I have a Lutron Series 8 processor with serial ports and an Ethernet port. When I telnet to the Ethernet port it presents the same text interface as a direct connection on a serial port.
Is there a way to configure the Lutron binding to connect to this Ethernet interface? Or do I need to set up a serial adapter?
The Series 8 processor is Lutron’s predecessor to the QS processor. I’m trying to port an AMX-based automation to OH so I can finally retire the AMX system.
OH now talking to the Lutron Processor. Bit of plumbing since OH is running in a VirtualBox VM on a computer several rooms from the processor, but managed to get a remote serial port working and plumbed from the VM to the host.
Next question is where does IntegrationId come from? My Lutron config software shows processor/module/link/device but no ids.
Also curious if anyone knows why a scan returns a bunch of load modules and doesn’t see all the keypads.
Just to close this out after another day of working with it.
The Lutron binding discovers and supports dimmer modules (load modules) for Series 8 Processors. It does not support keypad bindings, shades or contact closures as far as I can tell. Which explains why the scan only returns dimmer modules. I can work with this but it means duplicating a lot of the Lutron configuration inside OH scripts.
Also more general issue is serial passthrough on VirtualBox was losing characters. My configuration is OH on Ubuntu hosted on VirtualBox on Windows 10 on a NUC. The serial port is supported remotely on a Moxa nPort ethernet to serial converter.
I initially configured a virtual driver on the Windows host and then configured a serial port on VirtualBox. That worked for receive but got jammed up on transmit.
I switched the configuration to the Moxa Linux Kernel driver in the OH linux instance and that appears to work better.
I run this configuration because it’s great to be able to spin up a new OH instance for testing without messing with a production instance. It’s also easy to bring up companion services, like MQTT, for testing without interfering with any production versions.
Integration IDs are not a thing with Series 8. I address the dimmer modules like this [01:01:02:06:04]. HW Illuminations lists all the addresses in the Processor Link Report.
I did some work recently in the Lutron binding, changing the way that Keypads are defined and how button presses are defined. I used Claude Code.
I asked Claude Code to read this thread and if it could see a way forward. Here is it’s response
Hi Asher,
I’ve been contributing to the Lutron binding recently (most recently making keypad and Pico buttons fire events), and your thread caught my eye. My own setup is a Caseta Pro hub with Pico remotes and plug-in dimmers, so I don’t have first-hand Series 8 experience, but I’ve been reading the binding source and Lutron’s HomeWorks RS-232 guide.
IntegrationIDs. You’re right that Series 8 doesn’t use them. The HomeWorks RS-232 things are addressed by address, e.g. [01:01:03:02:04], which matches the format in your Processor Link Report.
Why only dimmers show up. The binding’s README lists only the processor and dimmers for the legacy HomeWorks RS-232 support, with keypads and keypad LEDs listed as future work. So your scan is behaving as designed.
If you’d like keypad support. The protocol guide defines keypad press, release, hold and double-tap events, so adding keypads looks feasible. To scope it, it would help to know:
What does your AMX system do with the Lutron today (for example, react to keypad presses and drive dimmers)?
Which keypad models do you have? The part numbers are on the faceplates or in your Processor Link Report.
Do you need keypad LED feedback, contact closures or shades, or only button events?
If you telnet to the processor’s Ethernet port, what text appears on connect?
Since I only have Caseta Pro hardware, I couldn’t test any Series 8 changes myself. I’d need someone with a processor to try them. Would you be willing?
With OH bindings I usually try them before I go deep into the README because I’ve found that often the hard part is getting the bridges to communicate to the targets. No fault of the binding but it took me a couple tries to get the serial ports connected and communicating reliably. Then I noticed that only processor and dimmers are supported in the legacy binding. Good news is I can talk to the dimmers.
One strange thing about SCAN and discovery is initially the scan only returns the [01:01:00:x:y:z] dimmers. But if I leave it connected for a few hours I slowly see more dimmers populating the inbox.
I’ve also observed I can manually enter an address and command a not-yet-discovered dimmer.
To answer your questions:
I didn’t write the AMX code but it appears to emulate keypads. I could be wrong and I don’t understand the roles of global scenes and phantom keypads. Most keypads in my house control more than one dimmer address.
My keypads are all ST-4S-NI.
Contact closures and shades (QED) would be nice, but my goal is not to write a whole-house overlay controller. I just want to set up a couple of UI buttons with scenes. To the first order I can do that with what I have now.
Telnet to the processor gets a login: prompt. The login user/password is configured in HW Illuminations software. After successful login the interface appears identical to the RS-232 interface. Which events are sent to the telnet session are user configurable in Illuminations.
A side note is that Illuminations software is old - released around 2005. It’s built for Windows 7 and runs reasonably well on Windows 10, but cannot reliably program a processor from Windows 11. No idea what the issues are. Mine is more reliable talking on the serial port, which is why I’d prefer the OH integration to talk on the telnet port. But I may have a workaround if serial is the only choice.
I’m happy to run tests and DM me about other options. I have a spare series 8 processor if the need is for a static testbed.
On what the binding supports today: the legacy HomeWorks binding (hwserialbridge) handles dimmers only. It doesn’t see keypads, contact closures or QED shades. That fits what you’re seeing, with [01:01:00:x:y:z] dimmers turning up in the inbox as they report in.
Keypad support looks feasible. The HWI RS-232 protocol guide documents keypad monitoring (button press, release, hold and so on), so those events could be exposed as openHAB channels, probably as trigger channels, like the Pico/keypad work that was recently merged for the other bridges. I haven’t read the whole guide yet, and I haven’t confirmed that the Series 8 processor behaves exactly as the guide describes.
That’s where your spare processor would be very valuable. If this goes ahead, a development build of the Lutron binding jar can be published for you to download and drop into your openHAB addons folder. You would run it against the spare processor, so your production setup stays untouched. You’d report back what the processor sends when you press keypad buttons, and whether the new channels behave. Each round of that would show what the guide says versus what the Series 8 actually does, and neither of us can get that without a real device. It also means the code gets checked against hardware before it’s proposed for merge.
Contact closures and QED shades are a separate matter. They’re likely covered by the protocol too, and they could follow as later work if there’s a demonstrated need. Since you describe them as nice to have rather than required, I’d keep the first step small: keypads, and the telnet connection if it proves practical. If you or others later need closures or shades, the same test setup would work for them.
The telnet interface you describe is a nice find. If the session behaves like the RS-232 one after login, a TCP connection mode would have real advantages:
no serial adapter or cable run to the openHAB host
no flaky serial link, which you’ve already seen
the processor can sit anywhere on the network
you already configure which events are sent in Illuminations
The challenge is that the current bridge only opens a serial port. A telnet mode would need a TCP connection plus a login step, and it would need to share the existing parsing code. That’s an interesting problem rather than a hard one, and the spare processor would be a good testbed for it too.
I’d suggest opening a GitHub issue on openhab-addons to start the discussion. Describe your hardware (Series 8, ST-4S-NI keypads, telnet with login), say what you’d like to control, and say you’re willing to test development builds. If other users need contact closures or shades, the issue is also the place for them to say so. That puts it in front of the binding maintainers, and a tester offer there is more likely to get attention than a forum thread. I’m happy to help word it if useful.
This could be about a week to implement in my spare time
If I were to prioritize, telnet access would be the most important thing so I don’t have to tie up the RS-232 port on the processor. I need that for programming the processor.
I’m building an OH test configuration that talks directly to the dimmer modules. The only real hairball is when I create a room item, I have to map out all the dimmer things associated with the room.
Then I’m configuring scenes using Blockly scripts to do things like turn on the first floor