have/use them. I like working with my hands now and then, it's viscerally satisfying in a way a keyboard never is.

> I didn't understand how this would be an actual problem with CAT5

I know that twisted pair is supposed to be resistant to that, but I wasn't going to risk it and didn't have a good way to test it. I didn't want even a teensy bit of hum, and there isn't any :-)

Although that hum is annoying on analog signals, it has no effect on digital high speed communications.

Still I agree it isn’t worth the headache of issues to save $20 and it’s more futureproof to use better cables.

It has no effect unless it lowers the SNR too much and errors can't be corrected. Try wrapping RG6 a few times around a thick, active AC line. I saw that a few times back when I was a field tech.

Just out of curiosity, would shielded cat-cables help in a situation where you have it next to A/C wires?

It would have to be a very noisy line and probably not the average house ac - this is from what my CCNA instructor commented.

They also said that stp is a lot harder and expensive to install.

> if you run a phone line parallel and close to an A/C line, it will induce a 60 cycle hum in the phone line

Isn't twisting the pairs supposed to prevent that? Isn't that what makes it feasible to run telephone lines on the same poles as electricity, thus allowing the POTS system to be practical?

Yes, it is. Also true for data. The problem is that common mode rejection is limited for unshielded pairs, so it works - but not perfectly.

POTS is legacy now, so I'm not sure how much of a consideration it is for modern builds.

There are safety issues with running network and power cables close to each other, but generally on small domestic systems crosstalk between data and power shouldn't be a serious problem.

The exception is if you have a lot of hash on the power lines from USB chargers, unfiltered noisy motors, or white goods that generate both digital and analog noise.

Ferrites or other kinds of filtering can help. So can using shielded cat cable.

I would seriously question the overall competence of any electrician who ran any low voltage cable in direct contact with AC power cables. This practice is not permitted by the electrical code. An electrician who does not respect that code requirement might be cutting corners with the code in other areas, too.

While the level of knowledge among electricians regarding network wiring is gradually increasing, the conventional wisdom for anyone who wants network wiring installed professionally is to hire a contractor that specializes in low voltage wiring.

This was 20 years ago and the general contractor thought I was nuts to pre-install all this cabling. This all changed a year later, and network cables pre-installed became normal.

40 sockets is pretty impressive. Running cat-something (I got CAT6) everywhere makes absolute sense though! Especially now that everyone and their dog is working from home, and I live in a typical nineteen-thirties Dutch street with terrace housing, not having to rely on wifi too much is pretty awesome.

I think by now I've got rid of all the R59 cable sockets and POTS stuff too except for one POTS socket in the cabinet that holds the fuse box. If anyone ever wants to reintroduce a wired telephone they can just reuse a CAT6 cable, but I doubt this will happen.

I didn't put any POTS cable in the house, but ran two cat5 cables to each room. One for POTS, the other for network. The configuration is done at the star hub. Cat5 is just fine for POTS.

These days, of course, I don't use the POTS much.

When I started with that idea a lot of people considered it crazy. But for some jobs I've had the trouble of not having a network in some places was so annoying that I thought: not in my place. For me it totally paid off, it's just convenient to have a proper connection everywhere. Also because we were living in the house during renovations so moved desk multiple times. Granted there are sockets which I haven't actually used yet, but there's equal amounts for which at one point I thought 'damn, so convenient that this socket is here' :). E.g. FM reception here sucks for some stations, now I just carry a little RaspberryPi+speakers mounted on a piece of wood around as 'radio' and can just plug it in in any room. Also if I see the mess some people go through in a brand new house, dealing with wifi repeaters on all floors and sometimes long cables to feed them just to get a signal which is still worse than a fixed line, I think this way is more convenient.

Could you explain what the star configuration is please? I'm not familiar with it and am currently assuming that just means cables to each wall from the ceiling.

Star network topology:

https://en.wikipedia.org/wiki/Network_topology#Star

Each endpoint connects directly to the hub, rather than as a ring, daisy-chain, tree, or mesh.

The easy solution is to have the intelligence in the lightbulbs the way Philips, Osram and IKEA did on their Zigbee based systems.

I have all lights (100+) in the house on Zigbee but without changing any wiring. Only non standard is that I replaced all the wall switches with a variant that's electrically "always on" and sends a Zigbee command when pressed (and can be pulled out to access a traditional physical switch as a backup in case of full system failure). They fit the original Siemens outer wall plates that my house was built with so everything looks standard. Returning evening to non-smart in case I want to sell the place is an afternoon of work replacing the switches behind the plastic face plate with their old versions. Those are all in a box in the attic :-)

The benefits of smart lights for comfort are pretty big, auto-on lights by motion detection is great for bathrooms, storage rooms etc. Having the color temperature match outside and time of day is also much better than traditional dimmers. Voice control is useful in some situations (mostly when cooking or carrying kids or groceries). And being able to walk out the front door saying "Alexa, turn off all lights" is a great way to save energy and make things easy.

What Zigbee switches are you using?

not GP, but I use Xiaomi 2-gang switches, IKEA 5 button remotes, the old IKEA dimmers, and their on/off remote. At the moment I'm eyeing LIDLs remotes as well... In Denmark

Everything is controlled via Homeassistant+Deconz and a Phoscon Zigbee Dongle.

It's even possible to hack the IKEA switches into the danish wall switches.

Here friends of hue Gira and Ikea 5 button remotes mounted on a blindplate.

Still wondering what @t0mas88 is using...

The upcoming Lidl remote is very interesting indeed, because it has 4 buttons which could be wired to a regular 2 button up/down push switch (e.g. Gira 014700) as cheaper friends of hue alternative.

The Gira here as well, and left physical switches behind them. So in an emergency I can pull them off the wall push the switch that's behind to turn off power.

Also 3d printed a holder for a few places that slots into a Gira frame and holds the Philips Hue Tap button. Allows more control of scenes etc, but the color isn't exactly the same.

Maybe this is used? US (and Canada?) only though...

https://www.lutron.com/en-US/products/pages/standalonecontro...

Also found this : https://www.samotech.co.uk/product-category/samotech-adapter...

Made for UK wall switches.

> I have all lights (100+)

Your house is massive or your castle is very tiny. Either way, that’s an usual amount of lightbulbs.

Easy to reach if you use spot lighting. I counted a bit less than 30 in a (tiny for American standards) 100sqm house.

I just counted my Hue lights in the apartment I live in and got up to 30, 90sqm.

I have some bulbs, 4 spot lights, lightstrips around the living room. Lamps around the studio, bedrooms, living room. If you add bias lighting and indirect light by bouncing off the ceiling you add up quite fast.

Granted, I built up this setup over some time, finding the best use for each set of lights I got but I can definitely see how you could get up to 100+ in a larger house.

Well, the Hue bulbs and most of the competition are not very bright.

Pretty normal house, just under 200 square meter. But for example the kitchen alone has 2 ceiling lamps with 4 spots each, 6 lamps for countertop lighting and 2 in a wall of cabinets. So that's already 16 lamps in only the kitchen. It's more in the living room.

Even a bathroom easily has 6 or 8 in the ceiling, a few near the mirror etc.

I tried the IKEA bulbs. They have coil whine turned off any made my unable to fall asleep. Apparently it's a known issue: https://i.reddit.com/r/tradfri/comments/arlyiz/coil_whine_an...

At the moment I have standalone lamps with Sonoff D1 dimmer and halogen (yes, really) bulbs that can actually be dimmed 0-100%.

I outfitted my apartment with ikea lights for the last few years, and I am currently in the middle of replacing them with philips Hue.

It is insane how unreliable the ikea products have been and how terrible the interface is vs philips.

I am using both, the bulbs are equally reliable here, the Ikea bridge & app have had their problems in the past, but also work well atm.

I am moving to Deconz (Phoscon) for full flexibility though..

The Philips bulbs are much better. Ikea also doesn't dim as low as the Philips Hue. But the Ikea bulbs are cheaper so I still use those for storage rooms where the quality is less important.

Please tell me what switches you are using. Are you in Europe or in the US?

Can you please provide a name /brand of the switches you use?

Friends of Hue (Gira and others) with EnOcean comes close, but you'll typically bridge the switch AC wires, so there is nothing to pull out and manually control things as backup.

Lutron used to make an amazing switch that worked with Hue. It was the Lutron Connected Bulb Remote. You hot-wire the switches behind the wall plate, and screw in a panel that sits right behind the rocker hole. The remote then slots into that. They're great, and when the battery dies you can just pop them out and change it without taking the wallplate off.

If they were held in by magnets then I'd call them the perfect switch for the Hue series.

Unfortunately they discontinued them, and on the used market they're often >$100 per switch.

I use the Friends of Hue but left physical switches (in the on position) under them. So if you pull them open you can turn off the power in an emergency.

And 3d printed a holder for a few rooms that fits the Gira frame and holds a Hue button. Works to control more things (and used it that way in my old house), but the color of Hue isn't exactly the same as the frame.

Out of interest, what light switches did you choose?

Are the originals German style modular switches (e.g. https://new.siemens.com/global/en/products/energy/low-voltag...)? If so, what did you replace them with as I haven't seen any ZigBee switches that fit those.

> I did all that work myself because the electrician didn't understand words like "crosstalk" and I didn't want to find out after I moved in that the wires in finished walls didn't work.

You hired a bad electrician. Any good electrician knows you can’t run power and data in the same conduit or pathway. NEC requires a divider in any junction box that contains low-voltage and line-voltage power, and also requires separate conduits/pathways for low-voltage and line-voltage power (can’t use the same stud hole to feed both through)

Spend more money on the electrician, and they’ll be better at their jobs.

Edit: I run electrical work

> Any good electrician knows you can’t run power and data in the same conduit or pathway. NEC requires a divider in any junction box that contains low-voltage and line-voltage power, and also requires separate conduits/pathways for low-voltage and line-voltage power (can’t use the same stud hole to feed both through)

Sure, but aren't those rules for safety purposes and have nothing to do with crosstalk/interference?

Yes. My point is that an electrician who ignores (or isn’t aware of) NEC rules isn’t anyone I’d let near my home’s electrical wiring. Almost all NEC rules are for safety.

More than that, there are laws (at least in Australia) about how data cables must be run.

Similar to how close an electrical outlet can be to a sink, there are rules about how close a data cable can be to a current-carrying cable.

In the UK CBus have CAT5 cable that's rates for installation alongside mains cables (back of lightswitches etc) https://www.cbusdirect.co.uk/cbus-product-range/cbus-cable-a...

The low voltage in the house is completely separated. The wall outlets are widely spaced. It even comes into the house in a different location and a different trench.

Yeah the guy in the article went farther than I would. I have a lot of connected switches but they are still functional as dumb switches. If you moved into my house they'd all work like normal, you might notice they didn't feel exactly the same as an old-fashioned switch but when you push up they go on, down they go off. I'd of course let the buyer know they had options to connect them, but they could just forget about that and go on with their life.

Smart lighting only makes sense in niche cases. I use the same Z-wave switches in my 60s-era house for a basement guest bedroom where the light switch is bizarrely in a storage closet outside the bedroom. I have another Z-wave switch to turn on an existing front porch light at dusk. This spared me needing to install a separate photocell.

The main value for smart home is in monitoring. I mostly use it for flood and door sensors without having to pay for a Ring/ADT subscription.

About 70% of my lights are smart lights at this point, and I disagree.

They are super useful for things like adding "sunrise" light in a bedroom with few windows/north facing, great for situations where you have a non ideal arrangement of light switches (one of my living room lights is on the switch circuit that my entry stairwell light is on). I also am very affected by light color temperature, so being able to easily transition lights from cool white to warm at night is very helpful for my sleep quality.

> I considered these systems when my house was built, and am very glad I didn't get them.

You don’t need an elaborate, complex system to get a pretty big benefit from a little automation.

I just built a home, and bought and had the electrician install Lutron Caseta light switches for most of the switches (just the ones we might want to automate, not switches like closets or bathrooms).

Doing this has allowed us to even just turn off lights that were left on without having to get up, or turn on some exterior lighting at sunset/off at sunrise.

Pretty easy to set up, just works, and no overhead/undoing if you want to just use them as normal switches.

Around 2010 I designed the "computer room" for a 3 story apartment building that had computers for the residents to use. It was all very basic. The only complexity was that the ethernet run was across the whole building and 3 stories from the 3rd story computer room to the first story utility room.

The day came when it was time to test it out. All the Ethernet lights came on, but throughput was on the order of less than a few kbps. I let the project manager know that something was wrong with the cabling.

He got back to me a week later. The electrician who ran the line ran out mid way and used wire nuts to splice another length of cable. The fix was to simply terminate each end correctly on a splice connector. IIRC they might have even just re-ran one single length.

But the instinct not to trust electricians to run CAT5/6 was 101% correct!

