Ankündigung

Einklappen
Keine Ankündigung bisher.

Migration von einem alten benutzerdefinierten System

Einklappen
X
 
  • Filter
  • Zeit
  • Anzeigen
Alles löschen
neue Beiträge

    Migration von einem alten benutzerdefinierten System

    I hope it’s OK to post in English. Ich sprecke ihre Sprache nicht gut.

    Also, I apologize for the size of this post. But I have a really big job ahead ;-)

    We have a 22 year old custom automation system. The company that created it went out of business 7 or 8 years ago.

    Since that time, I’ve been self supporting. I’ve been in software development for over 35 years, and have the source code of the system. Luckily the system has been very reliable, and most of the parts that have needed replacing (like PSUs) have been available.

    But eventually the custom hardware will break and I might not be able to repair it. Also, some day I will retire and we will likely sell this house. As it is the current automation system is a huge negative (unless we can sell to another engineer who likes to tinker with old technology).

    I’ve been researching possible system replacements, which led me to look at a number of products (including Control4). But none of them really line up with our use case.

    I’m interested in KNX because I think it’s the closest fit. Also it will let me still tinker without being locked into a closed platform. Further, when we sell our house the owner can still choose to move to C4 or a similar product and use that as the server that integrates with the KNX system.

    First I should mention some characteristics of our current system so that the problem set is clear:
    • Our main server is a PPC Mac desktop running OS 9. It provides a time base, night/day decisions, a user database of card key access, and other miscellaneous functionality.
    • The DCU units that do the actual work are Motorola 68k based VERO backplane units. They can largely operate independently of the main server.
    • Most of our light switches are dry contacts connected from the switch plates via one or more CAT5 cables. We have 58 switch plates and 106 switches.
    • The switches are intermittent contacts that can be pressed up and down, meaning that each switch results in two inputs to the system (on and off).
    • Most of the lights are controlled by a DCU that translates network commands to DMX protocol to a set of 20 commercial dimmer packs. Luminosity ramp up and down is very smooth.
    • One DCU handles relays, for outputs like fans and non-dimmable lights. We have around 50 relays.
    • One DCU handles alarm inputs, including motion sensors (used to automate the lights) and magnetic reed window/door sensors.
    • We have a normal burglar alarm system, but our alarm DCU manages all of the inputs. This means that from the alarm’s perspective we have a single virtual zone, plus fire sensors and manual fire alarm pulls.
    • We have 122 alarm inputs, which are a combination of motion sensors, magnetic window alarm contacts, and glass break detectors.
    • Smoke detectors are managed by the alarm system itself, not the automation system.
    • We enable and disable the alarm via touch panels (amber monochrome touch screens with laughably low resolution).
    • We have an exterior light sensor that determines when it’s night or day (with a different set of lighting behaviors).
    • We enter the house with HID card key access sensors.
    • We have Nest cameras, including a door cam.
    • The automation company never delivered integration with our HVAC. Today we have Ecobee and are pretty happy.
    • The automation company never delivered pool integration. Instead I have an Autelus unit that provides a web interface (and a restful API).
    • Most of the components of the system communicate via AppleTalk (yikes!)
    • Making binary changes to the system involves breaking out an ancient PPC iMac (or similar) and rebuilding with Code Warrior. So you can guess how often this happens ;-)
    Here’s the features of our system that are important to us:
    • We hardly ever hit any light switches. I might touch 5 in a day.
    • Our light sensors decide when to activate light groups based on the internal motion sensors.
    • We rely heavily on the motion sensors (especially at night).
    • We can change local dimming by holding the switches up or down.
    • We can override the local light group behavior by double clicking on or off switches in the area. Our default values are decent, so changing the behavior locally is rare.
    Of course we’d love to be able to do more, and we can with a modern system like KNX. Today we can’t even construct scenes. The automation company never delivered that, and I didn’t bother to implement it myself.

    Here’s some basic goals for a possible migration to KNX:
    • I’d like to keep as much of our infrastructure as makes sense, including the wired switches, relays, alarm sensors (motion, opening, glass break).
    • I’d like to change out some of the switches to nice KNX switches, if it’s possible (there’s mixed information about being able to drive KNX over CAT5).
    • Clearly we’ll have to throw out our touch screens, so replacing some of them with KNX units would be required.
    I’ve looked into the various product offerings. It’s nice that there are so many choices.

    Power

    Several power supply models (like from ABB) support a voltage range that includes 120v (I’m in America), so no problem there.

