Aqara T2 Dual Relay (LLKZMK12LM): State gets stuck "ON" when using Pulse mode (Firmware v33)

[GuiPoM]

Hi everyone,

I am seeking some technical expertise regarding an architectural behavior I am encountering with the Aqara Dual Relay Module T2 (LLKZMK12LM) on firmware v33 (but same on previous firmwares).

I am currently in contact with Aqara’s technical support and R&D team, who claimed they are looking into it, but I want to double-check from a Zigbee2MQTT perspective to understand exactly where the issue lies.

The Setup
Application: Motorizing a wired roller shutter (230V wet contact configuration).

Z2M UI / Cluster Settings: Both relays are configured with Interlock enabled and Pulse mode active (tested with durations ranging from 200ms to 1s). But same behavior with interlock off.

The Problem
When triggering a relay from Z2M or Home Assistant, the state switches to ON.

On the hardware side, the module executes the pulse perfectly: the relay closes, stays active for the configured duration, and physically clicks open (OFF) locally.

The bug: The state in Zigbee2MQTT and Home Assistant remains permanently stuck on ON.

From my analysis, when the device automatically cuts its own power via its internal hardware timer in Pulse mode, the firmware seems to fail to broadcast its new standard Zigbee on_off cluster attribute state change back to the coordinator.

My Question
Before I suspect a pure Aqara firmware limitation:
Is anyone aware if the T2 module actually broadcasts anything at the end of a local hardware pulse when Interlock is on? Or is there a specific reporting configuration/converter tweak needed in Z2M to force or catch the on_off status update for this specific dual-mode combination?

Thanks a lot for your insights!
Guillaume

[andrei-lazarov]

Hi. It's likely you just need to enable Reporting for the onOff attribute.
Devices usually don't report attributes if they are not asked to. This must be activated by the user.
(Some are automatically activated by Z2M)

[GuiPoM]

Hi Andrei,

Thanks for the hint! I actually just tried explicitly configuring the attribute reporting for the genOnOff cluster (onOff attribute) on both Endpoint 1 and Endpoint 2 (as shown in my screenshot below).

Unfortunately, it does not help here. Even with the reporting active (Min rep interval: 0), the state still remains permanently stuck to ON in Zigbee2MQTT and Home Assistant as soon as the hardware pulse ends.

Unless there is another specific/hidden cluster attribute that needs tweaking to force this broadcast specifically for the dual-mode combination !

[andrei-lazarov]

That should be enough. It does sound like a firmware limitation..
Check the Bind tab (Lier) just in case it wasn't configured correctly. Binds tell it where to send the attribute reports.
They should be sent to Coordinator, of course (or to a group, that would also reach the coordinator)

What version is your device currently running? There are some updates submitted to zigbee-ota.
(Aqara changed version numbering at some point, so Z2M can't automatically push all updates)

https://github.com/Koenkk/zigbee-OTA/blob/master/images/Lumi/20250918173547_OTA_lumi.switch.acn047_0.0.0_0033_20250917_8609D6.ota

[GuiPoM]

Bind was already okay

Firmware is up to date, at least the latest one:

lumi.switch.acn047
Aqara

LLKZMK12LM
Dual relay module T2
Aqara

Firmware
33
App: 0.0 build 0 | Stack: 0.0 build 33

But the issue was the same on a previous one, 28 if I remember well.

[andrei-lazarov]

In that case I hope Aqara support will be helpful 🙁

We can only do an ugly workaround in Z2M code to match the real state:
relay on -> wait pulse duration -> publish relay off
Not really keen to do that..

[GuiPoM]

Totally agree with you. I wanted to double check if I was missing something here as Aqara is now reporting working on this issue for two weeks !
I also tested the native onWithTimedOff command via the Z2M dev console (using the correct ontime and offwaittime payload syntax) but it seems not functionnal.

I would have loved to cross-verify and validate this behavior under ZHA as well to see if it behaves the same, but that's a real pain right now. The module is fully installed and enclosed inside the physical roller shutter housing, so I can't easily factory reset and re-pair it on a test stack.
Anyway, given that the bindings are clean and the native cluster commands seems ignored, the diagnostic should be solid. I will update Aqara with this info, but I still find it strange to be the first one to face this issue, pulse with dry contact of wet is not that unusual and many automation or custom integrations in HA are broken when the relay does report back its state.

Thanks for the help ! 😊

[andrei-lazarov]

Ah, also try to manually read the attribute when it's displayed wrong. Use the blue arrows in Z2M UI, next to the toggle. Curious if that gives the correct value

[GuiPoM]

No, this is the first thing I tried when I noticed that issue.

(both are off)