Hey Wendel :) In Austria is a small company called Loxone which builds a PLCs for Homeautomation. I think you might like it. Anyway nice series, I really enjoy it
This was great. I love the ramble and just actually addressing the faults of current IoT tech in a very practical and succinct way. The practical examples are perfect.
I never really thought much about smart home, but all the scenarios you mention are great. The big downside: Even if off-the-shelf products would exist for this you would still need to do a lot of setup yourself. And I think that's the catch. Things like nest don't actually do much smart things but they are inexpensive in the setup. You just buy it, install it yourself and you are done and can feel good. But if you really want a smart house you have to actually think about it (or pay someone to think about it). Every home is a bit different, there is not really a solution that fits everyone. This increases cost because it's a custom solution.
Never thought I would hear someone talk PLCs in one of the channels I follow. I work with Allen Bradley PLCs at work, so I was primed for home automation. Currently using home assistant.
The one wire protocol is a really neat protocol and very well thought out. Basically all sensors on the line are always on "Listening mode" waiting for a Reset pulse from the 1-wire Master. When a reset pulse is sensed, all sensors send simultaneously a presence pulse. Then there is a Scan algorithm to detect all the devices on the line, which is amazing. It can be slow (a few seconds) to scan hundreds of devices on the line, but you only need to scan if you connect a new device. Once you have a list of devices their MAC address contain their model so you can get all their capabilities from the mac and you can program the master to adress one device at a time to get or set sensor data.
I've been around and seen a few houses with actually decently electrically and computationally educated sensor set ups and they make me so excited to have a home of my own to do it to. Along with this series which has sent me on nostalgia lane to these set ups with your inspirational ideas. Beautiful !
Also if you're like me you have everything on two different flash drives or SD card so when it comes to 10 years down the line and the SD card of my Pi dies I have a copy of it on my computer and on another SD card. Redundancy never hurt anyyyybody
As someone who has been forced to work with DeltaV as an ICS you are correct, Windows won't update while everything is running, but it will break everything you can think of and more when trying to apply simple updates. Going so far as to have to rebuild machines from scratch just because Windows doesn't know how to update right anymore.
Wendell, as a PLC programmer of 8 years and counting, definitely saw the potential of home automation with a PLC. Heard of some colleagues who rewired their whole house to be controlled off of a PLC. Every outlet, every light switch. Sounded awesome at first, but then questioned what the next home owner will do with it... How do you future-proof it. I have to maintain a windows XP virtual desktop so I can reprogram some of the 20-year old PLCs that are still in operation. Also, BTW PLC firmware changes do shut down the assembly line while it is downloading, but is always a manual operation and can only be done one PLC at a time. Pros and Cons to this, the engineer is dedicated to making sure the new firmware works before moving onto the next PLC but also have every possible version of firmware on each machine across the plant, because who has time for that. PLC tech is also about 10 years behind consumer technology. Gigabit ethernet isn't really a thing for PLCs, partially because the bandwidth isn't needed. Partially because every PLC brand had their own network implementation. (See Data-Highway, Remote-IO, Controlnet, Devicenet, modbus, CAN, etc.) Personally, I'm interested in wireless devices. At least for now being in an apartment where I can't leave permanent changes to the building. Wifi devices are super convenient, and we know convenience to the consumer usually wins out. Zigbee and Z-wave haven't been too bad either. Would be awesome if apartments came pre-wired with ethernet so I could setup POE devices. I always wondered about implementing a commercial building automation solution rather than an industrial solution. Like some system that I could put temperature sensors in every room and had control of the air ducts and the HVAC to control every room to the temperatures that I want them to be at. Industrial PLCs are waayyy too pricey for a home use.
As someone who has worked for a for a HVAC company and has programmed PLCs, I have a much more grim look on PLCs. They usually don't crash because of how simple the programs are, but they certainly requires maintenance. Preprogrammed PLCs sometimes do have software issues though and requires rebooting/updating/etc. Industrial systems constantly have issues but there is people which monitors and deals with these issues. It is hooked up to hardware which has non-trivial interactions with the real world after all. In a small housing complex it might just be a janitor which calls for assistance when pressing the "Stop" and "Start" buttons doesn't fix it (and a lot of times it is just because they don't understand what is going on). Larger organizations will have a team of people which can actually fix hardware issues, with alarms set up to fix common problems before it becomes an issue. Live updates are possible to some degree, but normally stuff is shut down. Either because it is not an issue if it is offline for a few minutes, or because you are doing it while that section is not in service (say after work hours or doing scheduled maintenance). The grim look is this: They are usually programmed by people who knows nothing about computers. They are managed by people who know even less. And there is usually no security. The PLCs are usually not designed to be secure and the people who set them up knows nothing about it. By default many do not have a password, and if they do it is the default one because nobody changed it. (Many PLCs run Linux btw.) And yes, they are often directly connected to the internet. One of the hospitals we worked for had everything on a separate network, but anyone could walk in from the street to the "server" room and walk away with the computer and nobody would notice. We also once used the password on a sticky note we found to access one of the systems we usually did not work on. Another hospital had a lot of stuff break due to a power outage (including something I worked on because I forgot to move a switch, causing the CT scanners not operating because they depended on the ventilation working). And at a third hospital one of their technicians died in a pit of boiling water. These kind of things happens all the time, you usually just don't hear about it.
I have a friend who had an angle stop burst while he was on vacation. It flooded the entire house causing over $100k in damage. It took him over 4 years of battling the insurance company in court (he had restoration level insurance, which is far beyond what most people have and they didn't want to payout). With a moisture sensor or flow control cutoff setup like described this would have saved him years of headaches.
Over the lockdown I was actually looking into PLC programming to use on a swamp cooler. I grew up just outside of Las Vegas where you can run a swamp cooler as your main cooling source ~6 months out of the year. During the summer (with the exception of monsoon season during Jul/Aug), you are running it full tilt all the time, so control isn't as important, but the beginning (Feb/Mar) and end (Oct/Nov) of the cooling season, requires more intelligence than the commercially available thermostats can provide. That translates into controlling the pump and the fan controls manually based on time-of-day and outdoor weather conditions, which can get tedious. I haven't yet started designing it yet but I have decided to build a swamp cooler controller for my parents' house using some variety of PLC.
The water manifold control is pretty smart. But my dog has her favorite spot and it just so happens to be in a bathroom. But I'm sure there is a way to figure out how to exclude the animals. But, for the most part I have zero desire for IOT running my house via google home or feeding more data than necessary to the overlords at Google. Thanks for all the great informative videos Wendell.