    Inputs/outputs

    There are some interesting mass input devices that handle dry contacts. The KNX Multi IO 570 from Weinzierl is particularly interesting because it supports lots of inputs and outputs (48), and it also supports 24v outputs (which I can use to drive our existing relays).

    Weinzierl also has a DMX interface, which we will need to drive the DMX dimmer packs. Looks like Eelectron makes one too, but it doesn’t appear to be rail mounted.

    Alarm

    I’m a bit concerned about our alarm motion sensors. They’re powered by 12v DC. When “closed” they output a little under 5v. When they detect motion, the voltage goes up to 10v briefly.

    While there are KNX analogue input modules that sense 10v DC, they also seem to require a 24v AC PSU. I think the intent is to power the sensors. Perhaps I can ignore this since I’m already powering all the motion sensors separately with 12v?

    I’m also a bit unsure about the smoke detectors. Ours are very old and need to be replaced anyway. The wiring between them is 14 gauge romex, so it might be possible to move to KNX smoke detector units. But since the romex is unshielded, this might be a poor idea.

    Also, I don’t know how we could set up monitoring of the alarm system (which we have today).

    As an alternative, I could move to a newer IP based alarm system that integrates with with KNX (like Konnected). It could handle the smoke detectors and provide a single zone that the KNX system can trigger (like a phantom window) when there’s a break in.

    If it’s too hard to integrate the alarm inputs with KNX, then perhaps I would flip the problem and let the Konnected system handle all the inputs, providing information to something like Home Assistant to tell the KNX system when motion sensors are triggered. I’m pretty sure this would work but it feels like a potentially brittle design.

    Access

    We use card key access HID readers and electric door strikes to enter the house. Basically our automation system works in conjunction with a server that maintains a list of valid cards.

    I haven’t found a clear integration path to KNX. Although there are lots of DIY projects that integrate into various automation servers, so perhaps that’s possible.

    As an alternative, I might move to using KNX fingerprint sensors instead of card readers, if we can reliably connect them via our existing CAT5 wiring.

    Cameras

    While we hardwired for cameras (coax) we never put them in. But eventually we installed Nest, after having a decent experience with it elsewhere.

    I realize now that Google has been a poor development partner. It sounds like that’s changing somewhat and it’s possible to get developer access via a small ($5) fee. Much less than the cost of ETS ;-)

    Anyway, if Nest cannot be integrated, I’m happy to move to different cameras.

    HVAC

    Hopefully there’s some way to integrate our Ecobee’s with the automation system. But perhaps that’s more of a server issue and unrelated to KNX. I’m interested in Home Assistant as the server.

    Pool controller

    As mentioned, I have an Autelus unit that provides a web interface (and a restful API). I’ll probably have to integrate that with the server. Again, not a KNX issue. I’m not interested in trying to replace my pool controller.

    Garage experiment

    Assuming that I can figure out these remaining problems, my plan is to do a test migration of our detached garage/guestroom first (which is is a much smaller project). The garage’s current light manager DCU is offline now anyway after our power company accidentally overvolted our panel (and 53 of our neighbors). I can probably get some of the parts used on eBay to reduce the cost of the experiment.

    Here’s a summary of the garage infrastructure:
    • 4 light switches, so 8 dry contact inputs.
    • 7 alarm motion sensors.
    • 7 magnetic reed alarm door/window openings.
    • 19 relays, but only a very small subset are actually used now. I could get by with a lot less for the experiment.
    • 2 commercial DMX dimmer packs.
    I haven’t been able to find a good match for a small number of outputs that drive 24v, so I might have to forget about keeping my garage relays. I could could use a switch actuator in that case (I’ve found some rated for in my voltage range).

    If you made it this far, thank you for your persistence ;-)

    Hopefully the above makes sense. I’d appreciate any suggestions regarding this approach.

    Thanks in advance.

    Tschüss!

    #2
    Hi Alen, welcome to the club

    quite a bit for your first post, and a pretty big project, which lies ahead of you.

    Changing from one system to another, with a different infrastructure is a difficult job and needs good planing. Many of you systems are unknown here, and systems, we combine, should be hard to get in the USA.

    The changeover has to be well planed. KNX Partners in California are rare, support from here is chalanging, just the 9 hrs time difference makes it complicated.

    You should try to find someone, with whom you can discuss problems and start making a project plan, with small steps, so you home will be working even when things are changed.

    I wish you all the success you need,
    Regards Florian

    Kommentar

    Lädt...
    X