[ST Edge] SONOFF Hydro ONE / DUO family (SWV-ZFE / SWV-ZFU / SWV-ZNE / SWV-ZNU / SWV-ZF2E / SWV-ZF2U)

I see we wrote practically the same thing at the same time XD

See my post right above yours that I posted at the same time. The valve works fine, it’s just a few changes needed within the edge driver.

Thanks for the feedback. Please keep in mind that I don’t have the device and a lot of the development is just trial and error. Even Zigbee2MQTT is still trying to figure it out. I found two details behind the 10-minute cutoff.

First, SONOFF’s manual_default_settings at 0x501D is not quite a normal ZCL array. SONOFF/eWeLink private-cluster arrays appear to use a one-byte element count, while a standard ZCL Array uses a two-byte count. That can shift the payload by one byte, so the device may not receive the duration value where it expects it. There is a current (one day old!) Zigbee2MQTT issue about exactly this kind of parsing problem with manualDefaultSettings / 0x501D and related eWeLink array attributes:

Second, the device has both a manual watering duration and a separate manual fail-safe timeout. The previous build could change the duration, while the fail-safe value could still remain at 10 minutes.

The new build therefore writes 0x501D using the SONOFF/eWeLink array format and sets both values together:

Manual watering duration = selected value
Manual fail-safe timeout = selected value

So if you set Manual watering duration to 60 minutes, both internal limits should now be set to 60 minutes. Make sure to have the latest driver version 2026-07-02T22:59 installed, set the value again, press Refresh, and test whether the valve stays open beyond 10 minutes.


The waterSensor = dry part is still expected: in this driver, dry means no leak detected, not “no water supply.” That wording is confusing in SmartThings, because waterSensor means water leak sensor.

2 am here. Going to bed now.

I did all that and verified the driver version, but unfortunately the 10 minute issue is still happening. I’ll be glad to help test if you want help.

New version: 2026-07-03T07:38.

I made one more adjustment to the 0x501D manual watering setting.

The previous build was based on a Zigbee2MQTT finding that some SONOFF/eWeLink private array attributes are parsed differently from normal ZCL arrays. That was a useful clue, and manualDefaultSettings / 0x501D is one of the affected attributes.

However, the working deCONZ implementation for the Hydro ONE Lite writes 0x501D more specifically as a normal ZCL Write Attributes payload with a standard array header. So this build now follows that deCONZ write format, while still writing both the manual duration and the fail-safe timeout to the selected value.

Please test again by setting Manual watering duration to 20 minutes, refresh, and then checking whether the valve stays open beyond 10 minutes.

Will test a little later this morning, thanks for your work on this driver, much appreciated!

Hi @Andreas_Roedl ,

Unfortunately I have to report that it’s still 10 minutes. Requested 14 minutes, but closed after 10:

Settings:

Well … I think it’s time for some logs. I had a guy who was providing me logs, but he’s on vacation or something.

Anyway. New version: 2026-07-03T19:54

Added a keepalive to override the firmware timeout.

  1. Make sure that the latest version is installed.
  2. Keep Timed watering mode = driver timer.
  3. Set Manual watering duration to 20 or 60.
  4. Refresh.
  5. Request Open valve for minutes = 14 or 15.
  6. Check whether it now runs beyond 10 minutes.

No worries, I can provide logs for you too, but I just left for a short holiday until Tuesday. This isn’t urgent by any means, but I’ll let you know how my additional testing goes. Hope you have a good weekend and your temperatures are cooler.

Outside and wearing a jacket now.

Hi @Andreas_Roedl

Ermmmm… Since the driver updated to …07-03… it has failed to operate my valve at all, just not responsive :man_shrugging:t3:

Re-added device but no joy, I’ve had to resort to ST stock :frowning:

Sorry about that. That version introduced an regression while trying to work around the 10-minute timeout, and that could make the valve stop responding. I’ve pulled that build and replaced it with a newer one (2026-07-04T19:41), which rolls back the risky changes and is meant to restore reliable basic valve operation first.

The timed-watering cutoff is still under investigation, and I’ll need logs before trying another fix.

Edit: version 2026-07-04T20:14 changes Open valve for minutes to use Zigbee On with Timed Off with strict ZCL timing, while keeping the driver-side close timer as a backup and leaving normal Valve/Switch control unchanged.