With a flow sensor and motion detector you can do other things as well. Like if there is motion and water flow for more than say a minute, kick on the bathroom ventilation fan. Can also do a hot water recirc loop and have that turn on whenever there is motion in that bathroom, so when someone turns on the faucet or shower it already has hot water right there ready to go, and you don’t have to wait for it.
Nice series, thank you for making it! I personally would be quite interested about a deeper dive into decent alarm system: as in, with decency, without cloud or phoning home or an external company having data or ownership over your home/business system. What exist, what's possible, where are the pitfalls, examples, and so on.
In Industry wifi is no-go, no direct internet access-vlan and vpn to specific machines/locations and all wired, redundant and failsafe with PLC's. For home we don't need cloud at all. And standard for homes should be derived from good industry practices with some cost considerations.
I worked on tram systems, all the safety critical stuff was programmed into the PLCs, this then allowed the developers to build all the high level management stuff without the worry of creating a conflicting/dangerous movement.
The owner of Dutch tech-site Tweakers.net made his Dutch farmhouse 'smart' with his own system, years ago. I think he used industrial systems like this to make it work. There is a video tour of his house online somewhere.
The esp32 (at least with the esp-idf) makes doing firmware updates like that easy; the examples use a default factory partition and two OTA partitions. Download the update to the partition then reboot when ready.
It's funny that you mention the 2 OS partition setup on routers. Google's pixel phones (at least the 2 I know from experience) have a similarly set up A-B partitioning. So, some people at google know how to set things up well. I suspect the Nest has a lot more input from marketing people than serious engineers.
Yep that's what I thought.. Or the team program this are inexperienced and they wanted to maximize profit :-( Having no read only flash for iot things and updates in restart is just dumb
Absolutely, my moto G6 Plus has A/B partitions and I believe it's been a thing for some devices since 2013... It's not 100% failproof and it's a little bit more work for me to root and put a different rom in my device, but when I screw something up and can just revert back to the partition B, that makes it all worth it.
Been around 20 years since I've written in Ladder... There's one thing I have to say after the last time I did. Any time you change the program, which isn't often, back it up... Not once, but three times in three different places. The only backup this company had when it's PLC puked was a printout from the original tech that set it up. It took me over a week to rewrite the code. I had to hunt down every sensor, motor, actuator, light, etc. One viable backup would have saved that company over $200,000 in production losses for the half hour of replacing the cpu unit alone and reloading a backup. Ladder isn't hard, but with no diagrams, no backups and a lots of modifications after the original install you can imagine it was a wire nest. That's another thing. In the box, label every wire. No labels equals a seriously bad situation. I've carried that over to server racks too. Labels are cheaper than head scratches that take hours.
Nest Thermostats don't stop working when the Internet/cloud is out, they'll still run by the last saved schedule, additionally I'm pretty certain firmware updates do not prevent it from running either as I think the baseboard (what you mount it to) has low level control over the HVAC system using PLCs as you're talking about, to prevent damage to your HVAC.
@@jefffrilot9667 They thought putting windows 10 on the network was a good idea to allow it to download security updates, which it promptly did (when everyone was sleeping). They have since figured out how to set that behavior to manual.
Agreed, happens more than it should. Tried to start up my assembly line we use once a year and oh, windows 10 updates killed half my PC's in the area. But the PLC controlled stuff works fine :)
Hey, I am leading a project for a IoT device which is basically Industrial Automation for the home. The hardware is almost complete (I only need a propper uninterrupted PSU circuit to finish, but I have prototypes for it ready). And I've hired a programmer to help out with the software. It will have wifi, Bluetooth, 1-wire and touchscreen. All plug and play several sensors and actuators with SSR and buttons and temperatsensors. It can be used completely offline. It has a sd card and a database inside. Next version will have Lua scripts on the sdcard to customize the pseudo PLC logic.
Finite state machines should be taught in every degree plan. Basically a PSA to all human beings to always account for a failure state. Everything from economics to rocket science could benefit from understanding how to make systems fail gracefully.
Always, always set that "none of the above" state to the "failsafe" condition. It'll suck when the shower turns off mid shower, but it will suck more when the whole bathroom floods because your 3 year old turned on the upstairs bathtub and wandered off.
The water valve stuff could be really useful for accessibility. I have MS and my hands haven't worked properly at multiple times in my life. Having a smart device that could open a faucet for me would have been great but we need the stability of the industrial devices like you say. I've been using some smart devices to help with my disabilities for a little while now and I really wish I could recommend them to my friends with disabilities, the potential for them to do good in this use case is astonishing... it's just so frustrating that the standards aren't there yet.
I wanted a smart thermostat, but all the problem you pointed out was blocking me from buying one....and I didn't find time to implement by own version. You should make a tutorial on how to make our own with a parts list.
I know I'm late to this, but your comment about PLC's always working and windows update struck a nerve. We use Allen Bradley stuff at my work. When the Spectre exploit happened, Windows patched. FactoryTalk then broke and wouldn't load. The PLC's worked, but the HMI was dead.
I think CANbus isn't far off from what you'd really need. OneWire is neat.... but it's also a propreitary there are sender and receiver address concepts, but it's manually assigned. there is isolation pre-defined in the standard there is noise immunity pre defined (twisted pair, differential signaling), and bus termination defined, and the output/input is just serial. CAN really is the closest you can get to ethernet on a small scale, but it does increase your per device cost. Because of the CAN receiver, and if your device doesn't natively speak CAN or serial, then you end up with attiny84's all over the place to interface your SPI or onewire devices to the CAN bus. The upside is that you can now do multiple things with a single endpoint. Hook your PIR sensor to the attiny84 along with the temperature/humidity sensors, add LDRs or photo transistors. And CAN has the bonus of being fast and bidirectional, so bundling all those sensors together with a slightly faster mcu, you could drive a small touch LCD for an interface or monitoring panel. It's not even inconceivable to use cat6 for the CAN wiring and carry power over the same cable. Keep in mind, the space station's control data bus, that all the systems talk between each other, was built with a dual redundant CAN Bus (pedantry: it is actually MIL-STD-1553, but that standard is pretty similar to CAN with a few extra specifications). Adafruit also recently demoed a way to send i2c over 100ft of cat6, which is an interesting avenue to explore but more limited.
Good show sir...your feeling about Nest is the same I have for Honeywell Home Thermostats..They are unreliable and for a large company their totalconnect server is down more than my home alarm system I got that connects to china for connecting remotely. The alarm system might go off line once or twice per year yet the total connect would go offline almost every other week and they would do maintenance whenever they wanted with no warning to the users....