They did use the correct wiring standard T-568B ? and the runs where < 90m

> and the runs where < 90m

What happens if you somewhat exceed the 100m limit? The only sources I can find boil down to just don't do that but I can't find a chart showing what happens to performance/frame loss.

Would it help to have a heavily shielded cable?

It may work if you run it over RG6 coax.

10BASE2 uses Rg58 :-) not even sure if you can buy adaptors that could use Rg6

It depends a lot on cable quality, what you have on either side and the connectors in between.

At work we had to repurpose a pair 170ish meter cat 6 cables for some cameras and it has worked nicely for three years (with PoE even). We tried again with a 100m cable rated for exterior (plus female + patch cords) and it wouldn't even link at 100mbps

It helps to have a good combination of hardware on both ends. Some switches happily do 200-300m but others don't.

The 100m eithernet distance restriction used to be to ensure that if two machines tried to start talking on the same line, they would hear the noise from the other before they finished sending their packet, and could therefore know there was a transmission error.

That is, at least, what was described to me in Computer Networks I back in 2006. I'm sure things have changed as we've moved to always using full-duplex wiring and fully-switched networks.

It is really about the modulation scheme limits (distance v noise). Most ethernet networks have been full-duplex on an individual link level since the early 2000s.

> What happens if you somewhat exceed the 100m limit? The only sources I can find boil down to just don't do that but I can't find a chart showing what happens to performance/frame loss.

It depends on your system. If you're running old school shared medium Ethernet (with hubs, not switches, for cat5), or even just half-duplex mode, if your run length is too long, the big issue is that you've lost guarantees about collision detection; in that case, receivers may see collisions that senders don't, and you're left with transport layer resends (if any) instead of Ethernet layer resends. This is because of transmission speed, minimum packet lengths, and the length of the cable mean that one sender could finish transmission before seeing the second sender's preamble.

If you're running full-duplex dedicated medium, collision detection is a non-issue, there are no collisions. GigE transmits and receives on all four pairs simultaneously and uses echo cancellation to remove the echo'd send that comes back from the far end, the extra distance will make the echo come back later than expected, and that might be outside the ability of the network cards' signal processing to adjust.

Assuming that's not a problem, you would have more resistance, and probably more? capacitance, which may make signal recovery more challenging. 100BaseTx would probably be more likely to work than GigE, in my opinion, but I'd certainly try both. Ethernet autonegotiation runs on a single pair at 1Mbps, so it's quite possible that the cards will negotiate to 1G, and then not be able to actually communicate at that speed because of the wiring parameters.

One important thing to consider is that Ethernet wiring specifications are such that minimum standard wiring will work at a very low error rate at maximum length when in a bundle with many other ethernet cables/other low voltage signaling cables. If you don't have a lot of bundled cables, you have a bit more wiggle room. Of course, if you run the wires parallel and close to AC wires, you have a lot less wiggle room.

Oh, and one more thing. If there's a midpoint somewhat accessible, there are two-port PoE powered switches you can drop in the middle of the span; some of them even will chain PoE; I think i've seen something that would let you put in three of the chainable switches between the power source and your endpoint (you could probably chain a couple more if the endpoint is also a powered switch).

It would certainly make network engineers cry, but you have to work with the site and wiring you have, and only sometimes is it easy or convenient to run new cabling.

The same goes for telco technicians. Just as I've caught electricians wire-nutting UTP cables I've also caught telco technicians putting "Scotch Lock" splice connectors on cables to "extend" them.

I was confused as all get-out about what a "wire nut" is, turns out Marettes are not what everyone calls 'em!

without looking it up, sounds like you are referring the brand name or the manufacture/distributor of wire nuts available at your local hardware store.

It's actually based on the name of the inventor, a Canadian, and is commonly used in Canada as a generic name.

Electricians are bad at low voltage wiring. There’s a different contractor license for low voltage work in many states.

Electricians also almost always use the wrong colors when wiring speakers with something like a 16/4. I have ran into this a lot.

Spot on with the colors. Why don't they pay more attention to that?

My dad’s an electrician. I worked for him once upon a time and later I did low voltage electrical construction. He says that the standards change too often, and it’s different enough to be considered a totally different field.

When I went to do low voltage later I realized what he meant. Stereo and security installations just isn’t something you want your electrician, whose knowledge mostly revolves around huge, 1000+ amp panels, to be doing. There’s way more money in commercial installs, so most electricians only do residential work to tide them over.

My dad doesn’t even want to terminate anything more complicated than 4 conductor phone wire, and would pull me into a job just for that reason.

I did the same thing twenty years ago, when my house was built. I ran 2 CAT5 and 2 RG6 to every corner of every room. Some areas got more. For example the garage has 12 CAT5 feeds distributed between three walls. Everything goes to a standard 19 inch EIA rack in the garage and is terminated and labelled on patch panels.

Rather than bring out the connections in each room, I use a toner to find the feeds and only bring out what I need when I need it.

I stayed clear of smart appliances for a long time for the reasons I'm sure we're all well aware of. However, recently we had our central heating boiler changed and my partner was very keen on having Hive. It's almost useless. I can't believe how incompetently it seems to have been designed.

Essentially there are three parts, a receiver connected to the boiler, a network bride connected to the router and the thermostat connected to both.

Within the first week of having it installed we had times when we realised the heating hadn't come on as scheduled. Checking the display on the thermostat would show no signal. In effect the receiver on the boiler had disconnected from the thermostat and as such you have no heating. After several occurrences of this I broke down and moved the thermostat to the closest reasonable place I could.

The thermostat is now in the worst place in the house for taking a temperature reading but at least it vaguely heats the house up when I expect it to. However, it is now almost constantly disconnected from the network bridge which enables the "smart" part. The solution to this if I can ever be bothered is to run a network cable from the router to the same room as the thermostat.

Why it was designed like this is beyond me. Why not have the schedule stored on the receiver, at least in memory. That way if it lost contact with the thermostat at least it could put the heating on full during the schedule.

Why make the wireless bridge a separate part? If I could add my WiFi details to the thermostat I guarantee it would have a reasonable connection.

I have to say my experience with Hive has been nothing but exceptional. All the way from having it installed back in 2016 (when we moved in) to this day when it came on for an hour this morning. The controller is right next to the boiler in the eaves of my house and is connected via ethernet through a TP-Link TL-PA9020P powerline adaptor. In all fairness the thermostat is very close to the receiver, in the landing, so maybe that's why it's never been an issue. I don't think I've ever once had a single issue with using the app, whether that's from downstairs or from the other side of the Atlantic. It will be the first thing on the shopping list when/if we ever move.

Thanks for the hint on ethernet over power. I can plug one in reasonably close to the thermostat and another close to the router and they'd be on the same ring main so should work pretty well.

EoP has huge RFI issues, it basically turns your house wiring into a giant antenna broadcasting broadband noise.

Alas, if it makes my heating work the way it is supposed to then it's a price I'm willing to pay.

Spectrum is a shared resource, it's a cost we all pay.

Part 15 doesn't give you a free pass to blast RFI everywhere. If you're over the legal limit and found to be broadcasting you will be asked to stop using it.

I've had a Hive since, I think, late 2013 and my experience is a little closer to the GP's. The Hive hub crashes and needs a hard reboot on a very regular basis, at least once per month, sometimes 4 times a week.

Do you think your partner will opt for the less techie option next time or do they incurable gadget lust (like many of us have to fight within ourselves)

She is currently slightly obsessed by the ring doorbell. I feel her techie lust is incurable.

Not all smart systems are terrible, I've the Honeywell evohome system and it has been flawless for the 18 months or so that it's been installed, bit it looks like a different strategy was used - the main thermostat/bridge/controller is linked via wifi, but everything else is run over a different RF standard, so all the TRV's, thermostats and the boiler controller talk to the main controller directly (iirc it's in the 900MHz range)

The main controller runs the scheduler and is the only device which reaches out to the internet (to sync with honeywell's cloud systems so you can control heating etc from an app remotely when away from home).

Other than the interface on the controller being a bit clunky, I've no complaints, even the batteries in the TRVs seem to be lasting well.

I have installed a Tado smart thermostat and it's nothing short of phenomenal.

I've done a setup once and now it just runs automatically. Can't remember the last time I've even touched the thermostat.

It shuts down when I leave home, it heats up when I'm on my way home. When it's a sunny day it let's the sun heat up the house in stead of using the heating.

It's almost recouped it's costs in the first year by just gas savings (as opposed to my time based thermostat).

Good grief... who would sign up for such a thing? The biggest lighting system I worked on was Philips Dynalite. It's a gigantic RS-485 bus - you just make sure all the dimmers and light switch panels are on the segment. It can be over a kilometre long and you can control/reprogram everything from everywhere on that bus. The protocol is simple enough that you can bit-bang subsets of it from dumb devices. Do recommend.

Simple engineering rule applies here... “if you can run a wire, run a wire”.

Notice over the whole article he talks about the how, but not really the why?

I have a wifi access point in the house to supplement the lan, and so I don't have to use the data plan for my phone.

The durned thing needs to be power cycled now and then. No way I'm putting the lights on wifi. The wired lan is 10x more reliable, but even it needs the router rebooted. (I've had lots of routers. All needed rebooting.) Never had to reboot the lights or the power sockets.

Do you think there's something inherent about forwarding packets or beaming WiFi that make devices unreliable or have you considered the possibility that purchasing a more expensive device might resolve these issues for you?

MikroTik routers a good for the price, WiFi is decent. Ubiquity APs are good for their price, WiFi is about as good as it gets. HP ProCurve switches has been to space.

You could use any device that runs Linux to DYI these things. Though you need one 2ghz and one 5ghz NIC in the device acting AP.

Since i started working IT 5 years ago and got some gear from work I've never had to reboot my device because it stopped functioning though I have scheduled automatic software updates checking (and installing if available) every day at 5am which implies automatic reboots every once in awhile.

IMO x86(_64) machines running a standard Linux distro with ath9k/ath10k cards is the way to go. Embedded access points are crappy, and loading OpenWRT on them just makes them crappy in different ways. I'm sure enterprise gear is better than the consumer stuff on average, but it's still painting myself into a corner.

In a similar vein, it pains me to hear people complain about SD issues with Raspberry Pis. My Kodi terminal will die occasionally due to SD card issues, and the only local disk usage it does is booting and the occasional dist-upgrade. The things are meant to be cheap and accessible, not reliable. If you're setting up a server in which you will be investing your time, an RPi is the wrong tool for the job.

The filesystem (ext4) should handle power failures just fine, so a more expensive SD card might do you well. There's a reason some are 5x the price of the cheap ones.

A better SD card would be more expensive. But that doesn't mean if I buy a more expensive SD card it will be better. I've already been using name brand consumer ones like Samsung EVO+. And I don't have the patience to dig through spec sheets on Digikey to see which ones purport to handle power loss gracefully and then do web research to find out if they actually live up to that claim.

My next use of RPis (security cameras) will just use SD to load the kernel/initrd, and then run from nfsroot.

BTW this hasn't been a once a month problem, maybe once a year. But once a year is still too often for me, as I would assume it would be for anyone who relied on one to control their lights.

I've bought multiple access points, they all required rebooting. My HD TV requires rebooting once every couple weeks or so. Also my Roku box. Also the cable modem. My Grace Digital media server. Every damned thing I own that has a computer in it needs rebooting now and then.

Including my Windows and Linux boxen. Or they just slowly grind down, likely from not doing a good job cleaning up unused resources.

Stop buying consumer networking devices and the reboot problem goes away. Last I checked my Ubiquiti EdgeRouter X uptime was just over 2 years. 0 issues. Same with the Unifi APs I use in my house.

I have to reboot at least one of our Ubiquiti switches each month on our office network. I'm very skeptical of Ubiquiti hardware. We had similar problems with their access points and swapped to them out for Cisco Merakis. I don't recall ever having to reboot our Meraki access points or firewall.

Is this much different to DMX?

DMX is a one-way single-master bus.

That means you can't have multiple light switches on the same network without quite a bit of complexity, you can't detect what's on the network, you can't update firmware or do device setup over the network, and so on.

In stage applications that's not a problem because all the light switches are in one place - the lighting desk - and the lighting guy there can see all the lights at the same time.

I don't get it. What's the point in these systems? Genuinely. Why would I ever need this? Just like a "smart" doorbell, fridge, meter. I would love for someone to be able to explain the benefits to me over the normal, cheap, completely working "dump" version. The only "smart" device I own is a TV and it is crap, worse thing I've ever bought.

I didn't get it either but now i'm doing even more.

Our Lightsetup works really well and solves 'real' problems. In my corridor, there are 2 motion detectors and they switch those lights. At night less bright; That works flawless!

And that shows how nice it actually can be if it works.

I also use alexa quite often to switch between light settings; From Eating to watching TV to 'energize' where all lights on at 100%.

and it gets more interesting after i hooked up alexa to our blindes. I wanted to control them with a schedule anyway and now i can also use it thourgh alexa.

