Using Direct Valve Based with MCP Predictive Calibration still sets the target temp on my SONOFF TRVZB

[DrBlackRat]

Heyo,
I have a bit of an odd issue, but before I start, here is my setup:

Devices / Versions

• Better Thermostat Version 1.9.3
• 2x SONOFF TRVZB Thermostats; Firmware: 0x00001404
• 1x SONOFF SNZB-02D Temp & Humidity Sensor; Firmware: 0x00002300

All connected to Home Assistant via ZHA.

Config / Setup

One Better Thermostat that controls the top SONOFF Thermostats and used the Temp & Humidity.
I'm using the "Direct Valve Based" mode with "MCP Predictive (Beta)" Calibration.

BT Config

TRVZB Config

Question / Problem

The odd behavior I'm seeing with this config is that Better Thermostat seems to still just set the Target Temp on my Thermostats?

For example when I set the BT Thermostat to 21°C the TRVZB will be at like 17 or 18.5 and not the set 21°C, I would have expected it to:

• Set the target temp on the thermostats to what is set in BT
• Use the Valve Opening Degree to control how open the Valves are

So, is this the expected behavior? Also, is this the best setup for these SONOFF TRVZB or should we use something else?

[cl445]

Hi, thanks for the detailed write-up and the screenshots.

Yes, this is expected. In "Direct Valve Based" mode Better Thermostat controls the valve through the opening degree, but it still writes a setpoint to the TRV, because the TRVZB firmware keeps running its own regulation on top of the valve limits. Which setpoint it writes depends on what MPC decided for that TRV:

• MPC opens the valve (above 0 %): the TRV setpoint is set to your BT target, 21 °C in your example, so the firmware doesn't hold the valve back.
• MPC closes the valve (0 %): the setpoint is pushed below the TRV's own temperature reading. Otherwise the TRV would decide on its own to heat, even though BT just told it the valve should be closed. That's where values like 17 or 18.5 °C come from.

With two TRVs on one Better Thermostat there's a second effect: MPC computes one valve command for the room and splits it between the TRVs. The warmer TRV gets less opening, possibly 0 %. So at the same moment one TRV can show 21 °C and the other a lower value.

To confirm the valve path is active on your setup, could you check two things?

1. The attributes of the BT climate entity: calibration_balance should show a valve_percent value.
2. A debug log (custom_components.better_thermostat): you should see lines like TO TRV set_valve: ... to: NN% (source=mpc_calibration).

If BT doesn't treat the valve as writable, MPC falls back to steering through the setpoint alone, scaled between the TRV temperature and its max temperature. That would also produce setpoints that aren't 21 °C, so the log helps tell the two cases apart.

One more thing that affects you: on 1.9.3, a TRVZB paired through ZHA never receives the room temperature from your external sensor, because BT only knows the Zigbee2MQTT entity names for it. Valve degrees are also rewritten on every control cycle. Both are tracked in #2767, with a fix in #2768. For direct valve control this matters less, since BT decides the valve opening from your SNZB-02D anyway.

As for the setup itself: direct valve with a TRVZB and an external room sensor is a sensible combination, so no need to change it because of the setpoint values.

[DrBlackRat]

Thanks for the quick and thorough reply!

I've checked the calibration_balance which shows a valve% present, as well as saw on the TRVZB itself that the valve opening degree is being changed, so that all seems to work.

I guess I will just enable the child lock on the TRVZBs and tell my partner not to look at what is set on them directly anymore, but that should be doable!

[DrBlackRat]

Ah, I do have one other question though!

Currently, when heating, it sets the target temp on the TVRZB to, well, the target temp.

This is confusing me a little since, if the TVRZB decides it reached the target temp, it won't heat even though BT would want it to. Especially since the built-in temp sensor of the TVRZBs is usually quite a lot higher than what external sensors measure.

Should I e.g. set the temperature offset on my ZVRZB to -10 so that it always heats unless BT doesn't want it to?

[cl445]

You're right. While the valve is open, BT writes your BT target as the TRVZB setpoint. The TRVZB still regulates on its own sensor, and next to the radiator that sensor often reads above the room. Once it reads at or above the target, the firmware closes the valve, whatever opening degree BT has set.

So yes, a negative offset on the TRVZB works around it on 1.9.3. With the offset, the TRVZB reads lower than the radiator, stays below the setpoint and keeps the valve at the opening degree BT gives it. The closed case keeps working: when MPC wants 0 %, BT puts the setpoint below the temperature the TRVZB reports, and that reported value already includes your offset. You don't need the full -10. Pick enough to bring the TRVZB's reading clearly below your external sensor; a few degrees is usually plenty.

I've opened a fix for 2.0 (#2889): while the valve is open, the setpoint is kept at least 1 °C above the TRV's own reading, and never below the target. With that, the offset trick is no longer needed.