I teach control engineering and mechatronics and PLC's are our bread and butter, they are incredibly simple to program and super reliable. Our students can get simple miniature factory lines running after only a few lessons. after that its just the same thing but bigger. the only real downside is the price... anything industrial costs crazy ammounts of money
It all comes down to a few things though... Your solution is more expensive, generates less revenue and doesn't offer any features a regular customer is wowed by until it's too late. In any case... I strongly agree and try to make as many things as possible work in all cases. I don't want my wall switches/buttons/etc. to stop working when my internet or home assistant is down... that's insane...
Folks, I need some tech help today. I just did my first fancy computer build, and decided to replace my Corsair PSU cables with a braided-type from a local computer store. They just happened to have some in the clearance section that were nearly $ 15 cheaper than a new sack of cables. The cables seem to fit okay with my PSU and motherboard, although I could not get the video-card cable to plug into the PSU. I flipped on the switch and nothing turned on. I put my old cables back on and flipped on the switch and nothing turned on. I think I may have accidentally put the PCI-e cable in the CPU connector. Which you would think would not be feasible. Right ?? Anyways, I can't find anything that works and I do not have any tech-savvy skills or tools. So assuming I fried the motherboard, what else did I destroy ? Both NVMe's ?? the video-card ? the Ryzen processor ? the PSU ? The computer's front switch ?? I have no more money left, and I don't have any spare parts. I can't really install a cheap power-supply, because I need long cables. I have a giant case the size of a Mini-Cooper. LOL !
I do commend you on this older video in this series. BUT I do have to note that Windows updates have shutdown factories, hospitals, railroad engines and more; and PLC logic failures have resulted in injuries and deaths on carnival rides, on amusement park rides, in elevators, in factories, in nuclear processing facilities and more.
Wow wendel I was thinking about mqtt and that whole rabbit whole there. ... But getting a video about plc with single wire bus, ladder logic, petri nets and a whole lot more - what a throwback to uni 👍😅
Hi Wendell, love the video series keep up the great work. Have a look into a standard called ‘KNX’ it’s quite expensive but it looks interesting and without me doing any deep digging looks quite robust. It also fits onto DIN rails and into breaker panels so integrates well with electrical installations
We need to get Wendell on some show like Shark Tank or The Dragon's Den with ideas like this-if only the normies saw the need… _engagement_ Edit: Watching this a year later, I am not sure if I did not pay enough attention last time but I am not sure how I did not see that wireless temp sensor actually has a MAC address and being as small as it is with a _single wire_…I am still blown away.
12:20 Bummed out that the Maxim ultrasonic sensor costs $525 (www.maximintegrated.com/en/design/reference-design-center/system-board/5969.html/tb_tab3) Haven't crunched the numbers, but would it be cheaper to connect multiple flow detectors at various locations throughout the house? (e.g. toilet, shower, etc.)
While I agree with many ideas in this video there are some things I simply can't ignore: 1:18 No, DS18B20 do not have a MAC address. They have a 64 bit UID that allows to operate multiple sensors on a one-wire bus, however that just is not a 48bit MAC address at all. 3:00 The whole argument about PLCs is just broken in itself. Firstly PLCs don't go down through firmware updates because they do usually never receive any. That is not to say that there are none available, but they are usually never installed. This is of course far from ideal because security vulnerabilities remain unfixed in most installations, which would of course be a particularly big issue with IoT devices since It would just place all "decent" IoT devices from reputabler manufacturers where cheap noname IoT is today security wise. 5:57 That is likely not the reason there is no proper interfaceing documentation at all. First of all documenting and providing interfaces takes time and thus costs money. Since this is a feature only a few ppm of the userbase would use it does not make any sense to implement it from a business perspective. Additionally it gives you a bit of the perceived "security by obscurity" advantage. Without proper (protocol) documentation security issues are less likely to be found. Not having to fix security issues results in less work and is thus favorable from a management perspective. If the protocol turns out to be insecure it is not really an issue for the manufacturer either, since that is pretty much the accepted industry standard. 14:44 No. This is just wrong on a technical level. The primary difference between ethernet magnetics and optical isolators is not that one costs more than the other. Rather they are suitable for different applications. For starters magnetic isolators can only convey changes in current (and thus applied voltage) while optical isolators can be used to isolate DC signals. Additionally optical isolators are not linear devices. In most practical applications they can only be used to transfer digital signals with relatively high voltage levels without requiring a separte preamplifier. Magnetics however offer (almost perfect) linear behavior if properly designed for the frequency of signal they are supposed to convey. Both the low power and voltage level make magnetics a better choice for isolating Ethernet. 16:10 Applying PLC concepts to consumer electronics (which home IoT devcices are) just does not work. The most important aspect of designing IoT devices is low hardware cost. This is usually achieved by having a single SoC handle all functionality. Thus it really is not possible to update any type of external control component seperately because there is none. This does also help reduce power consumption since a single SoC will consume less power than the SoC plus companion controllers. 16:30 That is actually a common design pattern in many IoT SDKs, too. I don't own a wide enough selection of IoT devices to judge how common the actual implementation of this feature is, but manufacturers are at least provided with all they need to implement it. 19:35 As usual companies build products that appeal to customers. How useful the product is is not reflected in revenue. To sell a product needs to be appealing, not necessarily useful. 20:05 The general concept of a reasonable system for smart home control is very simple. It needs to provide the exact same functionality that was there before, fulfilling all previous assertions. On top of that (all) control parameters must be adjustable externally and change of system status needs to be announced to external components. For preexisting devices that is literally all that needs to change. On top of that new components can then be integrated, adding a smart aspect by linking data, computing it and changing control parameters. However. the devil is in the details. There must be standardized protocols for both control of and data collection from devices. This protocol needs to be universal enough to represent all types of data that need to be transferred. Even worse combining all the data and making control decisions based on that data is complex. This might be the part where control theroy is most useful and that comes closest to your PLC concept. And all of that is not to speak of building an interface that a human can use wihtout going insane. However, that covers only the software part. For hardware one can either hack existing devices - there should be loads of ESP8266/ESP32 based IoT devices for almost any purpose imaginable out there - or design custom hardware. The custom hardware route is most sustainable since it could be built as open hardware, ensuring availability into the future. But designing custom hardware takes a lot of time, dedication and perseverance, thus hacking existing hardware might be more accesible. All in all it will take a lot of time and is only feasible for the most dedicated people with some hardware and software design expirience. That is likely the primary reason why even such people are ok with existing IoT solutions.
I have a weather station from Germany 1930's made by Lufft. It uses Copper coils to read temp and humidity. It is accurate to this day and i believe it does not contain any mercury which is very cool.
Ladder logic is simple and robust.. smart. It's the remote monitoring and wireless options that needs work. What about RF and Bluetooth limited range as features that prevents highjacks.