Honestly, i think alexa is a very interesting playground on how modern/future user interface might/will look like as it feels much more direct, nicer etc.

Heating is also quite interesting: Make sure to only heat in winter when the outdoor temperature is cold enough. Alert when co2 is to high in a room. Sync up the work schedule of my wife to heat the bath room at 4 in the morning for an hour.

What is the problem? The Problem is still complexity, cost and reliability. But this will become better every single day.

For me it started with a single wifi lightbulb.

My wife often gets into bed with the baby. If the baby falls asleep before the light is off, she doesn't want to get up to have to turn the light off since it will wake the baby. Now she can just turn it off from her phone.

Kinda just expanded from there.

Now a good chunk of the regularly used bulbs in the house are "smart".

Basically what it means is that if you turn a light on, it turns on like a normal bulb (or you can do a quick flick off/on again to turn it on at a dimmer brightness) but you can pull your phone out and dim it / change the temperature / etc. Closer to bed time? Easy to dim the lights and make them super warm.

So that right there is already kinda useful for <$5/bulb. My workspace is a normal warm temperature at a reasonable brightness until I'm trying to do some soldering or something then I can instantly make it bright and white.

From there I expanded out into adding a few basic sensors. Combination motion/temperature/humidity/brightness sensors. So now when I get up for a piss in the middle of the night as soon as I walk out of the bedroom the hallway sensor sees that someone's there and that the hallway is pitch black and turns the hall light on as a red light at ~5% brightness.

When you walk into the baby's room in the middle of the night, same deal. Dim red light. If you go in in the daytime, bright warm light.

Did a similar thing with some of the basement too. Now as soon as I walk into one end of the basement to take the dog outside, do laundry, grab a tool, etc... the lights just come on. They stay on for a couple minutes after the last time motion is sensed and then automatically turn back off. Especially since half the basement is on a single light switch, this results in a lot less wasted power.

The lights in my office come on when I walk in if it's dark, and when I unlock my computer it shoots a message out over a MQTT broker which my automation takes as "someone is occupying this space now" and leaves them on. When I lock my computer, two minutes later if no motion is sensed it'll turn them off.

I don't know that I'd ever add 500 different controls to my house (not sure what I'd even use that many for...), but as far as I've taken it so far it's super convenient.

In the future I fully intend to get a smart thermostat (or replace my thermostat with some controlled relays) and use some "smart" air registers so I the temperature sensors I have can create a loop integrating the temperature sensing integrated in the motion/luminance sensors in most rooms with the heating to more precisely control the temperature throughout the house--if one room is already warm enough but the rest of the house is cold, can automatically close the vent to keep the temperature pretty consistent throughout the house.

> The lights in my office come on when I walk in if it's dark, and when I unlock my computer it shoots a message out over a MQTT broker which my automation takes as "someone is occupying this space now" and leaves them on. When I lock my computer, two minutes later if no motion is sensed it'll turn them off.

What did you do to send the message when the computer locks/unlocks? This is an exact use case I could make use of

IOTLink could be used for this: https://iotlink.gitlab.io/

As far as I'm aware it doesn't report locked/unlocked status, but it does report system idle time which you could use for a similar automation.

So, short answer: an MQTT client. On lock/unlock I have a MQTT client called to publish a message to a topic. I think I'm using HiveMQ's mqtt-cli because it was the most immediately obvious option for my OS.

Working out from there:

* I use Windows as the host OS for my desktop. You can use Task Scheduler to set a task to run on lock/unlock. I have jobs set that mute/unmute my audio and call a MQTT client to publish the messages to the broker.

* Those messages are picked up by Node-RED. It's a ladder-logic-ish environment for automation. It has flows that control all the bits of logic based off of all the various data sources including MQTT.

* Node-RED calls Home Assistant, which handles providing a UI as well as actual integration with all the various bits of hardware (light bulbs, z-wave controller, etc)

I quite enjoy this stuff (which is probably half of how I got here), so feel free to ask if you've got follow-up questions.

Indo the same idea but use a ZWave smart plug with power use monitoring with my monitors plugged into that. Watts > some number = I'm working.

Via zWave2MQTT and NodeRed it triggers other logic like throttling RClone uploads.

Note, this is what I use things but I also have dumb switches to control if things fail. I also dont put control in hard to reach places. I just use outdoor sensors + smart bulbs.

Use case 1: Open all lights in yard at night when exterior gate is open or motion detected. Shut off after set amount of time of no activity.

Use case 2: light in shed/basement/garage on when door is open.

Use case 3: control lights via telegram / get notifications as alarm system.

Use case 4: water sensors anywhere you can get a leak. (have been flooded while away)

Use case 5: you dont want to rewire the house but want to link multiple lights together.

Use case 6: you want to remotely enable lights outside from second floor to check what the dog is barking at.

I can go on.

What the guy did wrong was ignore failures and not have a way to deal with them

I have a couple motion sensing lights. They randomly click on and off. P.O.S.

During the day they are mostly useless outdoor.

During the night position them where there's not much infrared pollution.

I used to get random on/off from car lights reflected off home.

I used to get notifications from cats and birds too in the middle of the night.

I got small aqara sensors that you just glue to places and just moved them around until I got good spots.

Also, if they go on they just turn on lights so not much of an issue. Alarm is not driven by them but by smarter cameras.

I have been fortunate to have never been flooded but I'm curious if you got the alarm would the situation turn out much better than if you didn't?

There are motorized whole house water valve shut offs that can be connected to leak sensors. Have a water heater rust out, leak, fill the pan, trip the sensor, shutoff the house water, continue to slowly leak its 30-60 gallons out, but at least it’s not continuous, pressurized leaking until you notice.

Even without an automated valve, you could shut off a manual valve yourself if alerted to a slow leak.

Leak sensors can also help if you have a failed sump pump. My sump pits have a main pump, about 6” higher a second battery-backed pump with a (local) alarm. It’s a good idea, but the battery pump often fails to run when needed because of sitting so long, so the alarm is a wise addition. (If it ever goes off while you’re not testing the system, something needs attention.)

Yes, it really just depends on the circumstances.

At a friend's house at 2am in San Diego, first hard rain in months, outdoor drain that parallel garages cement slopes into was clogged. Water got to be hilariously high, he'd mentioned that before he rented the place his ground level bedroom adjacent to the garage had been remodeled because of a flood. So like, 8 inches high. Anyways, we had thousands of dollars of lithium batteries on the garage floor... shit got dangerously close, and all it took was clearing some leaves out of the drain to fix. If nobody would've happened to be there though, would've been quite the bad situation.

That's a bit not answering your question tho - with regards to alarms tripping, know somebody with a warehouse in Sacramento - owner hadn't been at the shop for a few days, employee had. One night a security alarm trips, he makes the 20 min drive to the shop, turns out the security alarm circuit had shorted from water 4" high water level caused by a paper towel clogging a sink that was left on. Fortunately he had mostly everything on raised shelving.

If you you have a reason to have some alarms, and they go off when you're away, you'll likely be able to say whether or not it's worth it to call emergency services and say literally break my windows and go inside to see what happened. In the case of a burst water pipe or something, they'd be able to shut off your water mains saving you from further damage. That's the jist or so.

Manual shut off water pipes and power outlets if it looks bad. (as others mentioned you can automate this)

Then check source of flooding and try and fix it, see other people's comments.

It's better then to have your home flooded for a long period of time while you're off at work or sleeping or somewhere else.

Also better then having a short / potentially getting electrocuted.

I'm building a stable and veterinary practice building for my wife (who is a horse vet), 22x10 m, 2 floors. She wants several types of lighting in the stable part which should be switchable from several locations. Switching a light source from 2 locations can be done easily, 3 is doable but more than that requires a separate switching circuit with momentary switches and a contactor to do the actual switching. It is here where a system like this, especially a wireless one can help by making it possible to have switches anywhere you want them without having to pull loads of cabling (which by regulation has to be mounted on the walls instead of in them in agricultural buildings). For now I've settled on 2 switching locations for most of the lighting so as to avoid any "smart" solutions, should she insist on having to be able to switch the "mood light" in the stable from the observation window in the office on the 2nd floor I'll have to go that way with all the problems that come with "smart" electronics on the countryside - power failures, lightning, etc.

I'm replying to myself in an effort to reply generally to everyone who replied to me, thanks for your use cases.

I can see some of the reasons and benefits and to me, no offence meant, the seem pretty trivial, especially when you take into account the privacy issues.

I have dumb timers for lights I want on at certain times, dumb movement sensors for outside lights that only turn on at night, my boiler has a timer which I manually set if needed, and I've yet to see any actual benefit to a smart meter. I can read and calculate my usage.

My point is everything I need can be accomplished without network connectivity or providing multi-national tech companies with all my data. I don't need a digital assistant and don't want to talk to my lightbulbs.

As one of the other commenters said, you get all the problems of IT infrastructure for light bulbs and sensors. Hell if I wanted some automated stuff I'd look at a raspberry pi or Arduino and roll my own version (half joking, I'd definitely attempt it).

Regarding privacy issues, IMO if that's a problem you've screwed up big time. Stuff like lights should only use local control and not be tied into the cloud or at least it should be optional, and any lights should fall back to working like normal dumb lights if they lose contact with the hub.

When it comes to rolling your own devices with an Arduino or similar, this is much harder to do safely and in a code-compliant manner than you would imagine. Making your own sensors is fun and harmless, but making stuff that is hard wired into mains is a bad idea. It's not totally impossible to engineer a properly safe device if you really know what you're doing, but it'll probably end up costing more than a commercial device and won't come with the compliance testing marks.

I agree, however... even though I have a lot of effort sunk into my own home automation setup it's not for everybody and there are plenty of downsides.

For commercial systems that will do that, Hue lights work fine on an isolated network, and HomeKit as a broader ecosystem doesn't require an internet connection to work and is explicitly designed to always treat the local hub as the primary controller.

There's plenty of this stuff that's entirely local and has no "privacy issues".

I roll my own version on a NUC with some ESP8266 based sensors, but frankly I don't think many people care if multi-national tech companies know when their lights are on. I don't really, I roll my own more as a hobby.

I've found it to be pretty useful. My partner and I tend to go to bed at different times. I put pressure sensors under the mattress, and whenever one of us goes to bed the subwoofer in the living room turns off. Little things like that are pretty nifty.

>especially when you take into account the privacy issues.

HomeAssistant seems to be the most popular open source "smart devices" manager.

As one of the general use cases, I'll hop in on the privacy note. That is, there are zero privacy issues from my end.

Nothing leaves my house for any of this. I can rip the internet out tomorrow and everything will keep working. None of the devices even have a route to the internet.

The Z-Wave stuff is all Z-Wave+ and is supposedly pretty secure in its own right, but is just general RF and talks to my controller which is plugged into my server via USB. None of this stuff can talk to the internet.

My light bulbs are 802.11, but they connect to a dedicated wireless network which bridges to my "IoT" VLAN.

My IP cams are all wired and connected only to my IoT VLAN.

A VM runs Home Assistant and Node-RED (both open source) which sit on my general LAN as well as the IoT LAN. That provides my interface and controller for all my smart home devices.

Another VM runs Blue Iris to act as my DVR for my IP cameras, do motion detection, etc.

All communication between everything is either done directly or through a MQTT broker running in a container only accessible on a bridge internal to the hypervisor.

All the VMs and containers run on a server running Proxmox sitting in the corner of my basement. The IoT VLAN does not even have a route out to the internet. DNS only resolves a couple internal hosts.

Basically this is "I already had a server running in my basement and I dug out an old piece of MikroTik gear". It's not gonna be simple for a non-technical person, but for most people on HN it's likely not a huge investment of time/money/etc.

There's no need to go roll your own interface for "how to communicate within your house". Z-Wave, 802.11, and ethernet all work perfectly well and provide you lots of great options to work with in existing hardware and existing physical and link layer technologies (cabling, PoE, switches, etc). They don't need to be insecure or privacy-violating unless you let them.

I have added a Sonnoff mini with Tasmota to my light switch in my room. Whenever I cannot see my USB ports on my computer at night I just press a macro button on my keyboard, and without taking off my headphones I can see everything. Kinda dumb, but it makes me happy every-time I don't have to move my ass.

Nevertheless sometimes (once every 6 months maybe?) the Tasmota box looses connection with my WiFi network and I have to power cycle it via the breaker box to reset it.

Turning the lights off without getting out of bed is my #1 use case.

For me it's a set of automations that make some of the minor inconveniences of life just sort of magically resolve themselves:

- Automatically closing the skylight shades after sunset (when they just let ambient city night glow in) and open them just before sunrise.

- Automatically opening the window shades just before sunrise.

- Turning on the overhead lights and increasing brightness in several stages during sunrise (my place can be a bit dark because I'm under the eaves, this helps the effect).

- Automatically turning off several light-producing devices (coffeemaker, Instant Pot, etc) via 'smart outlet' when I go to bed in a certain time window, then turning them back on when any light is turned on in the morning. (This second-order automation means they still have the same 'magical' effect even if I disable the other auto-lighting for a day because I desperately need to sleep in.)

- Automatically playing NPR in several rooms the first time I leave the bedroom on a weekday during a certain time window.

All the automation here happens entirely locally via HomeKit, and the only reason the system even needs an internet connection at all is so the Apple TV acting as a hub can get updated sunrise/sunset times. (For a totally self-contained system you'd probably want to grab a cheap iPad to act as the hub instead, wall-mount it, and put it in 'kiosk mode' locked to the Home app.)

For all of this, my Hue lights/smart plugs/motion sensors and the Apple and Sonos speakers have been completely reliable. The Hunter-Douglas skylight shades have been mostly reliable (maybe one or two failures over the course of a year, though I did have to get an RF repeater for that to work well).

The weak point is the window shades, which use generic rebranded functionality from a reseller because most of the industry of window coverings is awful and behind the times. Hunter-Douglas top-down-bottom-up shades are about the best replacement available at the moment for automation purposes (so you can adjust the exact amount of window covered for privacy purposes and do it multiple times per day in a way that would be a massive pain to handle manually), but awfully pricey.

I'm also considering rigging up a proper bed occupancy sensor (basically turning the whole bed into a giant scale with load sensors and triggering events on human-approximate weight being added/removed) to replace the bedroom motion sensor, but I've got one of those Ikea slat-bottomed bedframes that probably wouldn't work very well for that.

Spend a whole post writing about the mistake it was to invest in an overly complex solution in search of a problem and conclude by doubling down and redoing it all over again but with Wifi instead. Perhaps the problem wasn't with the protocol, but with the fact that something that just works became a months long project worth thousands. What a rabbithole.

The ONLY wireless lighting system I will ever use is Lutron Caseta light switches.

They are magic and don’t require a neutral cable, unlike just about every other smart dimmer, which means they can be swapped in place of literally any light switch in any home.

If the Lutron hub is unplugged, the switches continue to operate as normal light switches.

They also operate as normal light switches at all other times. So no need to teach every guest how to control the lights.

They are HomeKit compatible, so you only have to deal with Lutron’s trash app when pairing new switches with the hub.

100%. I've been using Lutron Caseta switches & hub for over 5 years and have never once had a single issue -- no matter how minor. The dimmers work like normal even if the wifi is down, and they don't need internet in order to work with HomeKit. I've never once had a switch not work or fail to respond with HomeKit. I have friends with Z-Wave/Zigby/Phillips/etc and they are always talking about how they are finicking with Home Assistant or Smart Things, or how the lights all returned to full brightness after a power outage in the middle of the night.

I want to install my switches and have them work, always. If my wife even once complains, that's a total failure. If a guest can even tell that they are anything but normal dimmers, that's a failure. With my Caseta system, my wife has never complained -- only asked me to install more of them.

And h/t to HomeKit and Home.app. I've tried Google's version (Home) and it's absolute garbage in comparison.

ZWave switches and dimmers also continue to work fine without the hub.

I do enjoy Caseta's smooth dimming and "nice touch" switch controls however.

What I'd really like to find is some variety of smart switch with a mode change between 'control the wall circuit' and 'leave the wall circuit always on and send commands over the network', so that the same switch hardware can be used everywhere but you can change the mode for specific circuits that have smart bulbs.

I have no idea how one would design this in an intuitive way, though.

Inovelli (Z-Wave) supports this - https://inovelli.com/

They have a competitor whose name I can't remember that does the same thing.

If you never program the switch, it behaves just like a regular switch. If you program the switch, you can specify if it will trigger the relay, or just send z-wave commands. There's even a physical override switch you can pull out to cut power in case you have it programmed in "soft" mode.

The "red" product line doesn't require a neutral, either.

Works great for smart bulbs as long as you can set the hub up to do the control.

If you have z-wave bulbs (harder to find than zigbee, due to licensing), you can set up the z-wave pairing so that the switch controls the bulbs directly, so theoretically no hub required.

From having just stumbled upon it, it looks like Insteon switches can behave similarly, and have the additional nice touch of syncing the indicator LEDs between multiple switches if you wire/configure them correctly.

Think it’s worth mentioning that the author is in the UK.

Unfortunately Lutron gear just isn’t compatible with our switches. Everything 240V rather than 110V, and to make things even more complicated our standard switch dimensions (and thus standard wall box dimensions) are different to both the US and EU. Meaning nothing physically fits either, unless you replaced every wall box with an imported wall box.

Lutrons stuff looks amazing, it just sucks that it’s completely not compatible with UK homes, or even UK culture. I think a lot of Brits would find Lutron’s designs very jarring.

Wow, this guy is either willfully ignorant of ZWave technical realities, or just doesn't have as much knowledge in the realm as he purports to have.

Basically every gripe listed has to do with using old original ZWave devices. If you started a 150-node mesh today with strictly ZWave+ devices, an Aotec ZWave hub, and perhaps Home Assistant to control and automate it all, you'd have a much more consistently positive experience.

So I'm left boiling down the complaints to "I was an early adopter, so I got stuck with the crappy stuff, boo!"

Also, if you're paying >$50USD per common ZWave fixture (like a dimmer, stuff like pool controllers can be more), you're seriously doing something wrong.

Oh well, to each their own. I have a mesh of about 50 nodes across 2 stories, and it all works great. Can't remember the last time I had to turn a light on when I entered a room. Among many other nice touches, which I am fine expending a bit more effort to upkeep than normal 'analog' dumb homes.

The thing that stuck out to me in that article was complaining about failures.

I would think, if you have 150 components of anything, you're going see failures on a fairly consistent pace, especially as time goes by.

For example, I'm replacing LED lights that were supposed to last for decades every 6 - 18 months in my house, and they aren't even the smart kind.

The one thing I'm not happy about my small z-wave setup is the HomeAssistant upgrade cycle breaking things way too often. It's gotten better in the past year or so, but I'm pretty sure I'm going to see that happen again in the future. Definitely would not recommend HA for anyone who doesn't like tinkering.

At that scale, personally, I'd really want to just be looking at good dimmable LED bulbs, and put any 'smart' hardware purely into the wall switches, so that there are far fewer devices to worry about.

For Home Assistant, look at ZWave2MQTT. ZWave devices use that as their hub, Home Assistant is a client. Now restarting Home Assistant doesn't cause problems. I was ready to ditch ZWave or Home Assistant until I switched.

Due to regional frequencies ZWave products can cost a lot more outside the US. For example here in New Zealand due to the AU/NZ frequency region you'd be lucky to find a basic switch for less than NZ$80, which is around US$55. Typically they're over NZ$100.

On the plus side not using 2.4Ghz means they're a lot more reliable in urban areas.

OOI where are you finding ZWave dimmers for less than $50?

Here in the UK, £50 per device is about the going rate. Finding something for about £40 is pretty good going. But even then just doesn’t compete with a £10 Shelly.

https://inovelli.com/black-series-dimmer-switch-z-wave/

I've also bought Zooz switches, but wouldn't recommend them. Inovelli has had major issues with stock in the past, but otherwise makes a good product.

Ah, that’s the difference. It’s a complete physical switch and dimmer module.

Unfortunately it’s an American style switch with isn’t compatible with UK style switches. Ours are smaller and square, so are our wall boxes, so these simple wouldn’t fit.

I installed an automatic sprinkler system with several "regions". I struggled with that thing for years. Programming it required reading a hefty manual. The user interface was near the worst I'd ever seen (the worst is on my automated cat feeder). I was constantly repairing the lines, the heads, the valves, the control box. I'd winterize it, de-winterize it. The coyotes would chew on it.

Finally I simply turned it off and let whatever would grow without constant watering grow, and the rest died. Makes me a much happier man.

I've looked into automatic sprinkler systems with scheduling based on the weather or moisture sensors, but they to have too many failure modes: moisture sensors corroding, incorrect weather readings, blocked outlets.

I'm now planning to use a simple drip irrigation system. A long hose with nozzles plugged into the mains water that has to be manually turned on. It may not be automatic, but turning things on and off when I need them isn't the problem, like using light switches isn't the problem. There aren't any coyotes around here and the winters are mostly frost free, so I'm hoping this doesn't become a burden like the tech you mention!

You're right, a drip irrigation system is much better. You wouldn't have to run regions, either, just one system.

But watch out for squirrels and other fur-bearin varmints chewing on the pipes. I'd consider copper pipes. I was thinking about digging up my whole sprinkler system and replacing it all with copper pipes, but then just opted to abandon the whole thing.

I couldn’t understand how smart watering systems were so expensive. I bought a TP-Link smart plug and a solenoid that defaults to off. It took 48v so I got a transformer, plugged it into the smart plug and wired up the solenoid. Once this was done it was just a matter of making it smart.

I set it to water for 10 minutes after sunset in Home Assistant.

I’ve always intended it too hold off if it rains, but our weather forecasting is too inaccurate for that so have been messing about with soil moisture sensors, but haven’t had success yet.

This is something that silicon valley gets right. Hardware isn't easy for inexperienced companies and there are a lot of companies out there that make excellent reliable hardware. It's all pointless though if that hardware is hard to use. There's sometimes bad engineering from Silicon Valley hardware startups, but at least the design is usually user oriented and has been iterated on.

The cat feeder invented all their own cute icons, but what those icons are for cannot be divined without the manual. Lose the manual, and you might as well throw the machine away.

It's a freakin' catfeeder, jeez. It should just work.

I have a similar set up using AeoTech Z-Stick on a Raspberry Pi running HomeAssistant with Jasco light switches. I'm ready to toss the whole setup, too.

First off, HomeAssistant is not as easy as people claim. They love to release breaking changes. You have to read the release notes or else a routine pip update will foobar your whole system. In the four years, I've run it, they have 1) Introduced and then retired a recommended install method. 2) Completely redone the web architecture. 3) Completely redone the Z-Wave architecture. 4) Routinely release UI features that are just placeholders and useless. It's completely maddening.

The problems are exasperated by the Z-Wave devices themselves. The light switches may be branded GE or Honeywell, but they are all made by Jasco. The first generation has been complete garbage. I've had a 40% failure rate over these five years for them. They will randomly die, or fry during a power failure. Knock on wood, so far no failures with the 2nd Gen so far.

HomeAssistant has tried to be more end-user friendly and have the UI be able to handle all aspects, but this is limited by the cooperation of the devices themselves. It's like herding cats. "Heal Network" never actually does anything. Try actually deleting a dead node through the UI. You can't. It never actually works. You have to ssh into the box, delete the entry from a JSON file, systemctl restart.

How a non-linux user can manage a HomeAssistant setup is beyond me.

Sounds like you have a similar setup to me. I'm running HA with mostly Jasco switches. It's been okay, but there have been hiccups. Z-Wave devices can be annoying -- ESPECIALLY door locks. I've gotten them all working, but the setup was finicky, mostly due to the security requirements. The only way I deal with HA breaking changes is that I don't update often. About once every couple years I've gone in, dedicated a couple hours on a Saturday to doing a major update. And I keep everything in git so I can roll back if something goes pear shaped.

I'm probably sticking with HA, but my latest batch of light switches have been TP-Link Kasa WiFi switches and I really like them. Very fast to respond. Easier to setup than Z-Wave.

Yeah, I'm probably sticking with HA for the time being mainly because I don't have the time to redo the whole thing right now. It's just frustrating to hear people recommend HA to complete home automation noobs with little or no linux experience.

Related to that, I keep the HA configs routinely backed up. I should probably switch to nightly backups because HA now manipulates those config files when setting changes are done through the UI. Every once in a while, I shutdown the system and completely clone the Pi SD Card image to save elsewhere. I (thankfully) haven't had to restore from this in a while, but before switching to a durable SD card, I had to do it with every power failure.

Backsups, cloning, industrial cards. Again, I have no clue how a non-linux user could even begin to use HA.

I'll checkout the TP-Link switches. Thank you!

Consider their multiplugs too. The plugs are individually addressable.

I end-run around the weird HA z-wave integration by using zwave2mqtt running on a pi with my HA running in a VM on a server in my homelab. I haven’t had nearly as many issues after I switched to this setup and it it running.

What is the point of trying to wirelessly link up a system that needs so much wiring in the first place, including electricity? This seems moronic, unless it’s for fun. And even then it’s pretty stupid.

Retrofitting new wires to every light fixture and switch in a house is unenviable. In some cases, impossible. Laminate floors can deprive you of subfloor access of a whole floor, running cables through existing stud walls is difficult at the best of times...