The whole "running-config" vs "startup-config" thing makes sense in this context, but is otherwise an absolutely horrendous design by Cisco, and they should be ashamed forever in history for keeping such malpractice up for decades. Companies like Juniper actually got it right with Junos.
Please link some flow sensors that you could use in the home easily. All the ones I can find are super cheap (so i worry about them working/leaking etc as they are inline), or super expensive :(
I’m hearing an opportunity for Level1 to make some deals with hardware manufacturers and tie in to OSS projects to make viable home automation products on a level of quality as the KVM stuff. If you build it with a proper design philosophy, they will come.
Pretty good introduction course to logic gates can be found in Fall Out 4, work shop. Only down side is once you get really complex systems built the game gets confused. But kind'a cool nun the less.
Really interresting, I would love some more about the hardware stuff, maybe concrete implementation examples etc :) Def. gonna check out the 1 wire bus stuff, hadn't heard about that
Can you reduce some of the security implications and garbage by our corporate overlords if you were to use something like Home Assistant for Pi and keep these devices from phoning outside your network?
Ah yes the ol copy run start. I appreciated that feature the other day when we had issues in our network Implemented a fix at every stack with a console cable since the management vlan was stuck in blocking state. Didn't have time to go back and fix it all before users would be back on the network so I didn't commit the changes in case I broke it further. Easier to tell the guys "rip the power out and back in" than "find a console cable, bring a laptop (yours better be charged!) , insert this command and don't f it up it's production time boiz"
Better use one title that is easily reconizeable for all 3 parts of this series and use subheadings for distinguishing them further. Wendel you might not like any computer or at least PCs, because most mainboards have extra chips on them that range from ARM, STM to PIC to bootstrap the CPU.
Great chat but I have some devils advocate items. So that sensor is $5. However I can get an esp for less than that and it's got the temp sensor built in.. As well as I butt ton of io Inc WiFi Ethernet. As for isolation, as you well know that can be on a separate WiFi network, logically it's the same sort of deal. I'd accept you doom, excess of processing power arguement but I'd not accept your security through ancient obscurity ;) Tbh quite happy that I just got all the esp tuya recessed lights off eBay recently. When I say all I just bought like 60+ of them and bought most of them around £3.50 so like $4.. So that's a light plus esp for not much. So that's my house lighting sorted. Loving this series btw.
I was actually looking at replacing my current thermostat with a honywell or other kinda smart but still not terribly smart device. I came across so many IOT crap I couldn't believe people actually put in their house.
PLCs have their own OSs which do get updates. You might have to take a factory down to do the upgrade. The industry is pretty religious about separating control from visualisation/analytics so in a factory you might be able to reboot windows boxes without taking the factory down. It’s not every factory, something like a car plant will have production critical windows boxes. Within the last 10 years PLC vendors have actually tried with security but before then with most systems you could just plug your laptop in and change the program or variable values. There’s a ton of legacy systems where that’s still the case. It will be a long time before secured systems are the majority. You can easily do home automation with a PLC. The traditional barrier was price but there’s affordable versions some times. I just saw this on eBay for $US170 for a plc on ethernet www.ebay.com/itm/Allen-Bradley-1763-L16AWA-Series-A-MicroLogix-1100-Controller/193719036931?epid=1179920194&hash=item2d1a8dec03:g:Jh8AAOSwSyBfhHZx Opto-isolation is sometimes built into the PLC IO but you can add it on. PLC outputs shouldn’t be the ones driving anything bigger than a status light. The idea is to use the outputs from the PLC to be the control signal to whatever is doing the control of the real power e.g. send a 24v control signal to the box that is built to control a megawatt motor. You can get a transient 600v spike off even a boring 24v solenoid for example so you’ll want to put things in place to protect the PLC controlling that.
This is why proper building automation is not done by any of these tech companies solutions, they simply are not up to the task. They have not enough features, can't be altered or expanded, they are not secure, they are not reliable, they are really expensive on big scale. Why can't we have these home scale proper automation systems from any company.
22:59 it won't in couple of months :D Using microprocessors for automation instead of microcontrollers is the same kind of mistake like Google's approach for Nest design, additional software layer that is always be prone to have bugs.
I wrote a giant comment but whittling it down to this, we need something simple, simple-simple-simple. I mean as plug and play as possible, I'm talking nintendo days of inserting a cartridge and hitting the power button simple. Including the instructions and learning, if the customer can't figure it out in 10 seconds then, even though it may be simple, if someone can't read a half-page brocure and understand everything then what is the point.
Hands-free control for a sink is actually somewhat useful? Your hands are covered in bread dough or chicken ectoplasm and you don’t want to gunk up your sink handle, but you need to wash them
I'm SO glad someone finally mentions Finite State Machines in Home Automation.
Love all the additional scenerios. Really helps apply this to my house.
Great to hear! ~Editor Amber
Hey Wendel :) In Austria is a small company called Loxone which builds a PLCs for Homeautomation. I think you might like it. Anyway nice series, I really enjoy it
Introducing Level1Things, an IoT device company that makes industrial PLC solutions available for elegant use in the home.
Welcome to This Old House - Electronics Edition with your host Wendell.
"The one wire bus will outlive x86" -- underrated statement!!
This was great. I love the ramble and just actually addressing the faults of current IoT tech in a very practical and succinct way. The practical examples are perfect.
I never really thought much about smart home, but all the scenarios you mention are great.
The big downside:
Even if off-the-shelf products would exist for this you would still need to do a lot of setup yourself. And I think that's the catch. Things like nest don't actually do much smart things but they are inexpensive in the setup. You just buy it, install it yourself and you are done and can feel good. But if you really want a smart house you have to actually think about it (or pay someone to think about it).
Every home is a bit different, there is not really a solution that fits everyone. This increases cost because it's a custom solution.
Your ramblings are my favourite thing on RUclips, I can't understate how much I enjoy this stuff!
Never thought I would hear someone talk PLCs in one of the channels I follow. I work with Allen Bradley PLCs at work, so I was primed for home automation. Currently using home assistant.
One Wire Networks get slept on. How high of a bitrate do you really need for temperature sensing anyhow? Air temps generally change slowly.
Home automation service is expected to grow but for an installer you want a standard. You present a good argument that such standards are needed.
wendell talking about plc’s, dream come true :D
Yes, exactly that. Now, how do we push this to the state its done properly?
The one wire protocol is a really neat protocol and very well thought out.
Basically all sensors on the line are always on "Listening mode" waiting for a Reset pulse from the 1-wire Master. When a reset pulse is sensed, all sensors send simultaneously a presence pulse. Then there is a Scan algorithm to detect all the devices on the line, which is amazing. It can be slow (a few seconds) to scan hundreds of devices on the line, but you only need to scan if you connect a new device. Once you have a list of devices their MAC address contain their model so you can get all their capabilities from the mac and you can program the master to adress one device at a time to get or set sensor data.
I've been around and seen a few houses with actually decently electrically and computationally educated sensor set ups and they make me so excited to have a home of my own to do it to. Along with this series which has sent me on nostalgia lane to these set ups with your inspirational ideas. Beautiful !
Also if you're like me you have everything on two different flash drives or SD card so when it comes to 10 years down the line and the SD card of my Pi dies I have a copy of it on my computer and on another SD card. Redundancy never hurt anyyyybody
I’m a PLC guy for work and I have dread about the continued push for IoT
As someone who has been forced to work with DeltaV as an ICS you are correct, Windows won't update while everything is running, but it will break everything you can think of and more when trying to apply simple updates. Going so far as to have to rebuild machines from scratch just because Windows doesn't know how to update right anymore.
Wendell, as a PLC programmer of 8 years and counting, definitely saw the potential of home automation with a PLC. Heard of some colleagues who rewired their whole house to be controlled off of a PLC. Every outlet, every light switch. Sounded awesome at first, but then questioned what the next home owner will do with it... How do you future-proof it. I have to maintain a windows XP virtual desktop so I can reprogram some of the 20-year old PLCs that are still in operation. Also, BTW PLC firmware changes do shut down the assembly line while it is downloading, but is always a manual operation and can only be done one PLC at a time. Pros and Cons to this, the engineer is dedicated to making sure the new firmware works before moving onto the next PLC but also have every possible version of firmware on each machine across the plant, because who has time for that. PLC tech is also about 10 years behind consumer technology. Gigabit ethernet isn't really a thing for PLCs, partially because the bandwidth isn't needed. Partially because every PLC brand had their own network implementation. (See Data-Highway, Remote-IO, Controlnet, Devicenet, modbus, CAN, etc.)
Personally, I'm interested in wireless devices. At least for now being in an apartment where I can't leave permanent changes to the building. Wifi devices are super convenient, and we know convenience to the consumer usually wins out. Zigbee and Z-wave haven't been too bad either. Would be awesome if apartments came pre-wired with ethernet so I could setup POE devices.
I always wondered about implementing a commercial building automation solution rather than an industrial solution. Like some system that I could put temperature sensors in every room and had control of the air ducts and the HVAC to control every room to the temperatures that I want them to be at. Industrial PLCs are waayyy too pricey for a home use.
As someone who has worked for a for a HVAC company and has programmed PLCs, I have a much more grim look on PLCs. They usually don't crash because of how simple the programs are, but they certainly requires maintenance. Preprogrammed PLCs sometimes do have software issues though and requires rebooting/updating/etc.
Industrial systems constantly have issues but there is people which monitors and deals with these issues. It is hooked up to hardware which has non-trivial interactions with the real world after all. In a small housing complex it might just be a janitor which calls for assistance when pressing the "Stop" and "Start" buttons doesn't fix it (and a lot of times it is just because they don't understand what is going on). Larger organizations will have a team of people which can actually fix hardware issues, with alarms set up to fix common problems before it becomes an issue. Live updates are possible to some degree, but normally stuff is shut down. Either because it is not an issue if it is offline for a few minutes, or because you are doing it while that section is not in service (say after work hours or doing scheduled maintenance).
The grim look is this: They are usually programmed by people who knows nothing about computers. They are managed by people who know even less. And there is usually no security. The PLCs are usually not designed to be secure and the people who set them up knows nothing about it. By default many do not have a password, and if they do it is the default one because nobody changed it. (Many PLCs run Linux btw.) And yes, they are often directly connected to the internet.
One of the hospitals we worked for had everything on a separate network, but anyone could walk in from the street to the "server" room and walk away with the computer and nobody would notice. We also once used the password on a sticky note we found to access one of the systems we usually did not work on. Another hospital had a lot of stuff break due to a power outage (including something I worked on because I forgot to move a switch, causing the CT scanners not operating because they depended on the ventilation working). And at a third hospital one of their technicians died in a pit of boiling water. These kind of things happens all the time, you usually just don't hear about it.
The PJON one-wire-bus is also pretty neat, and the standard also supports infrared/visible light transmission.
I have a friend who had an angle stop burst while he was on vacation. It flooded the entire house causing over $100k in damage. It took him over 4 years of battling the insurance company in court (he had restoration level insurance, which is far beyond what most people have and they didn't want to payout). With a moisture sensor or flow control cutoff setup like described this would have saved him years of headaches.
Thanks for rambling. Toilet handles not returning [flapper doesn't close] happens more often around here. I like the idea of a shutoff for that.
Over the lockdown I was actually looking into PLC programming to use on a swamp cooler. I grew up just outside of Las Vegas where you can run a swamp cooler as your main cooling source ~6 months out of the year. During the summer (with the exception of monsoon season during Jul/Aug), you are running it full tilt all the time, so control isn't as important, but the beginning (Feb/Mar) and end (Oct/Nov) of the cooling season, requires more intelligence than the commercially available thermostats can provide. That translates into controlling the pump and the fan controls manually based on time-of-day and outdoor weather conditions, which can get tedious. I haven't yet started designing it yet but I have decided to build a swamp cooler controller for my parents' house using some variety of PLC.
The water manifold control is pretty smart. But my dog has her favorite spot and it just so happens to be in a bathroom. But I'm sure there is a way to figure out how to exclude the animals. But, for the most part I have zero desire for IOT running my house via google home or feeding more data than necessary to the overlords at Google. Thanks for all the great informative videos Wendell.
With a flow sensor and motion detector you can do other things as well. Like if there is motion and water flow for more than say a minute, kick on the bathroom ventilation fan. Can also do a hot water recirc loop and have that turn on whenever there is motion in that bathroom, so when someone turns on the faucet or shower it already has hot water right there ready to go, and you don’t have to wait for it.
Nice series, thank you for making it!
I personally would be quite interested about a deeper dive into decent alarm system: as in, with decency, without cloud or phoning home or an external company having data or ownership over your home/business system. What exist, what's possible, where are the pitfalls, examples, and so on.
In Industry wifi is no-go, no direct internet access-vlan and vpn to specific machines/locations and all wired, redundant and failsafe with PLC's.
For home we don't need cloud at all. And standard for homes should be derived from good industry practices with some cost considerations.
I worked on tram systems, all the safety critical stuff was programmed into the PLCs, this then allowed the developers to build all the high level management stuff without the worry of creating a conflicting/dangerous movement.