Wirelessly linked smoke alarms are becoming extremely common now in the UK because outside a narrow envelope building regulations want linked alarms (linked either wired or wirelessly) on every floor. Imagine converting a basement and attic of a 1800s victorian house, and having to find a way to run a new cable from the roof of the attic to the ceiling of the basement. So instead it's a relatively dumb chip which shouts 'fire' on an unlicensed frequency block and a great deal of complexity is saved.

There is a chance you could do here what powerline adapters do, and use the ground wire to communicate. But they are notoriously inconsistent and there's any number of reasons it wouldn't work

And as I mentioned in another comment, many EU houses have concrete (internal) walls. Tubes for electricity wires come pre-installed in them so together with the other things mentioned, just running some wires can be hard to say the least.

That’s why ZigBee in mesh/star topology exists. No wires except for the hub.

> There is a chance you could do here what powerline adapters do, and use the ground wire to communicate. But they are notoriously inconsistent and there's any number of reasons it wouldn't work

This is exactly how Insteon works in conjunction with a wireless mesh. I'm extremely happy with my Insteon setup -- it's just a matter of replacing light switches and outlets with a bit of re-wiring on 3-wire switches (no new wires to run, just bypassing one of the switches so only one controls the light). The Insteon-branded hub is fine for a few light switches, but if you're doing a whole house and need something that can handle more complex configuration I would go for an ISY994 controller (don't know if anything newer exists but it's been rock solid for the last 5 years).

All that said, it was expensive and probably not worth it. I spent $4k on a somewhat large house and I was able to justify it as a learning experience for in-home IoT device use cases -- but I'm not sure there's much benefit to doing a whole house with it over just a couple rooms.

> Wirelessly linked smoke alarms are becoming extremely common now in the UK

Also becoming a baseline feature in detectors in the US, in my experience. New houses have been using wired interconnect for a long time now, but in the last couple years I've helped several older relatives upgrade their smoke detectors and even the cheap ones have wireless interconnect now.

Those are all reasonable comments, I have spent the last two years doing up my own house, but that is NOT what robdobson.com did.

Note: "Since Lorenzo (the sparky) was replacing most of the wiring in the house I managed to get him to bring a lot of the control cables together in around five separate locations."

Granted, not ALL the wiring, but enough that there are "at least three (control) units lost in the walls" that don't control anything.

The lack of access in a house to critical infrastructure is why that critical infrastructure has simplicity and reliability as its primary characteristic, and if you are going to experiment, maybe consider redundancy? The writer appears to have a lot of hardware expertise, and almost none about systems. Single point of failure? Check. Closed hardware systems that might be unsupported sometime soon? Check. No diagnostics? Check.

This was a fascinating project and I commend the writer for posting all the hairy details, but I am slightly appalled at the wastage.

I use Insteon dimmers and switches in my house. These replace the wall switch but look basically like a normal wall switch. They use a combination of power line networking and RF mesh networking for automation. This is the second house I've had entirely outfitted with them and I love it.

IMHO it's critical to maintain control at the wall switch - if I am walking into the kitchen and want to turn the lights on, I just hit the switch like any other house. But when I'm in bed and ready to go to sleep I can press one button in the app and all the lights turn off.

Anyone with basic electrical skills can replace a switch in about 15 minutes. I did them all myself in my first house over the course of a few months, and when I moved I hired an electrician for a few hours to just do them all.

The switches and dimmers are about $50 each, but they are highly reliable. I've had one failure out of about 40 total installed over the course of the 12 years I've been using it.

I did this on a prior home and then removed almost all of them just prior to moving. I haven't re-installed them in the 4-years we've been at our new home simply because the "failure rate" was so frustrating to me.

I don't mean hardware failure, I mean "intermittent, rare command failure" -- I've never experienced quite the same level of annoyance/rage from anything else, versus hitting a switch and having it either not turn on/off or having it be delayed by 1-second.

Granted, that was relatively rare, but having a wall switch not work even just 2-in-100 is enough to give me serious pause. I did love & do miss the ability to turn all lights off from a bedside switch, to run scenes, etc. -- but I'm just not sure it's worth the trade-off.

Are you saying you would hit the physical switch and it wouldn't activate? I don't think I've ever experienced that.

Early on, I would have command failures when the dimmers were powerline only, but with the dual band ones (RF mesh + powerline) that they have been selling for the last 5+ years I don't think I've ever had an error.

Yep, exactly. It was pretty rare, and I think it mostly occurred on "3-way" switches that were capped/originally wired as 3-ways but changed to be normal switches with Insteon links to get the dual-switch functionality.

It only takes once-a-month of hitting a physical switch, then having to do a "double-take" and re-hit it to get the light on, before the proverbial rage boils quickly.

All of mine are indeed the newer style RF + powerline; I also had my Insteon hub fail once, which was frustrating because it required some re-pairing that didn't "just work" IIRC. I was using a dedicated Mac Mini running the Indigo macOS app.

I may give it another go in my office at some point, we'll see.

Worked on a project for Kazakhstani real estate company to make a cheap "smart home" solution.

Went for 100% wired installation exactly because of that.

Second problem was every "star" topology hugely increases wiring cost, so a bus it was, and with bus came a problem of software MAC not being performant enough even to send audio, electrical isolation, CSMA...

Long story

I can't wait for 802.3cg

>Long story

It sounds like a story I'd like to hear more of, if you've the time

Ok, here it is:

1. Wireless is off for so many reasons.

Even if cost wouldn't be a problem, wireless has many other disadvantages.

First, it's not really wireless, you need to wire power, and cut walls.

Second, most of devices will spend their life deep in the junction box, or even cemented inside the wall. Having an external antenna for every one of them is impractical.

Third, no wireless standard is even 1/10th as reliable as wire. They are simply too complex.

Fourth, you want to have backup power for such vital things like HVAC in highrises without windows, security, and fire alarms, emergency water, and electricity cutoff, door locks, and intercom. This way, you need to have wiring anyways.

2. Star topology as I said hugely increase costs through: additional labour, cabling, cutting walls, switches, bigger junction boxes.

Main idea is that you don't want the builder to have to cut walls when they can not to.

2. For busses you have no alternative to CAN, or RS485 based busses, or multidrop ethernet hacks. Third is off because it's expensive (think of having 3 port T hubs for every device.) First is off because of bad hardware selection.

So we are left with many hacky RS485 based solutions.

Choose your poison:

Unreliable CSMA

Manual addressing

Super low performance

Polled nature

Near no existing ecosystem

In the end, MODBUS with hacks it was. The lesser of all evils.

4. It being a bus, and you having devices connected to other appliances, you have to provide electrical, and I/O isolation, as you don't want a random man plugging 220VAC into it in error, or deliberately frying all electronics in the highrise.

That's hugely expensive, and sometimes you want to power external devices through your box.

This necessitates to invent many case-by case compromise solutions.

It took me several days to do all the drilling for the star in my house. It was a lot of wire. But it was all worth it.

Tip: get a Milwaukee right angle drill. Worth every penny.

I was considering "smart" home active lightining stuff but finally went with 120% wires (some extra wires left for some future needs, double lamps etc) and traditional 2 state mechanical switches.

It is not about money but maintenance burden.

I can't imagine where would be my marriage when I was gone for some week long business trip and all that "smart" stuff had failed that time for my family. So far for 8 years of use I only had to replace one switch which got punched too hard by kid.

This. Every "smart" switch should also act/fallback as a physical analog switch any way. I don't understand why that's not the case... Edit : plus I don't want to pull out my phone or whatever device to switch the goddam light on

Well, there's two different types of smart switch:

- The first type isn't actually connected to wall power, and is basically just a small remote that tells your lighting to turn on/off.

- The second type is connected to the wall power, and works perfectly fine as a normal wall switch and controls the lights in the same way. Along with being a normal wall switch, it's a receiver and can receive commands from elsewhere (including the first type of smart switch) to trigger the normal wall switch functionality.

The second type is an easy retrofit for existing homes with lots of lights, but doesn't play nice with smart bulbs that need to be always on for advanced functionality (like having automatic step-up/step-down lighting levels at sunrise/sunset).

At the moment I feel like the best use of the first type are switches that fit neatly over normal hardwired wall switches (like the Lutron Aurora), or adding a spacer (see https://www.shapeways.com/product/35B9F36RD/philips-hue-dimm... for one example) so that your smart switch can just be popped off to access the normal hardware.

a few days ago I helped my neighbour install some light fixtures in her house. She was sold some smart lightbulbs with it because you can dim them and change the light color a bit from a remote control. There are 6 or 8 lightbulbs next to each others.

I'm impressed how unreliable this tech is. Standing 2 meters away from the lightbulbs, I had to press 3 or 4 times the off button to turn them all off. I just don't understand why you would knowingly install that kind of crap.

Depends on the bulbs. Our Philips Hue bulbs (ZigBee) work perfectly fine. None of the issues you or the OP describes are an issue: they continue to work as regular bulbs if the hub fails, can be paired via Bluetooth app without proximity to the hub, and seem to have a lot of range. Best thing is that no wiring/electrician is necessary.

Same experience here, Philips (and other Zigbee lights like IKEA) did a great job.

Only problem I had with the Philips hub is that it's memory limited and can't handle more than about 50 devices. Above that you need to replace the hub with something like Home Assistant, but can keep the rest of the system.

You can also use multiple hubs (HomeKit and various third party apps support this - the Hue app does, but IIRC you have to manually switch between them).

Yes I considered that, but it made some use cases that had buttons control lights that were on different hubs impossible.

Since lockdown started, we've gradually converted most of our house to Hue. A key feature is that _all the light switches still work_. We just don't use them all that much, unless something's failed.

From reading Matthew Garrett's exploits, I'm pretty sure we don't want WiFi IoT devices. We have Hue switches next to the physical switches, which don't depend on WiFi being up; Smart control works with only local networking. Add more smartness for extra control (SmartThings, remote access, Google Assistant), but it's no less functional than a non-smart system when the smarts are offline.

The killer feature for me is the ability to adjust the colour temperature and brightness of lights in each room -- I've got a 50W panel above my desk which makes my work-space brighter than it is outside at the moment (Edinburgh, UK, 1pm with rain) but automatically fades down to a warm, dim light as the evening progresses.

Would you mind sharing what panel this is?

It's a Hue ceiling panel that I'm not sure they make any more. https://smile.amazon.co.uk/gp/product/B07FXT4XH6

For those who don't have Hue, one nice feature of the design is that if the power to the bulb goes off and then back on the bulbs default to 'on'. This means that if you have a power cut, the lights come on automatically afterwards without the controllers being needed, which is useful at night so you don't have to wait for wifi. But it also means that you can turn them off and then on again at the normal light switch and they'll come on.

> For those who don't have Hue, one nice feature of the design is that if the power to the bulb goes off and then back on the bulbs default to 'on'.

After a recent update this is now also configurable. You can still have them default to full brightness, but also off or any other mode.

If I had a short power cut in the night and it resulted in all the lights coming on, I'd be pretty unhappy.... I'd rather just sleep through it.

I don't have them in my bedroom, so that hadn't occurred to me. It's configurable though apparently.

My issue with Hue is brightness. I have swapped all the living room bulbs (3 lamps, 2 ceiling lights) for a mix of Hue Colour ambience and White Ambience, but it's only just about bright enough for when I want the room well lit.

I'd love to put my office and some other rooms on Hue so I can tweak the colour temperature, but I've only got one ceiling pendant and it wouldn't be bright enough during the day.

We have multiple lamps in our living room. For the bedroom, which is also my WFH office, I've added a 100W LED that's intended for warehouse lighting but fits nicely on top of our Ikea uplighter. It's controlled by a Zigbee power socket, which means I can connect it to Hue as a "light" or SmartThings as a "switch" depending on which behaviour I want.

It's not quite as smooth a transition as pure Hue, but it means I can add a _lot_ of brightness when I want it.

The ceiling panel is really nice, and bright enough for the small area directly under it (which is my desk) but the rest of the room still seems a bit dim in comparison.

Unfortunately the Hue app doesn't really know about lamp brightness -- if you ask for all lamps on really dim, and one lamp's not dimmable, it'll switch that one on anyway.

What is a wireless lighting system? What is the advantage of such a system over a switch next to the door?

E: A wireless lighting systems seems to be a light bulb with WiFi capabilities.

It's an interesting academic exercise / hobby. Beyond that, nothing.

It's solutionism.

Never seen that. I like it.

Providing a solution before the problem exists is exactly what this is.

For example, not having to get up and go to the door. I find that extremely convenient when lying in bed, about to sleep. I usually read for a few hours with dark red light, when I’m ready to sleep I can just turn the lights off.

I've had both a switch for the main light (s) and a lamp bedside the bed all my life, multiple houses, none of them 'smart' or wireless.

Don't get me wrong - I want to use HomeAssistant or similar - but I don't want to ditch physical switches.

I have physical switches, I just don’t use them. They are my fallback for when the network fails, the hue hub dies or whatever.

What I mean is I think I prefer them even if I were technically set up to use my phone or a remote - they'd never be in the right place so I'd have to get up anyway.

(And in this case the (dumb) switches on the wall have always been in the right place, by the bed, in my bedrooms. (In addition to one by the door of course.))

I'd just have 'the system' 'aware' of the switches and current state etc. so that I could have certain automations, logging, etc.

The physical switches would either be network inputs themselves (easier implementation but reliability concern) or just report events but be electrically connected to the light still; I suppose also light-side state reporting for ultimate reliability (against missed switch events that did change light state because of electrical connection).

There is a technical solution for this which might need some research and I’m not sure it’s been proved out yet but if you search for ‘bedside lamp’ you might find a starting point.

Right, but then you have to go to the bed, turn the bedside lamp on, go to the door, turn the room lamp off, go to bed. Bit of a round trip.

I wound up with a LED strip attached to a RPi under my bed, driven by a clapper script. Doesn't work completely reliably, but well enough.

Why would you leave the room light on for reading in bed?

Because I don’t want to read in full darkness. I sometimes read for several hours before I sleep, the lack of light would make me fall asleep far too early (I already wake up at 4 in the morning when I sleep normally).

You don't really need a "smart" light for it, a traditional radio-controlled "dumb" light switch is good enough for this purpose. Just press a button, and the remote control sends a RF signal at 433 MHz to switch it on or off, there's no IoT complexity like firmware updates, Linux kernel, Wi-Fi, Bluetooth, an Internet connection or other nonsense. It can either be a light switch installed on the wall, or an integrated unit that comes with the LED light fixture (which may also allow you to change the color). Also, neutral wire is not a problem, some wall switches don't need it [0].

"Smart" lightbulbs are more flexible with more features, but personally I'm not a fan of them - many features with few additional benefits (with high price tags, possible bugs, and privacy concerns). Nevertheless, if you are not allowed to change any electrical wiring in the room, using "smart" lightbulbs can be a justified choice. Another genuine use may be controlling many wired-together lightbulbs on a large light fixture as separate groups without a laborious rewiring. Meanwhile I'll keep using my dumb wireless light switch.

[0] They solve the neutral problem by using an always-on high resistance in series with the "hot" wire. The lightbulbs on the light fixture completes the circuit, a tiny current starts to flow, which is harvested by the switch to power itself. Clever hack!

The question was for wireless lights. A radio-controlled dumb-switch used this way is still wireless lightning.

And what privacy concerns are there with lights that have no internet connection whatsoever?

First, some context. When I was replying to this thread, there were only three comments, the first said wireless lighting was an useless solution waiting for a problem, another said it was a hobby and not too useful, and only you said it was useful. Hence, it thought it was natural to add some comment under your comment. Furthermore, the original article described how the author designed a Wi-Fi based system, and other readers seemed to have "smart" or "Wi-Fi" lights or lightbulbs in mind when they were expressing their disapproval to wireless lightning, so I thought it was useful to make the point that a remote-controlled light is not necessarily a IoT "smart" light, which constitutes the first paragraph of my reply. And the second paragraph in my reply was my personal opinion to those "Wi-Fi" / "IoT" based lightbulbs - they do have legitimate uses, however they may contain "possible bugs, and privacy concerns" (as in: a Wi-Fi controlled lightbulb with a mobile app), thus "meanwhile I'll keep using my dumb wireless light switch".

I hope my comment makes sense to you now.

> A radio-controlled dumb-switch used this way is still wireless lightning.

Yes, and it was also the point I was trying to make.

> I hope my comment makes sense to you now.

It does, thank you :)

I have an interesting setup with home assistant similar to your needs

Power the bed reading light lamp with a zen15 zooz inline switch which is never shut off but can monitor current and power continuously. So now homeassistant knows when my bed reading light is on or off based on if the lamp is drawing more than 3 watts or less. Which means it can control my ceiling fan appropriately AND dimmable room light bulbs AND mess with the thermostat depending on outside temperature / mode. And maybe other options someday in the future.

Some effort to set up the hardware, maybe 30 lines of YAML automation code, but it is nice and convenient.

I have a "reading desk lamp" in my basement lab that does similar tricks. Desk lamp no current drawn means my soldering iron and some other apparatus has its power removed. Improves safety a little.

If my youngest kid were not an older teen, I've been thinking about wifi/bluetooth presence detection such that my table saw has no electrical power if my phone isn't within X feet of the saw. A kind of similar automation but different opportunity.

I’m actually contemplating exchanging all our halogen spotlights (as they’re no longer being sold) to IKEA Trådfri.

My rationale for it is that I want to be able to dim them, but exchanging halogens for LEDs would mean the transformer wouldn’t see enough of a load to dim properly. By exchanging everything for IKEA bulbs I save on changing the transformer since the dimming circuit is in each bulb.

Beware, the Tradfri spots are not as bright as regular bulbs by far. Try out one first before committing on a couple hundred dollars.

And if you're hunting for relays and other actuators... the compatibility between IKEA Tradfri products is excellent, but your mileage will vary when going with other Zigbee products.

I like and have bought into the IKEA system, but I have had a lot of trouble with their gu10 lights.

Some sets of three occasionally go into party mode and start flashing in a sequence. They reset themselves every month or so. I'm tall and my ceilings are not so it's quick for me to pair them again. It would be super annoying if they were too high to reach. Setting the white spectrum fails using the 5 button controller. Often one or two usually fail to set correctly resulting in a weird mix. I simply don't bother anymore. They work nicely with the app, when still paired.

I haven't had any problems with the e27's in the same rooms.

'Dumb' rf motion detector triggered under-counter lighting in the kitchen has been way better bang for buck. It just works. No smart home required.

I have found it useful when you have your hands full (e.g. hands full of shopping, or carrying a baby with a dirty nappy/diaper to a changing table are just two examples where you can just shout to google/alexa to turn it on), or turning stuff on/off when you are away from home (e.g. automatically at sunset, turning heating on when you are on your way home etc), or you left the lights on and you are already in bed etc.

First world problems, but it is handy.

I am sure that this is a cool feature, but I am really having a hard time wrapping my head around all the technological overhead required here to avoid having to flip a switch with your elbow or hip if your hands are full. "First world problems" seems like an understatement here.

The lights are just gravy for me. They're super cheap, and it is handy to be able to turn a bunch on or off with one command. Or have them turn on when you get home, etc.

What really keeps me going with home automation is the door & lock control. Being able to monitor the status of the garage doors and the deadbolts on all doors is worth the price of admission. The kids each have their own code to get in the house, I can lock the doors from afar if they get left open or unlocked, I can check the status at any time, etc. I have the system close & lock everything at bedtime if it isn't already, and it communicates status to me as things are opened/closed/locked/unlocked.

I could give up the lighting if I had to. I could just use the built-in scheduling on the thermostat. No biggy. But the locks and doors functionality is worth the effort.

> you left the lights on and you are already in bed

1970's technology to the rescue: https://www.amazon.com/Clapper-Sound-Activated-Switch-Each/d...

As for when my hands are full, elbows work just fine.

If/when I become bedridden or crippled, I'll probably get Alexa. Until then, I get up and walk to the switch. It's a privilege to be able to do that, and I'll do it as long as I can.

1970s technology appears to be 3 times the price of a WiFi switch though :) (...not including 16USD delivery...)

Do clappers hear through doors and across different floors? E.g. in bed upstairs and light is on downstairs? Plus would one clap not turnnoff everything that hears a clap, and not just the thing you want off but leave other things on? And can it auto clap itself off/on at certain times when you are asleep/not there? Can it hear you clap when it is at home and you are in your car/at work? And only 300% as expensive as a WiFi switch? Bargain.

But you know, when my hands are full I can just put everything down and rub two sticks together to make a fire just fine. The power of humanity's intellect and dominion over nature is a wonderful thing and making fire is a privilege that not all mammals have that I plan to use for as long as I can - 4000 BC's technology to the rescue </sarcasm to illustrate a point>

So much technology could easily be done "the old way" (fire to oil lamps to candles to gas lamps to electric lamps for example all make the room bright I'm the darkness), but it makes things simpler/more convenient/easier/etc. This is one of them IMO

You might be amused to know my car is a 1972 Dodge.

That does not sound like a good decision in terms of the safety tradeoff vs whatever benefit you get from driving it. Cheap? Am I wrong about that? Does it increase the risk of injury and death for other road users or are you sure it's just you and your passengers alone who are consenting to carry that (unnecessary?) risk?

It avoids having to run wires through the walls (the switch can be battery powered or even piezo-powered so it works just from the force of your press). Having said that most houses already have the wires in the walls, and even if the wires are old/unsuitable you can pull through new ones, so for most people it doesn't really make much sense.

There are some reasons to go for a wireless system, for example when you can’t just go and run new wires to have switches in places you’d want to. Either because it’s an old house or a rented unit or something like that. But when planning a refurbishment or new construction, pulling (ethernet) wires is the safe bet.

So what powers the switches then?

Batteries for example, or you may have power from other circuits that you can tap. A light circuit with multiple switches requires comparatively complex wiring - either all switches need to connect to a central point (breaker box) or you need to have some sort of loop that passes by all switches. It’s nontrivial to retrofit. Some sort of (wireless) bus system that transmits the signal digitally can help reducing that complexity.

There are switches that use the kinetic energy of the button push to work. No batteries, no wires.

https://www.enocean.com/en/products/enocean_modules_24ghz/

I have a system called KNX, its not wireless but each switch is independent of each actuated function (light, blinds, etc). The main advantage is the wiring is simple, a bus for all switches and all lights go back to a relay board. Switches, sequencing / grouping can be assigned at any time after install, scheduled on/off, easy to extend later, etc. For example I built a garage 10 years after my house and by running a single cat 5 between the buildings, I can now press a button (any button I wish) in the house and turn all garage lights off. I can also see the state of any light or actuator at any time, etc. So there are some nice advantages, but I'm glad it is wired for sure..

The author mentioned that his house is made of stone, so running wires to the lights is less easy than in a hollow walled house.

It's not hard, you just run the wires in conduit. Yeah, you can see the conduit, but make it part of the decor. You'll see this in a lot of retrofits.

But it is not as easy.

Stuff like this is the reason the only smart devices I have got so far are a Chromecast Audio, a Chromecast TV and a 2nd gen Google Home Mini (free from Spotify Premium). Those things are low impact if they don't work properly, and they do stop working with alarming regularity. I have spent less than £80 on the set up.

The Home mini has trouble with my regional British accent. The Chromecasts are within 2m of my router, and still manage to get disconnected randomly.

I'm not anti-smart home, but the stuff has to be 99.9% reliable for me to consider going forward with it. I've never had a light switch fail on me...

I don't think we'll start getting reliability like that until manufacturers 1) stop trying to hook them up to the cloud 2) default to wired and 3) implement failsafes (you can still turn off the light switch manually).

Until then it's the internet of malware infested broken toasters.

There are vendors who do just that, in various combinations.

Hue uses non-wifi wired protocol and offline-first hub. Lacks failsafe though.

A local vendor does the above, but instead of "smart" lightbulbs, offers smart switches - which still operate as physical switches when directly used. But they don't have huge marketing or anything like that, they appear to mostly sell to integrators (with their gateway then being integrated into popular solutions like HomeKit etc.)

I had almost exactly the same problems, albeit on a much smaller scale of perhaps 10% the size of this person's network.

I spent years and a fair chunk of cash trying and failing to get a reliable zigbee and z-wave network set up using various bits of kit.

To echo the article, the problem with zigbee & zwave for me was that things would stop working, and it was almost impossible to work out why. The only "solution" was to start the annoying repairing process as outlined in the article which was a huge ballache.

In the end I also binned all of the zigwave & zwave bits and moved to a 100% wifi-based solution using MQTT & Tasmota firmware (1) running on very cheap hardware that is usually <10GBP per node. It is very reliable - I don't think I have had a single unexplained outage in the couple of years I've been running this now. Wifi + MQTT is mature and just works. It is also super easy to diagnose problems - you can see the devices on your network, ping them, subscribe to their MQTT topics, manually send messages on MQTT etc. Its a dream compared to the black box of zigbee & z-wave where you essentially had zero control apart from on/off.

Wifi + MQTT + Tasmota appears to be the solution for me.

My main complaint now is that HomeAssistant cant connect to Google Home/Alexa (so that I can use voice control), so I am using OpenHAB (which has a free integration that works). OpenHAB seems absurdly complex though, and they seem to be in a continual process of rewriting the software (some stuff is stored in config files, some in a databases, and the location of the config files is not in the docs/the docs have the wrong location so backups are hard, the docs are only for V1 (but don't say that they are only for V1) but the current version is V2 that is also deprecated because they are working on V3 etc, the recommended UI can't do everything you need to do to get the system working so you have to manually update stuff in a shell but the aforementioned docs telling you what file to edit are just plain wrong etc because the are referring to some different version, running OpenHAB in docker requires some odd permissions/user set up on the host machine etc - it is a mess). I am planning on ripping out OpenHAB and replacing with NodeRED at some point since that seems sane.