The owner of Dutch tech-site Tweakers.net made his Dutch farmhouse 'smart' with his own system, years ago. I think he used industrial systems like this to make it work. There is a video tour of his house online somewhere.
Don't you dare interrupt my long showers, Wendell! ;)
The esp32 (at least with the esp-idf) makes doing firmware updates like that easy; the examples use a default factory partition and two OTA partitions. Download the update to the partition then reboot when ready.
It's funny that you mention the 2 OS partition setup on routers. Google's pixel phones (at least the 2 I know from experience) have a similarly set up A-B partitioning. So, some people at google know how to set things up well. I suspect the Nest has a lot more input from marketing people than serious engineers.
Yep that's what I thought.. Or the team program this are inexperienced and they wanted to maximize profit :-(
Having no read only flash for iot things and updates in restart is just dumb
Absolutely, my moto G6 Plus has A/B partitions and I believe it's been a thing for some devices since 2013... It's not 100% failproof and it's a little bit more work for me to root and put a different rom in my device, but when I screw something up and can just revert back to the partition B, that makes it all worth it.
Been around 20 years since I've written in Ladder... There's one thing I have to say after the last time I did. Any time you change the program, which isn't often, back it up... Not once, but three times in three different places. The only backup this company had when it's PLC puked was a printout from the original tech that set it up. It took me over a week to rewrite the code. I had to hunt down every sensor, motor, actuator, light, etc. One viable backup would have saved that company over $200,000 in production losses for the half hour of replacing the cpu unit alone and reloading a backup. Ladder isn't hard, but with no diagrams, no backups and a lots of modifications after the original install you can imagine it was a wire nest. That's another thing. In the box, label every wire. No labels equals a seriously bad situation. I've carried that over to server racks too. Labels are cheaper than head scratches that take hours.
Thank you so much. Really appreciate it.
I hope that you continue this educational series
You should really place a sticker over your laptop's logo. It's kinda bizzare :D
Nest Thermostats don't stop working when the Internet/cloud is out, they'll still run by the last saved schedule, additionally I'm pretty certain firmware updates do not prevent it from running either as I think the baseboard (what you mount it to) has low level control over the HVAC system using PLCs as you're talking about, to prevent damage to your HVAC.
"Can you imagine if when a windows update happens a factory line shuts down?" I can tell you from experience, this does happen. Blame weak IT.
"But we tested it before we pushed the update..."
@@jefffrilot9667 They thought putting windows 10 on the network was a good idea to allow it to download security updates, which it promptly did (when everyone was sleeping). They have since figured out how to set that behavior to manual.
Agreed, happens more than it should. Tried to start up my assembly line we use once a year and oh, windows 10 updates killed half my PC's in the area. But the PLC controlled stuff works fine :)
Hey, I am leading a project for a IoT device which is basically Industrial Automation for the home. The hardware is almost complete (I only need a propper uninterrupted PSU circuit to finish, but I have prototypes for it ready). And I've hired a programmer to help out with the software. It will have wifi, Bluetooth, 1-wire and touchscreen. All plug and play several sensors and actuators with SSR and buttons and temperatsensors. It can be used completely offline. It has a sd card and a database inside. Next version will have Lua scripts on the sdcard to customize the pseudo PLC logic.
It will have a webserver with PWA version for smartphone use.
Finite state machines should be taught in every degree plan. Basically a PSA to all human beings to always account for a failure state. Everything from economics to rocket science could benefit from understanding how to make systems fail gracefully.
Always, always set that "none of the above" state to the "failsafe" condition. It'll suck when the shower turns off mid shower, but it will suck more when the whole bathroom floods because your 3 year old turned on the upstairs bathtub and wandered off.
The water valve stuff could be really useful for accessibility. I have MS and my hands haven't worked properly at multiple times in my life. Having a smart device that could open a faucet for me would have been great but we need the stability of the industrial devices like you say. I've been using some smart devices to help with my disabilities for a little while now and I really wish I could recommend them to my friends with disabilities, the potential for them to do good in this use case is astonishing... it's just so frustrating that the standards aren't there yet.
I wanted a smart thermostat, but all the problem you pointed out was blocking me from buying one....and I didn't find time to implement by own version. You should make a tutorial on how to make our own with a parts list.
I know I'm late to this, but your comment about PLC's always working and windows update struck a nerve. We use Allen Bradley stuff at my work. When the Spectre exploit happened, Windows patched. FactoryTalk then broke and wouldn't load. The PLC's worked, but the HMI was dead.
I think CANbus isn't far off from what you'd really need. OneWire is neat.... but it's also a propreitary
there are sender and receiver address concepts, but it's manually assigned.
there is isolation pre-defined in the standard
there is noise immunity pre defined (twisted pair, differential signaling), and bus termination defined, and the output/input is just serial.
CAN really is the closest you can get to ethernet on a small scale, but it does increase your per device cost. Because of the CAN receiver, and if your device doesn't natively speak CAN or serial, then you end up with attiny84's all over the place to interface your SPI or onewire devices to the CAN bus. The upside is that you can now do multiple things with a single endpoint. Hook your PIR sensor to the attiny84 along with the temperature/humidity sensors, add LDRs or photo transistors.
And CAN has the bonus of being fast and bidirectional, so bundling all those sensors together with a slightly faster mcu, you could drive a small touch LCD for an interface or monitoring panel.
It's not even inconceivable to use cat6 for the CAN wiring and carry power over the same cable.
Keep in mind, the space station's control data bus, that all the systems talk between each other, was built with a dual redundant CAN Bus (pedantry: it is actually MIL-STD-1553, but that standard is pretty similar to CAN with a few extra specifications).
Adafruit also recently demoed a way to send i2c over 100ft of cat6, which is an interesting avenue to explore but more limited.
Good show sir...your feeling about Nest is the same I have for Honeywell Home Thermostats..They are unreliable and for a large company their totalconnect server is down more than my home alarm system I got that connects to china for connecting remotely. The alarm system might go off line once or twice per year yet the total connect would go offline almost every other week and they would do maintenance whenever they wanted with no warning to the users....
14:42 you have mentioned its optically isolated. Actually it is an transformer (1000Base-T Magnetic Module)
Turn to 11 video! Great insights and simple yet useful logic analysis. Thanks
where is the forum post for a detailed guide for how to setup the temp sensors around the home and tie it into your thermostat?
the way Wendell talks about the bathroom flooding makes me think his tub doesnt have an emergency overflow like literally every tub i ever saw XD
I teach control engineering and mechatronics and PLC's are our bread and butter, they are incredibly simple to program and super reliable. Our students can get simple miniature factory lines running after only a few lessons. after that its just the same thing but bigger. the only real downside is the price... anything industrial costs crazy ammounts of money
Thank you Wendell! This is genuinely useful.