1 - https://tasmota.github.io/docs/

I use OpenHAB purely as a ZWave I/O engine for Node-Red which works quite well for me.

When people have more money than good sense.

I can't imagine how large a house or must be, to require 150 lighting circuits.

My house had a kitchen extension built by previous owners. I count 27 lights in total, mostly in-ceiling spotlights.

At least they're controlled with a normal switch. With decent quality, high powered LED bulbs like we have in the living room, 2 or 3 would do.

Don't forget time.

Jesus. All of the cost and complexity of managing complex IT infrastructure... for light bulbs. Is this satire?

https://thedailywtf.com/articles/The_Complicator_0x27_s_Glov...

This article honestly made me want to quit engineering and limit the amount of technology I am dependent on. You must have the patience of a god. Also your family definitely hates you

I don't understand why there are no powerline network industry standards fir home automation other than the aging X10 and we have to rely on wireless comms. Maybe the PRIME or G3-PLC industry alliances will change this and come up with something useful. This is also needed for smart grids. It would eliminate the need to run parallel wiring for automation or use wireless tech. Separation could be done at the circuit breakers with lowpass filters and cable runs could use shielded cables with the shield connected to earth in order to eliminate crosstalk and prevent the power network become an antenna and pollute the radio spectrum with RF noise. Today's network tech for smart homes is immature and relies heavily on WiFi and industry standards designed for computer networking when in reality what is needed is something like a powerline bus for flipping switches, reading meters, remote controled power factor correction and so on.

How much power does the new system use when everything is nominally off?

I'm going down a similar path, although I have been doing it differently from the beginning. My home is still predominantly Z-Wave, but I never got on the hidden-relay bandwagon for lights. All of my lights which are on the Z-Wave network are wall switches, so they work as dumb switches if the controller is down. Number 1 rule, do not ever rely on the controller for basic on/off operation.

But I agree with his points about maintaining the network. It's a pain to add new devices. I bought a bunch of new switches, and they're all Kasa WiFi switches. Easier to set up, more responsive than Z-Wave, and a lot cheaper.

Perhaps the one downside is that a WiFi device by definition has one less (really big) hurdle to become Internet-connected. So I did have to be careful with my WiFi configuration and firewall rules. But I'll take that trade-off. The experience has been far superior.

It blows my mind that PoE LED drivers aren't an established dirt-cheap standard I can choose from multiple vendors supporting @ my local Home Depot/Lowes.

> there are actually at least three units lost in the walls somewhere which respond to Z-Wave commands but don’t seem to control anything I can find!

Reminds me of http://bash.org/?5273 haha

There's a better version of it -

https://www.theregister.com/2001/04/12/missing_novell_server...

Is different. Bash guy cannot find it. Other guys didn't know it was missing. If it was missing for four years and then found it would be a better version.

Never send a consumer IoT (where the 'q' stands for 'quality') product to do the job of a commercial lighting control system.

The right tool to automate a house with 100-150 lighting fixtures is something along the lines of a Crestron system.

I've been thinking and experimenting with making my home smarter. There is one main rule.

When building such systems one thing must be done always. It must work when there is no smart component to that system.

For example, if your smart home stops working or startes to glitch, it should still be possible to go to a switch and turn that light on.

This can be easily done when building a house or doing renovations. Retrofitting, however, very much depends on way your switches, lights and other related things are wired and installed.

I’ve been using lightwaverf for around 10 years now and find it to be a decent solution. I originally bought a light switch and some of their pass through plugs, but recently I’ve been expanding with more light switches and a relay, mostly to control hard-to-reach-the-switch things like outside lights but now I’m thinking about replacing more indoor lights and sockets for a more complete solution

Lighting-wise their dimmers just replace your regular dimmers, though they are push buttons rather than switches and they have led indicators on them (that you can turn off). They work without the hub, not that the hub has ever failed for me

I’ve never had a problem with range like tfa suggests, even in places that struggle to get a wifi signal and out in the garden (125 year old Victorian terrace over 3 floors)

Their stuff is fairly expensive but my oldest equipment is maybe 10 years old now, still works well and is compatible with the newer generation. My wife is even coming around to it - being able to say “Alexa turn on garden lights” instead of trudging outside in the rain to turn them on has convinced her I think!

If/when we move I can mostly just put the old light switches back in and a) keep my kit for the next house and b) not have to subject the new owners to my home automation addiction

This is exactly the sort of system I was picturing while reading the article, glad to hear it exists and works well!

I've been in tech for a while, and I am very reluctant to jump in on "smart" appliances. Having said that, I have dipped my toe into things by tackling things I can fully control and maintain. I also make sure it is simple to bypass as not everyone who lives with me is interested in learning or troubleshooting issues with something like a light fixture or thermostat.

Cloud is absolutely out for privacy reasons and for the simple fact that no device should be dependent on my ISP functioning. So I keep the smart stuff to things like lamps or shades. All things that are easy to maintain and are not required to function to enjoy the home. Shades are on Z-Wave Plus and I have just 4 shades that are motorized and it's not easy to maintain. The only other thing I use is Tasmota on some cheap plugs for lamps and holiday lights. All of these devices are controlled by a web panel built on a NAS that has an easy to remember local lan address. Control via cellphone is easy so long as you are on the lan.

I just haven't been totally sold on the idea that the convenience offered by IoT devices that are locked in or require cloud services outweighs the maintenance and hassle over the long haul.

Does anyone here have any experience with KNX? It seems like the best wired system (without being locked into a single manufacturer), but from what I understand it's expensive and not aimed at DIYers.

For context I'm in the process of building a house and like OP am considering Shelly modules, but it seems a bit silly to go wireless when I am doing everything from scratch.

I don't have hands on experience, but I read about it a lot.

Yes, it is not meant for DIY, but in Germany there is a big DIY community, unfortunately that also means a lot of hands-on experience is shared in German forums and Youtube channels.

The main issue is the ETS software you need to configure KNX devices. Full license is around 2k€, but if you have up to 15 KNX devices I believe there is a cheaper license.

Next apartment/house i buy will probably be wired with KNX for roller shutters, heating/air conditioning and for lighting either KNX or KNX-Dali Gateway.

There is an interesting channel on Youtube if you are interested: https://www.youtube.com/channel/UCnGAZMA4wzOYx_P7aXTRc7A

Hi, I have built my house using KNX. Actually, the fact that it is mostly cable based makes this system super reliabe.

It also combines various trades, lighting, heating, etc and is supported by 500 manufacturers.

Together with a friend who has more than 20 years experience, we published an ebook on how to get started without any prior knowledge. There is a free version at https://knxtutorial.com/knx

Contact me if you have any other questions.

We built our house with KNX this year. While our electrician did all the wiring, I did the programming myself. It has a steep learning curve, but if you're in IT, you'll probably get the hang of it pretty quickly. The system is very robust and offers a lot of flexibility. Almost all our componenents are from MDT (https://www.mdt.de/EN_Start.html). They build really great products and are quite affordable. Compared with components from other manufacturers we use (Theben, Weinzierl, Steinel) I think their applications and manuals are also far superior.

I have since connected the system to my home assistant installation, and enjoy the added flexiblity.

My Philips Hue stuff in my flat works very very well and i like it.

When i will buy a house, i will put in cables as i still think cable controlled is more reliable.

Nonetheless, i like my system but it could be better especially the heating: It is actually quite useful to control my radiators digitally. It also makes a lot of sense to let my radiators measure room temperature and let it detect open window.

Normally when i activate a radiator, in 2-3h the room is always way too hot and i do only put it to the middle number (3).

Is this critical? No Is this important? No Can i live without it? Yes

But i do think its here to stay and in the long run, IoT will become the default.

Light atmosphere actually works very well when you can activate it by one click Light switches are were you need/want them Voice command feels much more natural and is quite useful

I like to think how my future house could incorporate something like alexa (a smart agent) in every room but 1. with powersavings 2. privacy in mind.

Something like the line in I, Robot where the main computer can see everthing.

I have been gradually installing Shelly relays and dimmers in my place for some time, and they are excellent. They have a fully open and documented REST API, a simple but effective local web interface, and any Cloud connectivity is completely optional.

I have never had one fail (N=25). Since installation, I have never had one drop off the WiFi or require a physical visit for any reason at all (one is installed in my boiler and it takes an hour to get it apart, so I really appreciate this!). I honestly don’t think a single one of them has missed a single rest API call in the time they’ve been running.

When initially installing them, this can be done from a mobile next to the unit, at which point you join them to your WiFi.

I’m not in any way connected with Shelly, just a very impressed user of their stuff.

my opinion: z-wave for lights only is not efficient - it mainly can give you dimming capability, no control over warm/cold lights or even color. plus everything is an unfortunate combination of wiring and rf. zigbee based lamps like hue or others are much better here.

however, for controlling things like heating or blinds I find it very effective in combination with sensors. It is quite simple for everone to do simple things, like setting schedule based target temperatures. and reasonably simple for a dev to do complicated things, e.g. to control blinds based on exact position of the sun around the year. plus you can always use actuators that work without the base controller, so the base controller is just an option.

I greatly enjoyed the article, love the Shellys too. +1 also for Homeassistant. Just wish (minor point) the author had chosen a font-color with more contrast to the background, like a slightly less pale shade of grey.

I love browser's reader modes to easily apply the styles I like to articles like this. Instantly dark mode and huge font to any article

Can someone list productive smart lighting features?

I mean: I get smart heating. I get smart ovens. But smart lighting? I can't think of non-gimicky functionality that can be solved by a occupancy sensors.

The main thing I like about smart lights is the concept of programmed “routines”. I have a large family so perhaps that makes it more beneficial.

Morning routines that run automatically:

- Turn on the bedroom lamps at a low dim level around the same time the alarm goes off. This provides a bit of light when you first wake up without being annoying.

- Turn off all exterior lights. Most are probably off anyway, but this catches any they might have been left on, or any that were scheduled to be on through the night.

- Turn up the nightlights / stair lights to 100%. In the dark winter months, this makes it comfortable to navigate from bedrooms and bathrooms without having to turn on the full hallway lights, which can be jarring and perhaps annoying to others who aren’t quite up yet. (A bit after sunrise these go off for the day).

- Turn on a few lights in the main living areas (kitchen).

Evening routines that run a bit after sunset:

- Turn on outside lights

- Turn on nightlights and stair lights at 50%.

- Turn on the bedside lamps at 50%. This provides a relaxing light level in the bedroom through the evening. Before the smart switches we never really seemed to use the lamps. (The hassle of walking to both sides of the bed and setting them at the right dim level I guess was just enough to prevent us from using them I guess).

Bedtime routines that run automatically:

- Turn off all but one exterior light.

The other thing I like is being able to control lights from my phone. When we are in the backyard playing with the kids on a late summer evening, we can turn on all the backyard lights without having to go back to the house. If the kids have left a bunch of lights on in the basement, they can be turned off with Siri “turn off all the basement lights”. I also like it from a security perspective “turn on all the outside lights” which would normally take walking all over the house to hit all the various switches.

I also have appreciated being able to add light switches literally wherever, with no wiring or holes in the wall. Want a bedroom light switch next to your bed? Peel and stick and you’ve got it, and from the outside it looks and acts exactly like a real switch. If you move the bed to the other wall one day, you just pull the switch off and stick it in the new location.

For a long time I resisted anything “smart” in my home. I found the added complexity more than offset any benefit. But with the lights, we can just use them as normal (all the switches look and act like they always did) but the automated routines add convenience and comfort while also reducing our energy usage.

The main reasons I use smart lights are in my kids rooms after they fall asleep I can turn them down/off, and we leave the living room light on and my spouse and I argue who needs to go down to turn it off--now we don't argue!

For me, it's simply grouping lights.

Example: in the winter months if I'm waking up to take out my dogs, it's pitch black on the main floor, so a command like "Alexa, good morning" will turn on the lights for the path that I need to get to the back door and tell me what the outside temperature is so I can wear the appropriate coat.

The second most important use case for us (pre-pandemic era) was to have lighting routines that made it look like we were home while away on vacation.

I started with some wifi devices (WeMo) but eventually migrated to Z-wave, and found them to be much easier to setup and maintain (and generally more reliable) than wifi (ymmv, of course).

My Z-wave hub is basically a USB Z-wave stick connected to an old NUC that's running OpenHAB in a Docker container. I rarely log into it or do any maintenance on it other than when I add new devices, which is almost never. I chose OpenHAB over HomeAssistant (YMMV) because I found it to be easier to write a couple of basic scripts to handle state notifications for my smart locks. I had been on SmartThings which was fine prior to getting smart locks and then turned into a major pain afterwards (one firmware update basically broke all my integrations, forcing me to do everything all over again).