Glad it was helpful! ~Editor Amber
It all comes down to a few things though... Your solution is more expensive, generates less revenue and doesn't offer any features a regular customer is wowed by until it's too late.
In any case... I strongly agree and try to make as many things as possible work in all cases. I don't want my wall switches/buttons/etc. to stop working when my internet or home assistant is down... that's insane...
PLCs :D this video should be labeled 'back to basics'
haven't touch one of those in 10 years
Folks, I need some tech help today. I just did my first fancy computer build, and decided to replace my Corsair PSU cables with a braided-type from a local computer store. They just happened to have some in the clearance section that were nearly $ 15 cheaper than a new sack of cables. The cables seem to fit okay with my PSU and motherboard, although I could not get the video-card cable to plug into the PSU. I flipped on the switch and nothing turned on. I put my old cables back on and flipped on the switch and nothing turned on. I think I may have accidentally put the PCI-e cable in the CPU connector. Which you would think would not be feasible. Right ?? Anyways, I can't find anything that works and I do not have any tech-savvy skills or tools.
So assuming I fried the motherboard, what else did I destroy ? Both NVMe's ?? the video-card ? the Ryzen processor ? the PSU ? The computer's front switch ?? I have no more money left, and I don't have any spare parts.
I can't really install a cheap power-supply, because I need long cables. I have a giant case the size of a Mini-Cooper. LOL !
I do commend you on this older video in this series. BUT I do have to note that Windows updates have shutdown factories, hospitals, railroad engines and more; and PLC logic failures have resulted in injuries and deaths on carnival rides, on amusement park rides, in elevators, in factories, in nuclear processing facilities and more.
Wow wendel I was thinking about mqtt and that whole rabbit whole there. ... But getting a video about plc with single wire bus, ladder logic, petri nets and a whole lot more - what a throwback to uni 👍😅
You get a like for DS18B20. Nash, just kidding. Much more than that. This channel is an oasis of deeply thoughtful tech insight.
Thank you!
Forget the Level 1 KVM, time to expand into home sensors!
Hi Wendell, love the video series keep up the great work. Have a look into a standard called ‘KNX’ it’s quite expensive but it looks interesting and without me doing any deep digging looks quite robust. It also fits onto DIN rails and into breaker panels so integrates well with electrical installations
We need to get Wendell on some show like Shark Tank or The Dragon's Den with ideas like this-if only the normies saw the need…
_engagement_
Edit: Watching this a year later, I am not sure if I did not pay enough attention last time but I am not sure how I did not see that wireless temp sensor actually has a MAC address and being as small as it is with a _single wire_…I am still blown away.
12:20 Bummed out that the Maxim ultrasonic sensor costs $525 (www.maximintegrated.com/en/design/reference-design-center/system-board/5969.html/tb_tab3)
Haven't crunched the numbers, but would it be cheaper to connect multiple flow detectors at various locations throughout the house? (e.g. toilet, shower, etc.)
While I agree with many ideas in this video there are some things I simply can't ignore:
1:18 No, DS18B20 do not have a MAC address. They have a 64 bit UID that allows to operate multiple sensors on a one-wire bus, however that just is not a 48bit MAC address at all.
3:00 The whole argument about PLCs is just broken in itself. Firstly PLCs don't go down through firmware updates because they do usually never receive any. That is not to say that there are none available, but they are usually never installed. This is of course far from ideal because security vulnerabilities remain unfixed in most installations, which would of course be a particularly big issue with IoT devices since It would just place all "decent" IoT devices from reputabler manufacturers where cheap noname IoT is today security wise.
5:57 That is likely not the reason there is no proper interfaceing documentation at all. First of all documenting and providing interfaces takes time and thus costs money. Since this is a feature only a few ppm of the userbase would use it does not make any sense to implement it from a business perspective. Additionally it gives you a bit of the perceived "security by obscurity" advantage. Without proper (protocol) documentation security issues are less likely to be found. Not having to fix security issues results in less work and is thus favorable from a management perspective. If the protocol turns out to be insecure it is not really an issue for the manufacturer either, since that is pretty much the accepted industry standard.
14:44 No. This is just wrong on a technical level. The primary difference between ethernet magnetics and optical isolators is not that one costs more than the other. Rather they are suitable for different applications. For starters magnetic isolators can only convey changes in current (and thus applied voltage) while optical isolators can be used to isolate DC signals. Additionally optical isolators are not linear devices. In most practical applications they can only be used to transfer digital signals with relatively high voltage levels without requiring a separte preamplifier. Magnetics however offer (almost perfect) linear behavior if properly designed for the frequency of signal they are supposed to convey. Both the low power and voltage level make magnetics a better choice for isolating Ethernet.
16:10 Applying PLC concepts to consumer electronics (which home IoT devcices are) just does not work. The most important aspect of designing IoT devices is low hardware cost. This is usually achieved by having a single SoC handle all functionality. Thus it really is not possible to update any type of external control component seperately because there is none. This does also help reduce power consumption since a single SoC will consume less power than the SoC plus companion controllers.
16:30 That is actually a common design pattern in many IoT SDKs, too. I don't own a wide enough selection of IoT devices to judge how common the actual implementation of this feature is, but manufacturers are at least provided with all they need to implement it.
19:35 As usual companies build products that appeal to customers. How useful the product is is not reflected in revenue. To sell a product needs to be appealing, not necessarily useful.
20:05 The general concept of a reasonable system for smart home control is very simple. It needs to provide the exact same functionality that was there before, fulfilling all previous assertions. On top of that (all) control parameters must be adjustable externally and change of system status needs to be announced to external components. For preexisting devices that is literally all that needs to change. On top of that new components can then be integrated, adding a smart aspect by linking data, computing it and changing control parameters. However. the devil is in the details. There must be standardized protocols for both control of and data collection from devices. This protocol needs to be universal enough to represent all types of data that need to be transferred. Even worse combining all the data and making control decisions based on that data is complex. This might be the part where control theroy is most useful and that comes closest to your PLC concept. And all of that is not to speak of building an interface that a human can use wihtout going insane. However, that covers only the software part. For hardware one can either hack existing devices - there should be loads of ESP8266/ESP32 based IoT devices for almost any purpose imaginable out there - or design custom hardware. The custom hardware route is most sustainable since it could be built as open hardware, ensuring availability into the future. But designing custom hardware takes a lot of time, dedication and perseverance, thus hacking existing hardware might be more accesible. All in all it will take a lot of time and is only feasible for the most dedicated people with some hardware and software design expirience. That is likely the primary reason why even such people are ok with existing IoT solutions.
Wendell, you really need to take a look at home assistant. Cloudless capabilities, and various wireless protocols are supported
Nice comparison of industrial and home control alternatives!
I have a weather station from Germany 1930's made by Lufft. It uses Copper coils to read temp and humidity. It is accurate to this day and i believe it does not contain any mercury which is very cool.
awesome channel buddy, love your intelligent and dry presentation style, great work you got a keen subscriber.
Ladder logic is simple and robust.. smart. It's the remote monitoring and wireless options that needs work. What about RF and Bluetooth limited range as features that prevents highjacks.
The whole "running-config" vs "startup-config" thing makes sense in this context, but is otherwise an absolutely horrendous design by Cisco, and they should be ashamed forever in history for keeping such malpractice up for decades. Companies like Juniper actually got it right with Junos.
Please link some flow sensors that you could use in the home easily. All the ones I can find are super cheap (so i worry about them working/leaking etc as they are inline), or super expensive :(
L1 has the best rambles
Wouldn’t a fair comparison be a PLC firmware update to a Nest Thermostat Firmware update?
Have you looked into LoraWAN and helium?
I like these videos Wendell.
Thank you! ~Editor Amber
I’m hearing an opportunity for Level1 to make some deals with hardware manufacturers and tie in to OSS projects to make viable home automation products on a level of quality as the KVM stuff. If you build it with a proper design philosophy, they will come.
"Yeah, that doesn't happen. CAUSE THEY'RE NOT STUPID."
Hilarious, Wendell
Pretty good introduction course to logic gates can be found in Fall Out 4, work shop. Only down side is once you get really complex systems built the game gets confused. But kind'a cool nun the less.
Really interresting, I would love some more about the hardware stuff, maybe concrete implementation examples etc :)
Def. gonna check out the 1 wire bus stuff, hadn't heard about that
Thank you and I'm glad you enjoyed it! ~Editor Amber
3:22 - That’s right, it’s because of Allen-Friggin-Bradley baby!
Can you reduce some of the security implications and garbage by our corporate overlords if you were to use something like Home Assistant for Pi and keep these devices from phoning outside your network?
Thanks very much for this talk and creative ideas.
Ah yes the ol copy run start. I appreciated that feature the other day when we had issues in our network Implemented a fix at every stack with a console cable since the management vlan was stuck in blocking state. Didn't have time to go back and fix it all before users would be back on the network so I didn't commit the changes in case I broke it further. Easier to tell the guys "rip the power out and back in" than "find a console cable, bring a laptop (yours better be charged!) , insert this command and don't f it up it's production time boiz"
Without reinventing the wheel, ELK M1 (ladder logic built in, wired and/or wireless) $$$ though.
Better use one title that is easily reconizeable for all 3 parts of this series and use subheadings for distinguishing them further.
Wendel you might not like any computer or at least PCs, because most mainboards have extra chips on them that range from ARM, STM to PIC to bootstrap the CPU.
Great chat but I have some devils advocate items.
So that sensor is $5. However I can get an esp for less than that and it's got the temp sensor built in..
As well as I butt ton of io Inc WiFi Ethernet.
As for isolation, as you well know that can be on a separate WiFi network, logically it's the same sort of deal.
I'd accept you doom, excess of processing power arguement but I'd not accept your security through ancient obscurity ;)
Tbh quite happy that I just got all the esp tuya recessed lights off eBay recently.
When I say all I just bought like 60+ of them and bought most of them around £3.50 so like $4..
So that's a light plus esp for not much.
So that's my house lighting sorted.
Loving this series btw.
I was actually looking at replacing my current thermostat with a honywell or other kinda smart but still not terribly smart device.
I came across so many IOT crap I couldn't believe people actually put in their house.
Do you know where your refrigerator is?
PLCs have their own OSs which do get updates. You might have to take a factory down to do the upgrade. The industry is pretty religious about separating control from visualisation/analytics so in a factory you might be able to reboot windows boxes without taking the factory down. It’s not every factory, something like a car plant will have production critical windows boxes.
Within the last 10 years PLC vendors have actually tried with security but before then with most systems you could just plug your laptop in and change the program or variable values. There’s a ton of legacy systems where that’s still the case. It will be a long time before secured systems are the majority.
You can easily do home automation with a PLC. The traditional barrier was price but there’s affordable versions some times. I just saw this on eBay for $US170 for a plc on ethernet www.ebay.com/itm/Allen-Bradley-1763-L16AWA-Series-A-MicroLogix-1100-Controller/193719036931?epid=1179920194&hash=item2d1a8dec03:g:Jh8AAOSwSyBfhHZx
Opto-isolation is sometimes built into the PLC IO but you can add it on. PLC outputs shouldn’t be the ones driving anything bigger than a status light. The idea is to use the outputs from the PLC to be the control signal to whatever is doing the control of the real power e.g. send a 24v control signal to the box that is built to control a megawatt motor. You can get a transient 600v spike off even a boring 24v solenoid for example so you’ll want to put things in place to protect the PLC controlling that.
Keep this series coming! Love it! Check out Home Assistant if you haven’t already
Was that an aside to the side?
In ethernet card these are not optical isolators, these are transformers, so there is a galvanic isolation but not by optical means
This is why proper building automation is not done by any of these tech companies solutions, they simply are not up to the task. They have not enough features, can't be altered or expanded, they are not secure, they are not reliable, they are really expensive on big scale. Why can't we have these home scale proper automation systems from any company.
@12:40 either grandma fell in the shower, or the teen is taking way too long, shut the water off.
Really cool. Thank you for the ramble
Love your rambles
22:59 it won't in couple of months :D
Using microprocessors for automation instead of microcontrollers is the same kind of mistake like Google's approach for Nest design, additional software layer that is always be prone to have bugs.
Wendell saying "Carnies" really made me rofl.
I wrote a giant comment but whittling it down to this, we need something simple, simple-simple-simple. I mean as plug and play as possible, I'm talking nintendo days of inserting a cartridge and hitting the power button simple. Including the instructions and learning, if the customer can't figure it out in 10 seconds then, even though it may be simple, if someone can't read a half-page brocure and understand everything then what is the point.
"PLCs aren't a dumpster fire" me knee deep in a dumpster fire of a factory. "I disagree"
Hands-free control for a sink is actually somewhat useful? Your hands are covered in bread dough or chicken ectoplasm and you don’t want to gunk up your sink handle, but you need to wash them
Maybe time to revisit powerline ethernet? every home already has a electrical system.