Keep in mind, I didn't really go so far as automating everything "just because I can". 100+ nodes (as in the article) in a home situation just seems bonkers to me.

dumb occupancy sensors are incredibly unreliable.

Smart code based on MAC addresses seen on my wifi, timer based, power drawn by the wireless phone charger next to my bed, door positions and other things added together is not difficult to code or understand, but very reliable.

If I have all that on the input side may as well have it on the output side.

Some sensor work is utterly trivial. Door position plus sun elevation based on NTP accurate clock equals lights on / off in the laundry room. Position of garage door equals garage extra lights on/off kind like a giant fridge door and its fridge light. The complete set of presence code for my home office is a little more elaborate.

Our ceiling lamps and mini-split air conditioners are all IR remote controlled, so I just got a couple of IR blasters and set them up with Homebridge, so I can control the aircon/lights/TV with my iPhone or Apple Watch. If the smart stuff breaks down, I can just throw away the IR blasters and everything returns to how it was before. The only negative is since IR is one-way, you can't see the current status of things after you've manipulated it with a manual control.

If his house is in Dick place, I delivered "the Scotsman" to former owners every morning before school for years in the 1970s.

There was an albino blackbird living in the neighbourhood.

I use Shelly devices with Tamota, they talk to my Home assistant setup with MQTT.

When I can get power (with a neutral wire...) to something then WiFi is the best, but uses too much power compared to Zigbee and Zwave for running off batteries.

I have some climate sensors in each room to control heating, they're too awkwardly placed to get power to, so I use Zwave for these - they were just as big of a pain to setup as TFA describes, but at least I only have to replace the batteries once a year

I'm glad to see that he chose to do it himself, I'm building a house and am also leaning toward a DIY solution, because then I am in control of component quality, and I can have a fully documented system which I know I will be able to hack as I see fit, and for instance, design in such a way that even if the controller is down, the wall mounted switches still work as expected. So, great, comforting article!

I bought one of the philips wireless lights working on wifi or bluetooth, can't remember. It worked fine after 2 minutes of set-up, and then stopped working about 1-2 months in. The light still works with the regular switch, but I can't get the connection back whatever I try, to turn it on/off with my phone.

Definitely won't be coming back to this until 5 or 10 years to see if the tech has reliably moved.

There's also systems like Loxone where you have to wire everything but it avoids a star-like system as you can connect switches and lights in a tree-like fashion (ie, to connect another node, find the nearest node already connected and wire to it, that dramatically reduces the amount of cables).

I have just two of these lights, both phillip hue's, and Alexa. They are pretty problematic, but they are the best solution I've found in my two lights, which are dim lamps that I turn on and off frequently when I cannot reach them.

I don't want to come off as unduly harsh here but I could feel my blood pressure rising the more I read of this article. I got to about point 4 and had to stop.

Why, you might ask?

Because I am "fortunate" enough to live in a house where the previous owner bodged and kludged every single home improvement and DIY job he attempted. He did it in a different way to Mr Dobson (who at least employed a professional electrician) but, nonetheless, the point stands. In my place even the most basic jobs have been done in a way that is incorrect, sometimes unsafe, and usually adds significant time and expense in fixing or making modifications beyond what would reasonably be expected. I am gradually correcting all of this but it's slow and frustrating work.

This article contains numerous sections that would be absolute red flags, at least to me, for buying this house if and when Mr Dobson chooses to sell it. Here are a couple of examples:

> To this day I have a spreadsheet with around 500 lines (around 3 lines per circuit) which details all of the lighting circuits in the house, the circuit breaker they are connected to, the type of device that controls the circuit and its location. Unfortunately I wasn’t 100% successful in getting the control point of all the lighting circuits in one of these five locations and there are actually at least three units lost in the walls somewhere which respond to Z-Wave commands but don’t seem to control anything I can find!

> I have now gotten to the stage where I dread some modules failing. The one that controls a water-feature in the garden is in a sealed box, the one that controls an out-door heater is in a small loft area that can only be accessed by a rickety ladder, the one for the guest-room extractor fan is in a crawl space, etc. This can’t be the right way to do things can it?

That last sentence nearly floored me: no, it clearly isn't, so why have you done things this way?

This is an absolute horror show and I have little confidence that anyone who approaches home improvement with Mr Dobson's mindset is going to do a better job of whatever the replacement ends up being. [EDIT: I have now read the conclusion which, at least in my view, bears out this statement pretty well.]

You can obviously do whatever you want with your own home but if you're going to go nuts on the home automation front, especially if you tend towards early adoption and want to attempt something very ambitious, you might want to consider what effect that could have on your ability to sell your home again in future. Homebuyers are going to become a lot more savvy, and a lot more sceptical, about these kinds of systems and modifications as time goes on.

My home was built in 1927, I would never retro-fit with a system like this.

I saw all this kit on ebay!

This has got me riled.

1. I hope all this electronic and plastic junk gets properly recycled.

2. Learn to be more prudent with your money.

3. £6k would have funded a lot of solar lights in developing countries through SolarAid or similar.

I don't disagree with point #1 but can you honestly say you've never purchased something that you truly believed would work better than it did? I'm pretty sure the OP did not intend to rip out his system and waste £6k when he started this project.

I truly do not understand the desire to complicate simple things like lighting. Especially considering the state of “smart” appliances.

I’ll stick to my physical dimmer switches, and “dim to warm” LEDs

The trick to smart systems is to have it consist of dumb closed loops. So that the central hub becomes redundant. And also manual override so it becomes 3 way redundant.

Off-topic, but

>Quentin Stafford-Fraser (co-inventor of the webcam).

How can you be the inventor of the webcam? What is it even, besides from a digital camera? Did he invent that?

he has a wikipedia page https://en.wikipedia.org/wiki/Quentin_Stafford-Fraser

He was one of the team that created the first webcam: the Trojan room coffee pot: Quentin pointed a camera at the coffee pot and wrote the XCoffee client program which allowed the image of the pot to be displayed on a workstation screen. When web browsers gained the ability to display images, the system was modified to make the coffee pot images available over HTTP and thus became the first webcam.

It wasn't even a digital camera for that system - it was an analogue CCTV camera hooked up to a framegrabber. Digital cameras weren't really a thing at that time.

So, has anyone used this Shelly 2.5 "WiFi-operated double relay switch" thingy?

Looks pretty cheap and interesting in principle.

Yep. Works as intended, easy to set up though as with most of these wireless technologies the initial pairing can be a bit fussy. The built-in software and smartphone app is good enough for me, but if you want you can flash new firmware and/or use HomeAssistant. I use it for roller blinds and for outdoor lighting on a sunset/sunrise timer.

I'm not a huge fan of home automation and my partner hates the mere thought of it, so what's nice about Shelly is that these units have not just outputs but inputs as well. If you flip a physical switch you will still trigger your lights or appliances or whatever is connected to it. You're not locked into an ecosystem with its own remotes, switches etc. – it'll feel like the sturdy old system you already had, except that now you can program it and control it remotely, too.

I am using it for a zipscreen, it has never failed. Comes with good software & API, but you can flash your own if you want.

This ends up being anti-Zwave. He replaced his system with a Shelly WiFi system.

I think I would have gone for wired.

I’m rather confused - what exactly is a wireless lighting system? And what are the advantages?

Being able to switch lights on/off either with conventional wall switches or programmatically, i.e. from your smartphone. If you don't get carried away with the "automation" aspect of it, it can really enhance the UX in some common cases.

It's a system for controlling lights wirelessly. The lights themselves are of course still wired, but instead of having physical switches you have a wireless control network turning them on or off.

He had it coming as he chose a proprietary locked-up system such as z-wave.

I noticed no one has brought up X10.

This sounds like a hacker gone mad.

I have just never felt like flicking a light switch was a problem in need of a solution. It works just fine for me.

I'm so confused why he didn't just go Phillips Hue?

I have 30+ zwave devices on my homeassistant and some years experience going back to the X10 and insteon days and here's my comments on the article and other folks comments:

1) TLDR does not like the effort required for hard wiring. Solution I used? Never hard wire end devices. Makes it easy to take it all with me if I move. Pairing is insanely simple and done in infinite comfort in my office chair not on top of a ladder or whatever nonsense. Then once it all works I plug it in the official place. I hated replacing my hardwired insteon switches. Obviously wall switches have to be built in, but why would I have to climb a rickety ladder with a laptop per the linked article to access a wall switch? And I've had ridiculous good results with RF switches and with automation based on presence and timers I don't "need" wall switches nearly as often. There's no point in rewiring a wall switch for my driveway and side door lights if the lights are now near 100% controlled by elevation of sun and position of garage door and some software timers. Honestly I'm not entirely sure where my physical garage light switch is...

2) Never had a problem with healing

3) Yes I too ran into scaling problems where I have a "large" number of zooz Zen15 inline switches and they work great but the out of the box homeassistant config had them spamming the current voltage and current ... current like every second. Which looks nice when you're demoing one device. Not cool if you have 20 of them spamming the unholy hell out of homeassistant. There are config options to fix that and suddenly my network works better including healing.

You uh, do know you can have multiple zwave systems? I mean zwave the protocol only supports 254-ish devices off a stick, but another stick is like $15? And to some extent with MQTT and some virtualization you can have "infinite" HA servers running? I was amused with the idea of running a separate HA instance for every ROOM in my house on docker. Its not hard to pass usb access thru docker and the sticks work fine. I would imagine if I ran long USB extension cords and multiple zwave sticks each with only like 10 devices that network thruput and healing times would be AWESOME. But I'm lazy to set this up and I only have like 30+ devices so...

4) See my virtualization idea in #3. If an aeotech zwave stick failed (note, never happened to me nor heard of it happening to anyone) then I'd cry some if I had the protocol limited 254 devices to re-pair, but with only 10 in a room or whatever, eh who cares something to do for a half hour while listening to a podcast or something.

5) Hate to repeat myself see virtualization idea in #3. Again, if my entire system died I'd be sorely inconvenienced, but if my living room virtualized dockerized server died I'd only be out my automatic fish tank lights, my automatic presence detected and temp controlled ceiling fan, and some area lighting effects based on roku and other TV device status.

6) Seriously dude I lived thru Insteon AC line caps blowing every year or two, you have no idea how good you have it with zwave. But see answer to #1 again. Have a zooz zen15 ready in backup and its like 5 minutes work to swap them out.

7) I have not had a dead node problem. Interference related? Only when the microwave oven is on? Honestly I donno. It sounds anecdotal along lines of "Don't use Verizon worldwide because my local tower sux"

8) Confused what short lifespan Chinese outdoor christmas lights have to do with automation. Like if your christmas lights burn out too much it doesn't matter if you don't control them, or timer them, or zwave them.

9) Yeah well CLIs are user friendly they're just a bit selective about who they're friends with. General public maybe not, HN type people will be fine.

10) What price safety, convenience, security? So the "light switch" for my basement stairs is the door, making it impossible to stumble down the steps in the dark, isn't that worth something? Building code requires a light over my entrance door which is practically never used, but now my icy steps in the winter are at least very well lit.

I think the biggest problem OP will have with his esoteric wifi light system is theres maybe 1000x as many casual zwave users and eyes on zwave hardware. Bug strikes down zwave then HA development goes in crisis repair mode with excellent support on the forums. The one install in his state of this guy's wifi thingie goes down, well... reinstall zwave or be very very patient?

People who hate the upgrade process on home assistant: Dockerized HA and tar or whatever snapshot strategy you want. Life's a lot easier than "misterhouse" back around 2000.

Wait so you have a separate HA for every room in your house? What benefit do you actually get from that?

Edit upon further reflection: I’m doing something very similar. I have one bad mesh right now with lots of repeaters to transit my ridiculously long house.

The house is a barbell with the “house” part on one end and an office on the other, separated by a garage, with a detached shed in the back yard. My plan is to put a zwave stick + raspberry pi running zwave2mqtt in every “building envelope”, so four total. Those will all talk to an mqtt broker colocated with my singular HA instance.

> Wait so you have a separate HA for every room in your house? What benefit do you actually get from that?

I could. Its just a docker image and a $15 usb stick, so... The benefit is as per the linked article the hardware (and software?) doesn't scale beyond 100 devices or so, and I'll never have more than 100 devices in one room...

My house only has a couple dozen zwave things so I'm not bumping up against the scalability limits as seem in the linked article so I'm all good.

Strange room idea: Since I have presence detection I could redirect incoming connections from my phone to the current room LOL. "Microsegmentation" also makes some strange possibilities for security.

Wireless lighting control is an inelegant solution to a nonexistent problem.[0] How many years of his life has this guy wasted in order to save himself the trouble of spending 2 seconds using a wall switch a few times a day?

[0] https://www.theonion.com/horoscope-for-the-week-of-october-3...

Reply on news.ycombinator.com